Journal of Chemical Technology and Biotechnology J Chem Technol Biotechnol 79:954–960 (online: 2004)DOI: 10.1002/jctb.1066

The effect of operational parameters on theperformance of a bipolar trickle tower reactorPedro Trinidad,1† Frank C Walsh,1∗,‡ Sally A Sheppard,2§ Stephen P Gillard2

and Sheelagh A Campbell21Department of Chemical Engineering, University of Bath, Claverton Down, Bath, BA2 7AY, UK2Applied Electrochemistry Group, University of Portsmouth, St Michael’s Building, White Swan Road, Portsmouth, PO1 2DT, UK

Abstract: A bipolar trickle tower reactor (BTTR) (of 7.9 cm internal diameter and 75 cm length containing57 layers, each layer having 30 carbon Raschig rings, each of 1.25 cm outside diameter) has been studiedunder a range of operational conditions. The batch recycle mode of operation has been used for theremoval of Cu(II) ions (at an initial concentration of 50–200 ppm) from an acid sulfate solution (typically3000 cm3) at 295 K. Non-ideal flow and Peclet number values have been considered to establish the degree ofdeviation from ideal reactor flow models. Operational variables included the potential difference per layer(1.0–3.0 V), volumetric flow rate (8.3–50 cm3 s−1) and the effect of H2SO4 concentration (which increasedconductivity and lowered pH) in the electrolyte. The reactor has been shown to be best suited to thetreatment of a moderately high reactant concentration (eg 100–200 ppm) and low electrolyte conductivity.The final concentration can be as low as a few parts per million but the performance of the reactor (asjudged by the current efficiency and the rate of concentration decay) markedly decreased as the dissolvedmetal ion concentration fell. 2004 Society of Chemical Industry

Keywords: bipolar trickle tower reactor; Cu(II) ion removal; current efficiency; environmental treatment; flowdispersion; mass transport and electrode area; normalised space velocity

NOTATIONA Electrode area (cm2)

Bo Bodenstein number = Pe(Lp/L) (dimension-less)

C Concentration of tracer in the fluid(mol dm−3)

CB Bulk concentration of metal ions (mol dm−3)

C(IN) Concentration of metal ion at reactor inlet(mol dm−3)

C(OUT) Concentration of metal ion at reactor outlet(mol dm−3)

C(t) Concentration of metal ion in the reservoir attime t (mol dm−3)

C(0) Initial concentration of metal ion in thereservoir (mol dm−3)

D Dispersion coefficient (cm2 s−1)

F Faraday constant (= 96 485 C mol−1)

(C mol−1)

IL Limiting current (A)nr Number of Raschig rings per layer

km Averaged mass transport coefficient (cm s−1)

L Length of experimental section of reac-tor (cm)

Lp Length of single, bipolar electrode (cm)N Number of bipolar layers in a trickle tower

(dimensionless)Pe Peclet number = UL/D (dimensionless)Q Volumetric flow rate of electrolyte (cm3 s−1)

(Re)f Film Reynolds number = Q/[2πnr(ro + ri)ν](dimensionless)

ri Inner radius of Raschig ring (cm)ro Outer radius of Raschig ring (cm)sn Normalised space velocity (cm3 cm−3 s−1)

t Batch electrolysis time (s)U Mean linear electrolyte velocity (cm s−1)

V Volume of electrolyte in batch system (cm3)

VR Effective electrolyte volume in the reactor(cm3)

x Distance from entrance of reactor (cm)Vn Normalised volume of electrolyte (cm3)

∗ Correspondence to: Frank C Walsh, Electrochemical Engineering Group, School of Engineering Sciences, University of Southampton,Highfield, Southampton, SO17 1BJ, UKE-mail: F.C.Walsh@soton.ac.uk†Current address: Novochem 2000, Derivados Quimicos, Camino Viejo de Pliego S/N, Apartado 182, 30820 Alcantarilla, Murcia, Spain‡Current address: Electrochemical Engineering Group, School of Engineering Sciences, University of Southampton, Highfield, Southampton,SO17 1BJ, UK§Current address: Inco Europe Ltd, Acton Refinery, North Acton, London, NW10 6SN, UKContract/grant sponsor: EPSRCContract/grant sponsor: EU(Received 25 July 2003; revised version received 25 March 2004; accepted 6 April 2004)Published online 15 July 2004

2004 Society of Chemical Industry. J Chem Technol Biotechnol 0268–2575/2004/$30.00 954

Performance of a bipolar trickle tower reactor

z Number of electrons involved in the primaryreaction (dimensionless)

ν Kinematic viscosity of the electrolyte(cm2 s−1)

τR Mean residence time in the reactor =VR/Q (s)

τT Mean residence time in the reservoir =V/Q (s)

φ Overall current efficiency for metal ionremoval (dimensionless)

θ Dimensionless time = Ut/L

INTRODUCTIONThe Bipolar Trickle Tower Reactor (BTTR) consistsof a number (typically 10–60) of mutually insulated,bipolar electrode layers (usually made of carbon)inside an insulating column, with a trickle flowof electrolyte through the packing material. Thereactor has several fields of application includingorganic/inorganic electrosynthesis and, especially,pollution control.1–3 The space-time yield of thereactor and the final concentration attainable inpollution control studies depend upon variables suchas the volumetric flow rate, electrode and electrolyteconductivities, reactant concentration, the appliedvoltage and the type of electrical connection.

