Professional Documents
Culture Documents
09340, Mexico
a r t i c l e i n f o a b s t r a c t
Article history: This work presents the numerical modeling of the effect of hydrodynamics on mass trans-
Received 28 September 2014 port and tertiary current and potential distribution in a filter press type electrochemical
Received in revised form 29 April reactor used to study the copper recovery process. The operating conditions of the reactor
2015 were in turbulent regime and under charge and mass transfer mixed control.
Accepted 30 April 2015 For hydrodynamics, the Reynolds averaged Navier–Stokes equations and the stan-
Available online 11 May 2015 dard k–ε turbulence model were used. The mass transfer model was a combination of
the convection–diffusion equation and a wall function adapted for mass transfer. The
Keywords: Butler–Volmer kinetics for copper reduction, simplified Tafel equations for water oxidation
Electrochemical reactors and ohmic potential drop through the electrolyte were also incorporated into the model.
Tertiary current distribution The strategic part of the proposed numerical modeling is the concentration wall function
Turbulent mass transfer that allows linking the transport equations with Cu2+ concentration at the interface in order
Copper recovery to obtain, along with interfacial potential, the electrode kinetics. Using this approach it was
Butler–Volmer kinetics possible to model a very complex interrelation between physical phenomena and the elec-
trochemical reaction taking place in a reactor under a turbulent flow regime using moderate
computer resources. The numerical results obtained are in agreement with experimental
data of mass transfer coefficient and current–potential behavior.
© 2015 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.
∗
Corresponding author. Tel.: +52 55 56232071.
E-mail address: priveromtz@msn.com (E.P. Rivero).
http://dx.doi.org/10.1016/j.cherd.2015.04.036
0263-8762/© 2015 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.
chemical engineering research and design 1 0 0 ( 2 0 1 5 ) 422–433 423
Nomenclature
ε turbulent energy dissipation rate
A empirical constant a electrolyte potential at the anode interface
a,b kinetic constants for water oxidation electrolyte potential
B reactor channel width c electrolyte potential at the cathode interface
C constant of the turbulence model Karman constant
Cε1 constant of the k–ε model viscosity
Cε2 constant of the k–ε model T turbulent viscosity
C+ dimensionless concentration density
Ci concentration of the electroactive species electrolyte conductivity
Ci,0 inlet concentration of the electroactive species k constant of k–ε model
Ci,w wall concentration of the electroactive species ε constant of k–ε model
de equivalent diameter = 2BS/(B + S) shear stress
Di diffusion coefficient w wall shear stress
DiT turbulent diffusion coefficient
E potential
Ea anodic potential variations are very important, not only because of the amount
Ec cathodic potential of electroactive species which can reach the electrode sur-
Eeq equilibrium potential of copper reduction face, but also because of their effect on the overpotential in
Eeq,ox equilibrium potential of water oxidation each zone of the reactor. Mass transfer rate and overpoten-
E0 equilibrium cell potential difference tial are factors involved in the type of metallic copper deposit
E˝ ohmic drop formed (compactness, grain size, dendrite and powder pro-
F Faraday constant duction) as well as the flow characteristics are involved in
F external force the shear stress over the electrode surface for carrying the
J current density metallic copper out in the liquid stream. Rough deposit is
JL limiting current density formed at local laminar hydrodynamic conditions (current
Ja current density at the anode electrodeposition lower than the limiting current density con-
Jc current density at the cathode ditions), whereas a turbulent local regime provokes dendritric
J current density vector deposits (Bazan and Bisang, 2004; Robinson and Walsh, 1991a,
J0 exchange current density 1991b). In addition, high overpotentials may lead to unde-
k turbulence kinetic energy sired side reactions that may diminish the charge efficiency of
km mass transfer coefficient the reactor and lead to uncontrollable changes (pH, conduc-
L length of the reactor channel tivity, excessive gas formation, mass transfer rate, formation
n vector normal to the wall of solid byproducts, etc.). Therefore, coupling hydrodynam-
Ni mass transfer flux ics, mass transport, electrochemical kinetics and potential
Ni,w mass transfer flux at the wall and current distribution is necessary to obtain a comprehen-
P pressure sive understanding of the electrochemical reactor behavior.
Pc factor of integration of resistances in near-wall The importance of the fluid dynamics on mass transport and
layers electrochemical kinetics has been shown by modeling elec-
R gas constant trodeposition process on rotating electrodes (Mandin et al.,
Re Reynolds number = de v/ 2007a, 2007b, 2006a, 2006b). Also, in processes with inten-
S electrode spacing sive gas evolution reactions, as in water electrolysis, there is
Sc Schmidt number = /(Di ) a strong coupling of these phenomena since the bubble for-
ScT turbulent Schmidt number = T /(Di,T ) mation diminishes the electrolyte conductivity and modifies
ScT∞ constant of the Kays–Crawford model the hydrodynamic behavior affecting the current distribution
Sh Sherwood number = de km /Di over the electrode (Mandin et al., 2014, 2005). However, model-
t time ing tertiary current distribution in an electrochemical process
T temperature may be very complex, especially for reactors operating under
u velocity vector turbulent flow. Thus, several authors have studied the prob-
u velocity lem using some simplifications: the 2D current distribution
U cell potential = Ea − Ec over flat plate electrodes and other simple geometries where
u+ dimensionless velocity = u/( w /)1/2 electrolyte flows in the laminar regimen have been reported
Vin inlet velocity (Georgiadou, 2003; Boovaragavan and Basha, 2006). Also, the
x,y,z Cartesian coordinates relationships among mass transfer, ohmic drop and overpo-
y+ dimensionless distance to the tential have been studied for copper electrodeposition in a cell
wall = y( w /)1/2 / without any type of convection (Pohjoranta et al., 2010). How-
zi number of electrons exchanged in electro- ever, very few works have studied electrochemical reactions in
chemical reaction reactors under turbulent flow, which is a very important case
from a practical point of view. Even though the characteristic
Greek letters lengths of the laboratory scale FM01-LC electrochemical reac-
˛ transfer coefficient tor are small, the design of the inlet distributor produces high
ı+ dimensionless thickness selected for wall velocity streams or jets not parallel to reactor walls, which
function = ı( w /)1/2 / are discharged inside the reactor channel causing 3D flow
424 chemical engineering research and design 1 0 0 ( 2 0 1 5 ) 422–433
of the hydraulic channel. Such a configuration ensures a (5) only Cu2+ reduction at the cathode and water oxidation at
more uniform flow and electric potential distribution so the anode are considered.
that mass transport measurements are only associated with
the flow pattern in the hydraulic channel. For experimental 3.1. Fluid flow
determination of the limiting current, polarization curves
were constructed using sampled current density vs. poten- For the purposes of the present work, the bulk-flow dynamics
tial. It is important to mention that the authors decided to can be approximated by the Reynolds averaged Navier Stokes
construct the polarization curves in this way instead of by equation (Eq. (3)) using turbulence models as closure condition
using the linear sweep voltammetry technique, because a since it is still an effective and practical tool to approach this
voltammetric study always involves competition between the kind of problem (Bernard and Wallace, 2002; Eςa and Hoekstra,
rate of electrochemical reaction at the interface and the rate 2010; Lacasse et al., 2004; Trujillo et al., 2009). The standard
of potential change. This competition is eliminated in the k–ε turbulence model and the balances of k and ε given by
potential pulse technique. Eqs. (4)–(6) are used in this work. Since this approach is not
The polarization curves have been constructed as follows: applicable in the proximity of solid surfaces, log-law based
a series of potential double pulses was applied to the elec- wall functions are commonly used as boundary conditions
trochemical reactor. The direct pulse was started from open (Bernard and Wallace, 2002; Eςa and Hoekstra, 2010; Lacasse
circuit potential OCP (−0.34 versus SSE) in the potential range et al., 2004), i.e. Eq. (3) is solved only in the fluid bulk sepa-
of −1.0 ≤ E ≤ −0.4 V versus SSE during 30 s, with increments of rated by a distance ı from walls where the velocity is linked
0.05 V. The inverse pulse was 0.58 V to ensure the oxidation to the wall function (Eq. (7)). This wall function based on the
of the previously deposited copper. The corresponding cur- universal velocity distribution in the viscous, buffer and turbu-
rent time curves were obtained at different volumetric flows lent layers defines the parallel-to-wall velocity in the turbulent
between 2 and 7 LPM (1384 ≤ Re ≤ 4845). From these current layer as a function of its distance to the wall, where molecu-
time curves obtained for direct pulses at the applied poten- lar viscosity is considered negligible compared to turbulent
tial, the current density was sampled at 15 s and polarization viscosity.
curves were constructed (J–E) for a specific flow rate. The
T
sampled time ensures both a negligible contribution of the (u · ∇) u = −∇P + ∇ · ( + T ) ∇u + (∇u) + F (3)
capacitive current to the measured current and a pseudo-
steady state (the current is time-independent). k2
T = C (4)
ε
3. Mathematical model and simulation T
1 2
u · ∇k − ∇ · + ∇k = T ∇u + (∇u)T − ε (5)
k 2
Copper recovery in a continuous bench scale electrochemical
reactor may be carried out under either potentiostatic or gal-
T
vanostatic conditions where copper reduction (Eq. (1)) takes
u · ∇ε − ∇ · + ∇ε
place with a minimum effect of side cathodic reactions (hydro- ε
to locate the boundary in the turbulent layer and the logarith- where u is the averaged velocity vector, Ci the averaged con-
mic velocity distribution given by Eq. (7) could be applied. Fig. 3 centration field, Di the diffusion coefficient and Di,T the eddy
illustrates this wall function along with the location of viscous diffusivity or turbulent diffusivity. This last term can be deter-
(y+ ≤ 5), buffer (5 ≤ y+ ≤ 30) and turbulent (y+ ≥ 30) layers, the mined from the turbulent Schmidt number (ScT = T /(Di,T ))
mass transport wall conditions and other model conditions given by the following Kays Crawford model as described pre-
for the cathode surface. viously (Rivero et al., 2012).
−1
1 0.3 T
T
2 −Di
ScT = +√ − 0.3 1 − exp √ (9)
2ScT∞ ScT∞ Di Di 0.3T ScT∞
Ci + = Sc y+ (10)
1
Ci + = ScT ln y+ + 5.5 + Pc (11)
where A is a constant whose value (A = 9.24) has been deter- The ohmic potential drop can be obtained from the con-
mined from heat transfer experimental data in smooth tubes tinuity equation for the electrolyte between the electrodes,
(Jayatillake, 1969). according to
Considering the transport of Cu2+ ion toward the cathode
(i.e. subscript i is equal to Cu2+ ) where a reduction reaction ∇ · J = −∇ · ( ∇) = 0 (17)
takes place, the boundary conditions of the mass transport
model are: where J is the current density vector and the electrolyte
conductivity.
As the overpotentials are given by the Butler–Volmer equa-
• Ci = Ci,0 , at the reactor inlet.
tion for copper reduction (Eq. (14)) and Tafel equation for water
• n·(−Di Ci ) = 0, at the reactor outlet (convective flux)
oxidation (Eq. (15)), Eq. (17) applies for the potential between
• Wall function (Eqs. (11) and (12)) at the electrode, with
c and a . Assuming constant cathodic and anodic potentials,
Ni,w = Jc /zi F.
Ec and Ea respectively, the interface electrolyte potentials, c
• n·(−Di Ci + uCi ) = 0, in all other (impermeable) walls.
and a , vary over the electrode and are unknown. Therefore,
Dirichlet type boundary conditions of Eq. (17) for the electrode
where Jc is the local current density at the cathode, zi the num- interface are not possible, instead Neumann type conditions
ber of electrons in the reduction reaction (two electrons in this are to be used as follows.
case) and F the Faraday constant.
• −n · J = Jc , at the cathode.
3.3. Electrode kinetics • −n · J = Ja , at the anode.
• n · J = 0, at all other boundaries.
The most accepted mechanism for Cu2+ reduction involves
two successive monoelectronic transfer reactions, adsorp- where Jc and Ja are the local current density given by the kinetic
tion processes and ion migration over the electrode surface. equations (Eqs. (14) and (15)), which depend on overpotential.
Even though other mechanisms have been reported and there With these equations the mathematical development to
has not been a clear discrimination among several reaction obtain the tertiary current distribution in an electrochemical
sequences to elucidate the real mechanism, the experimental reactor for copper recovery under turbulent flow is complete.
electrodeposition rate data might be fitted to Butler–Volmer
kinetics, according to: 3.5. Simulation
zi (1 − ˛)F (Ec − c − Eeq ) The equations for hydrodynamic, mass transport, electrical
Jc = J0 exp
RT continuity, electrode kinetics and concentration wall func-
tion form a coupled set of partial differential equations and
CCu2+ ,w −zi ˛F (Ec − c − Eeq ) algebraic equations. However, since hydrodynamic changes
− exp (14)
CCu2+ RT caused by anodic and cathodic reactions are not considered
(assumptions 1 and 2), the hydrodynamic model was solved
where J0 is the exchange current density, Ec the cathodic first. It was made by use of finite elements using commercial
potential, c the electrolyte potential at the cathode inter- software (COMSOL Multiphysics® ) as follows.
face, Eeq the equilibrium potential, CCu2+ ,w the interface Cu2+ The hydrodynamic model was solved with the option of the
concentration, CCu2+ the bulk concentration, R the gas con- k–ε turbulence model, momentum transport, using calculation
stant and T the temperature (parameters are given in Table 1). options by default: stationary segregated solver with BICGStab
Eq. (14) corresponds to a mixed control (kinetics and diffu- iterative methods for u and P, and GMRES for logarithms
sion). The potential range for the copper electrodeposition in a of the turbulence model variables, log(k) and log(ε). Galerkin
mixed control was previously determined in a microeletrolysis Least Squares (GLS) was used as a stabilization method; tol-
study. In this potential range, HER is negligible. erance was fixed at 10−3 . Upon verifying numerical solutions
Assuming that oxidation of the medium (water) follows a reached at different values of ı+ and mesh size, the ı+ value
Tafel expression: was fixed at 100 (typical value used within the validity of
the logarithmic velocity distribution law, assuming no effects
Ja = a exp (b (Ea − a − Eeq,ox )) (15) of roughness caused by copper deposits) and the maximum
mesh size in the reactor channel at 2 × 10−3 m. Fig. 4 shows
the mesh in the reactor inlet used in simulation (37,042 total
where Ea is the anodic potential, a the electrolyte potential
mesh points and 185,250 tetrahedral elements). The results
at the anode interphase, Eeq,ox the equilibrium potential of
obtained were stored through the option available in the pro-
the water oxidation, a and b are empirical constants shown in
gram, in order to be used in the mass transfer model. The
Table 1.
diffusion–convection equation (Eq. (8)) along with mass trans-
port wall functions given by Eqs. (11)–(13) were then added to
3.4. Electric continuity
the model, linking the results of the hydrodynamic calcula-
tion: (i) the time-averaged velocity to evaluate the convection
The cell potential defined by the voltage difference (Ea –Ec ) is
term in Eq. (8), (ii) the turbulent viscosity to obtain turbulent
given by the sum of absolute values of overpotentials at both
diffusivity through the turbulent Schmidt number, ScT , and (iii)
electrodes plus the ohmic drop E˝ and the equilibrium cell
the eddy kinetic energy necessary for the concentration wall
potential difference E0 , as follows:
functions (Eqs. (11) and (12)).
The mass transport model requires the current density Jc
U = (Ea − a − Eeq,ox ) + E˝ − (Ec − ϕc − Eeq ) + E0 (16) at the boundary conditions, which is set by Cu2+ reduction
428 chemical engineering research and design 1 0 0 ( 2 0 1 5 ) 422–433
ohmic potential drop through the electrolyte given by the elec- 0.8
trical continuity equation (Eq. (17)). However, as the potential
drop is a function of the current density it is necessary to eval- 0.6
uate the boundary conditions of Eq. (17), Jc and Ja , from the z2
0.4
kinetic equations, Eqs. (14) and (15), respectively. The kinet-
ics of anodic reaction given by Eq. (15) are only related to one 0.2
interfacial variable: a , and must satisfy the current continuity
criterion i.e. anodic and cathodic currents are equal, although
0.0
z3
the current density distribution over the anode, Ja , could be -0.2
different from that of the cathode Jc . Taking into account all 0 0.01 0.02 0.03 0.04
these relationships the complete model can be solved for given
Cell width, y/m
electrode potentials, Ea , Ec and inlet Cu2+ concentration. The
mass transport and electrical continuity models were solved Fig. 6 – Axial velocity profile as a function of channel width
simultaneously by finite elements along with kinetics equa- at 0.005 m (continuous line) and 0.01 m (broken line) from
tions for Jc and Ja and the concentration wall function, using the channel inlet and different distances from the cathode,
COMSOL Multiphysics® . The solver employed was iterative, z1 = 0.005 m, z2 = 0.008 m and z3 = 0.01 m; Re = 4835.
FGMRES, and a relative tolerance of 1 × 10−6 was established.
The values of ı+ and mesh size were as those of the hydro- profiles at different distance from the cathode, z1 = 0.005 m,
dynamic model. Simulations were done using a 4-core Intel i7 z2 = 0.008 m and z3 = 0.01 m, and from the inlet, 0.005 m (solid
920 processor at 2.67 GHz, 12 GB RAM. The values of electrolyte line) and 0.01 m (broken line) are shown in Fig. 6, where
properties and model parameters used to solve the model are negative velocity indicates a recirculation pattern. These flow
shown in Table 1. instabilities caused by the inlet distribution and the absence
of calming zone make it impossible for the flow to be laminar
4. Results and discussion even at low Re. Also, it is clear that a fully developed flow
is not reached in the short axial length of the FM01-LC. All
4.1. Hydrodynamics these characteristics of the Reynolds averaged velocity field
reveal complex velocity variations in all directions making
There has been a number of works dealing with hydrody- impossible to find symmetries for performing 2D simulations
namics inside flat plate electrochemical reactors; some of so that simulations in 3D are necessary.
them have specifically treated the FM01-LC reactor (Rivero
et al., 2012; Vázquez et al., 2010). As they commonly used 4.2. Mass transfer assessment
CFD to describe the reactor behavior we only show the
main characteristics investigated in this work. The manifold Before the model for tertiary current distribution can be used
effects, mainly those of the inlet manifold, on the velocity for simulating Cu reduction in a FM01-LC reactor, the mass
field are shown in Fig. 5, where jets of different velocity are transfer rate should be adjusted by experimental results. The
clearly formed causing preferred flow paths, recirculation mass transport model given by Eqs. (8)–(13) uses an empirical
and low velocity regions. Two high velocity streams, one parameter (A) for correcting the concentration wall function.
higher than the other, close to lateral walls and a low veloc- The A-parameter value has been reported for heat transfer
ity region in between can be identified. The axial velocity over smooth surfaces (A = 9.24). This value has allowed the
chemical engineering research and design 1 0 0 ( 2 0 1 5 ) 422–433 429
0.12 -0.06
0.08
0.04
0
-10
-400 -70
-500 -85
Fig. 12 – Patterns of copper electrodeposition over the -0.5 -0.25 0 0.25
electrode of the FM01-LC reactor at (a) Re = 2769 and 5 min Potential, Ec vs SHE/ V
electrolysis time, and (b) Re = 1384 and 10 min electrolysis
time. Fig. 13 – Average current density obtained from copper
electrodeposition simulations at FM01-LC reactor under
different cathodic potentials (solid line) and experimental
sampled current (dots). Results of simulation (broken line)
time this high current density demands a higher overpo-
of a reactor channel with homogenous inlet velocity and
tential according to the Butler–Volmer equation. In addition,
constant sectional area are shown for comparison.
high current density regions provoke high ohmic potential
Conditions: 18.8 mM CuSO4 in 1 M H2 SO4 , Re = 4845.
drop diminishing the available overpotential in potentiostatic
experiments so that a dynamic balance is reached at the
steady state. The final results are a low overpotential and Fig. 12 shows the patterns of the copper deposits on the
high current density (absolute values) regions at the electrode electrode surface obtained after 5 and 10 min of electrolysis.
surface at the entrance of the reactor (Figs. 10 and 11). There The regions covered by copper follow the current density dis-
is then a calm region, where a high overpotential and low tribution pattern at the cathode shown in Fig. 11. At short
current density are evident. It is worth noting that the mass times the figure reveals the effect of jets on copper deposition.
transfer resistance is due not only to the low local velocity Also, the effect of the two high velocity streams near the chan-
but also to the low turbulence intensity. nel wall (Fig. 5) and the consequently high current density at
At the anode, the electrochemical reaction is not the cathode edges (parallel to the flow direction in Fig. 11), is
dependent on mass transport of reactants, however, the non- displayed in Fig. 12a. At longer electrolysis times (10 min), the
uniformity of current density at the cathode surface produces copper deposit shape follows the high and low current den-
an anodic current density distribution due to the electrical sity zones (Fig. 12b) depicted by simulation in the cathode in
continuity requirement. Nevertheless, the variations of cur- Fig. 11.
rent density at the anode are smoother than those at the With regard to simulations at different cathodic potentials,
cathode showing a steady decrease of current density from Fig. 13 shows the average current density obtained by surface
the inlet in the axial direction that reaches a minimum at integration of the current density over the electrode divided
13.8 cm from the inlet and then a slight increase close to the by the electrode area at each steady state simulation (solid
outlet. Since the overpotential depends on the current density line) and the comparison with experimental data (dots).
distribution, an ovepotential distribution at the anode is also Even though the current distributions are not uniform on the
evident. It is worthy of note that the oxygen evolution at the electrode, the average current densities calculated are close to
anode was not taken into account in the model. Although gas the experimental data. This fact proves the capability of the
bubble production may lead to several changes like decreas- approach used in the model. On the other hand, the very wide
ing the effective conductivity and modifying the flow pattern, sigmoidal curve (solid line) depicted in Fig. 13 represents the
their effects were considered minor as the high flow used can wide current density distribution on the electrode, i.e. at not
carry the bubbles away rapidly. very negative cathode potential some areas of the electrode
The general behavior obtained by simulation and depicted surface reach the limiting current density while others (with
in Figs. 8–11 for copper recovery in the FM01-LC reactor was lower mass transfer resistance) close to the entrance of the
found to be qualitatively similar to the results of simulations reactor are far from limiting current density; and the more
at different Re ranging from 1384 to 4845 at the same cathodic negative is the cathode potential, the larger is the area of
potential (with respect to the reference electrode). The jet for- the electrode surface reached by the limiting current density.
mation and the flow pattern developed determine the mass For comparative purposes Fig. 13 shows the current potential
transport rate in each zone of the reactor leading to equivalent curve (broken line) predicted by the model when ideal inlet
effects on current density and overpotential on the electrode and outlet reactor distributors are used. The simulations at
surface but with faster rate as Re increases. the same Re (Re = 4845) and inlet copper ion concentration
Experimentally, the type of copper deposits is different (18.8 mM) were performed only in the reactor channel (formed
depending on Re. At high Re, keeping the same cathodic poten- by the space between the electrodes) with a homogenous
tial, the high copper deposition rate leads to loosely adherent inlet velocity. Under these circumstances, the sigmoidal curve
copper deposits making the removal of copper deposits easier is more sharply so that the mass-transfer-controlled plateau
because of the high shear stress of the fluid flowing at high is reached at less negative potential. In this case, as expected
velocity. the current density is lower than that obtained for the actual
432 chemical engineering research and design 1 0 0 ( 2 0 1 5 ) 422–433
FM01-LC reactor since the additional turbulence generated up. The results obtained from the simulation are in agree-
by distributors in the last case leads to higher mass transfer ment with potential current experimental data before the HER
coefficients. takes place. Moreover, they allow explanation of the difference
Furthermore, as the cathode potential in the FM01-LC between the potential necessary to start the HER in the FM01-
becomes more negative, the overpotential of higher mass LC and that expected from microelectrolysis experiments. It
transfer resistance zones increases more (in absolute value) is important to note the advantages in approach presented
than the average overpotential value, which increases the risk here which allowed a complete description of the tertiary cur-
of side reactions in these zones. The experimental points at rent distribution in an electrochemical reactor under the most
more negative potential in Fig. 13 show that the current begins practical type of flow regime (turbulent flow) requiring mod-
to increase (absolute value) due to the appearance of the erate computer resources.
hydrogen evolution reaction (HER) which shows that higher
overpotential regions started to develop this side reaction. It References
is important to note that at this potential (0.386 V vs. SHE) the
HER is not expected to take place based on micro electroly- Bazan, J.C., Bisang, J.M., 2004. Electrochemical removal of tin from
sis experiments on an RDE (Rivera and Nava, 2007). However, dilute aqueous sulfate solutions using a rotating cylinder
in the FM01-LC reactor, higher overpotential regions might electrode of expanded metal. J. Appl. Electrochem. 37, 501–506.
be developing before the reference point can reach such high Bengoa, C., Montillet, A., Legentilhomme, P., Legrand, J., 2000.
overpotential owing to the non uniform overpotential distribu- Characterization and modeling of the hydrodynamic behavior
tion. Clearly, under these conditions the model deviates from in the filter-press-type FM01-LC electrochemical cell by direct
flow visualization and residence time distribution. Ind. Eng.
the experimental data as the side reactions are not taken into
Chem. Res. 39, 2199–2206.
account in the model, but the implications of the model calcu- Bernard, P.S., Wallace, J.M., 2002. Turbulent Flow: Analysis,
lations remain valid since regions of higher overpotential than Measurement, and Prediction. John Wiley & Sons, New Jersey.
that where the potential is measured exhibit an overpoten- Bird, R.B., Stewart, W.E., Lightfoot, E.N., 2007. Transport
tial high enough to drive the reduction of protons. Moreover, Phenomena, second ed. John Wiley & Sons, USA.
the overpotential and current density distribution explain the Boovaragavan, V., Basha, C.A., 2006. A novel approach for
computing tertiary current distributions based on simplifying
experimental results. The picture described for the overpo-
assumptions. J. Appl. Electrochem. 36, 745–757.
tential and current density distribution where the fraction of
Brown, C.J., Pletcher, D., Walsh, F.C., 1993. Studies of
surface area of the electrode reaches the limiting current con- space-averaged mass transport in the FM01-LC laboratory
ditions and the overpotential is high enough to develop side electrolyser. J. Appl. Electrochem. 23, 38–43.
reactions as the cathode potential becomes more negative, Doche, O., Bauer, F., Tardu, S., 2012. Direct numerical simulation
produces current–potential curves with ill-defined plateaux of an electrolyte deposition under a turbulent flow—a first
which is in agreement with experimental results. approach. J. Electroanal. Chem. 664, 1–6.
Doche, O., Bauer, F., Tardu, S., 2013. Direct numerical simulations
of electrochemical reactions in turbulent flow. Electrochim.
5. Conclusions Acta 88, 365–372.
Eςa, L., Hoekstra, M., 2010. Near-wall profiles of mean flow and
A comprehensive mathematical model was developed to sim- turbulence quantities predicted by eddy-viscosity turbulence
ulate the complex relationship between hydrodynamics and models. Int. J. Numer. Methods Fluids 63, 953–988.
distribution of tertiary current and potential for copper recov- Frías-Ferrer, Á., González-García, J., Sáez, V., Ponce de León, C.,
Walsh, F.C., 2008. The effects of manifold flow on mass
ery in a continuous filter press type electrochemical reactor,
transport in electrochemical filter-press reactors. AIChE J. 54,
where non-ideal flow patterns, like backmixing, preferen-
811–823.
tial flow, inlet effects, dead zones, etc., are present. The Georgiadou, M., 2003. Modeling current density distribution in
model comprises the following equations: Reynolds averaged electrochemical systems. Electrochim. Acta 48, 4089–4095.
Navier Stokes, convection–diffusion, Butler–Volmer kinetics Griffiths, M., Ponce de León, C., Walsh, F.C., 2005. Mass transport
for copper reduction, Tafel for water oxidation and electri- in the rectangular channel of a filter-press electrolyzer (the
cal continuity. The model was able to provide information FM01-LC reactor). AIChE J. 51, 682–687.
Hammond, J.K., Robinson, D., Walsh, F.C., 1991. Mass transport
on velocity, copper ion concentration and electrical potential
studies in filterpress monopolar (FM-type) electrolysers. In:
fields inside the reactor, from which important distributions
Kreysa, G (Ed.), Electrochemical Cell Design and Optimisation
could be calculated: overpotential, mass transfer flux and cur- Procedures. Dechema Monograph, vol. 123. VCH, Weinheim, p.
rent density over the electrode surface. As part of the model 279.
development the mass transfer coefficients were determined Hotlos, J., Jaskula, M., 1988. Densities and viscosities of
experimentally and a fitting parameter (A) of the concentra- CuSO4 –H2 SO4 –H2 O solutions. Hydrometallurgy 21, 1–7.
tion wall function was tuned to the mass transfer model. Jayatillake, C.L.V., 1969. The influence of Prandtl number and
surface roughness on the resistance of the laminar sub-layer
The A-value and mass transfer coefficients obtained account
to momentum and heat transfer. Prog. Heat Mass Transfer 1,
for the enhancement of mass transfer due to the roughness 193–329.
caused by copper deposition. Simulation results show the Lacasse, D., Turgeon, E., Pelletier, D., 2004. On the judicious use of
importance of the reactor geometry, especially of the inlet and the k–ε model, wall functions and adaptivity. Int. J. Therm. Sci.
outlet distributors in the FM01-LC reactor where some flow 43, 925–938.
characteristics such as back mixing, preferential flow and low Mandin, Ph, Hamburger, J., Bessou, S., Picard, G., 2005. Modelling
and calculation of the current density distribution evolution
velocity zones cause strong variations of the mass transfer
at vertical gas-evolving electrodes. Electrochim. Acta 51,
rate in different parts of the electrode and, consequently, a
1140–1156.
nonuniform overpotential distribution over the electrode sur- Mandin, Ph, Cense, J.M., Picard, G., Lincot, D., 2006a. Simplified
face. These characteristics have profound implications on the kinetic modelling and numerical simulation of a metal oxide
reactor performance that require assessment in order to ana- chemical bath electro deposition process at a rotating
lyze experimental data of copper recovery before process scale electrode. Electrochim. Acta 52, 1296–1308.
chemical engineering research and design 1 0 0 ( 2 0 1 5 ) 422–433 433
Mandin, Ph, Fabian, C., Lincot, D., 2006b. Importance of the electrode (RCE) reactor under turbulent flow for copper
density gradient effects in modelling electro deposition recovery. Chem. Eng. Sci. 65, 3042–3049.
process at a rotating cylinder electrode. Electrochim. Acta 51, Rivero, E.P., Rivera, F.F., Cruz-Díaz, M.R., Mayen, E., González, I.,
4067–4079. 2012. Numerical simulation of mass transport in a filter press
Mandin, Ph, Cense, J.M., Fabian, C., Gbado, C., Lincot, D., 2007a. type electrochemical reactor FM01-LC: comparison of
Electrodeposition process modelling using continuous and predicted and experimental mass transfer coefficient. Chem.
discrete scales. Comput. Chem. Eng. 31, 980–992. Eng. Res. Des. 90, 1969–1978.
Mandin, Ph, Cense, J.M., Georges, B., Favre, V., Pauporté Th, Robinson, D., Walsh, F.C., 1991a. The performance of a 500 Amp
Fukunaka, Y., Lincot, D., 2007b. Prediction of the rotating cylinder electrode reactor. Part 1: Current–potential
electrodeposition process behavior with the gravity or data and single pass. Hydrometallurgy 26, 93–114.
acceleration value at continuous and discrete scale. Robinson, D., Walsh, F.C., 1991b. The performance of a 500 Amp
Electrochim. Acta 53, 233–244. rotating cylinder electrode reactor. Part 2: Batch recirculation
Mandin, Ph, Derhoumi, Z., Roustan, H., Rolf, W., 2014. Bubble studies and overall mass transport. Hydrometallurgy 26,
over-potential during two-phase alkaline water electrolysis. 115–133.
Electrochim. Acta 128, 248–258. Santos, J.L.C., Geraldes, V., Velizarov, S., Crespo, J.G., 2010.
Nelissen, G., Van den Bossche, B., Deconinck, J., Van Theemsche, Characterization of fluid dynamics and mass-transfer in an
A., Dan, C., 2003. Laminar and turbulent mass transfer electrochemical oxidation cell by experimental and CFD
simulations in a parallel plate reactor. J. Appl. Electrochem. studies. Chem. Eng. J. 157, 379–392.
33, 863–873. Trujillo, F.J., Safinski, T., Adesina, A.A., 2009. Solid–liquid mass
Nelissen, G., Van Theemsche, A., Dan, C., Van den Bossche, B., transfer analysis in a multi-phase tank reactor containing
Deconinck, J., 2004. Multi-ion transport and reaction submerged coated inclined plates: a computational fluid
simulations in turbulent parallel plate flow. J. Electroanal. dynamics approach. Chem. Eng. Sci. 42, 1143–1153.
Chem. 563, 213–2020. Vande Vyver, O., Nelissen, G., Weyns, G., Deconinck, J., Degrez, M.,
Nelissen, G., Weyns, G., Maciel, P., Deconinck, J., Vande Vyver, O., Godet, S., 2008. Mass transfer and current distribution on a
Deconinck, H., 2007. Numerical study of the influence of the metallic wire. Electrochim. Acta 53, 6452–6459.
anode position and the electrolyte flow on the deposition of Vázquez, L., Alvarez-Gallegos, A., Sierra, F.Z., Ponce de León, C.,
copper on a wire. Electrochim. Acta 52, 6584–6591. Walsh, F.C., 2010. Simulation of velocity profiles in a
Newman, J.S., 1991. Electrochemical Systems, second ed. laboratory electrolyser using computational fluid dynamics.
Prentice-Hall, New Jersey. Electrochim. Acta 55, 3437–3445.
Pohjoranta, A., Mendelson, A., Tenno, R., 2010. A copper Weigand, B., Ferguson, J.R., Crawford, M.E., 1997. An extended
electrolysis cell model including effects of the ohmic potential Kays and Crawford turbulent Prandtl number model. Int. J.
loss in the cell. Electrochim. Acta 55, 1001–1012. Heat Mass Transfer 40, 4191–4196.
Rivera, F.F., Nava, J.L., 2007. Mass transport studies at rotating Wragg, A.A., Leontaritis, A.A., 1997. Local mass transfer and
cylinder electrode (RCE) Influence of using plates and current distribution in baffled and unbaffled parallel plate
concentric cylinder as counter electrodes. Electrochim. Acta electrochemical reactors. Chem. Eng. J. 66, 1–10.
52, 5868–5872. Zeng, K., Zhang, D., 2010. Recent progress in alkaline water
Rivero, E.P., Granados, P., Rivera, F.F., Cruz, M., González, I., 2010. electrolysis for hydrogen production and applications. Prog.
Mass transfer modeling and simulation at a rotating cylinder Energy Combust. Sci. 36, 307–325.