The deposition of copper, from Cu(II) ions in anacid sulfate electrolyte, has often been used in masstransport studies4 and is a well-known reaction inpollution control.5 The rate of Cu(II) ion removal viathe cathodic reaction:

Cu2+ + 2e− −−−→ Cu (1)

is often mass transport controlled, particularly atlow concentrations of dissolved metal.6,7 Failure toimprove the mass transport conditions within thereactor can lead to low production rates and asignificant loss of current efficiency due to hydrogenevolution as a secondary reaction:

2H+ + 2e− −−−→ H2 (2)

The reduction of dissolved oxygen provides anothersecondary reaction:

O2 + 4e− + 4H+ = 2H2O (3)

The fluid flow conditions are extremely important inreactors used for metal ion removal from dilute processstreams. Electrolytic flow within a reactor controlsthe magnitude and uniformity of mass transport.Reactor performance strongly depends upon fluid flowconditions.7,8 The influence of the flow rate on masstransport and energy efficiency and the deviation offluid flow from ideal models such as the plug flowreactor and a perfectly mixed reactor are studiedhere by stimulus–response techniques. In order toperform experimental measurements of dispersion,

tracer experiments were carried out. 5 cm3 pulsesof 3.8 mol dm−3 KCl were injected at the reactor inletand the response was analysed by a two-electrode(platinised Pt) conductivity probe at the outlet, whichwas connected via a conductivity meter to a strip chartrecorder. This method has been previously validatedand is fully described elsewhere.9,10

EXPERIMENTALThe electrolyte flow circuit is shown in Fig 1.Electrolyte was pumped from the reservoir andcirculated through a by-pass loop, which ensured goodmixing in the reservoir. The flow rate was controlledand monitored by a rotameter. A water-cooled coilcondenser was provided to help prevent excessivetemperature rise due to the joule heating. Ambientconditions were used, resulting in a temperatureof 295 ± 2 K, and the reservoir temperature wasmonitored. The electrolyte was sprayed into the top ofthe column, via a shower spray percolating downwardsthrough the packing material and returning to thereservoir. The normal mode of operation was the batchrecycle one, where a fixed inventory of electrolyte(typically 3000 cm3) was continuously recirculated.Several single pass reactor trials were carried out toinvestigate the reactor performance in this mode.

The electrolyte was a copper sulfate solution insulfuric acid, normally adjusted to pH 2 down thereactor. It was prepared using double distilled water

e

f

i

c

d

a

b

h

+ −

g

Figure 1. Electrolyte flow circuit. (a) Magnetically-coupled pump,(b) by-pass valve, (c) water-cooled glass coil condenser, (d) flowcontrol valve, (e) rotameter, (f) carbon electrode packing material,(g) electrolyte reservoir (3 dm3), (h) dc power supply (200 V/10 A),(i) bipolar trickle tower reactor containing 57 layers of carbonRaschig rings.

J Chem Technol Biotechnol 79:954–960 (online: 2004) 955

P Trinidad et al

and AnalaR grade reagents. Before application of thecurrent, the electrolyte was recirculated for 15 min toensure a stable, trickle flow down the reactor packingand to allow the temperature to stabilise at 295 K.An in-house, direct current, power supply was usedto carry out the experiments at a constant reactorvoltage; the cell current and cumulative electricalcharge were monitored. The Cu(II) concentration inthe electrolyte was established by removal of small(typically 2–5 cm3) samples of the electrolyte, followedby dilution to the range 1–4 ppm, then atomicabsorption spectrophotometry analysis (air–acetyleneflame). The analytical technique established thedissolved copper level to an accuracy of ±4%. Aftereach experiment, copper deposits were removed fromthe reactor packing using a recirculated stream ofair-saturated background electrolyte. Two methodswere used to clean-up the tower packing. In thefirst method, the polarity of the tower was reversedresulting in anodic dissolution of the copper deposits;to avoid a fresh deposition during this process, theelectrolyte loop was opened and diverted to drain.In the second method, electrolyte was sprayed downthe tower and open-circuit dissolution of the copperdeposit via oxygen reduction as the cathodic reactionin the corrosion process was allowed to occur. Aneffective cleaning solution was found to be aerated10% sulfuric acid, with a 20 min cleaning period.

The glass (QVF) column tower had a circular cross-section of 49 cm2 and was 75 cm high with an internaldiameter of 7.9 cm. It comprised 57 layers of carbonRaschig rings; each ring was smooth on its exteriorand hexagonally fluted internally, the mean internaland external diameters being 0.65 cm and 1.25 cm,respectively. Each horizontal carbon layer comprised30 Raschig rings of 1.25 cm height. Carbon layerswere separated by a high porosity (c 90% volume),monofilament polyester mesh of thickness 0.05 cm.Perforated carbon plates were used as current feedersat the top and bottom of the tower, resulting in a totalof 58 bipolar layers. At the top of the tower, electrolyteentered the reactor via a spray distributor, a randomarrangement of ceramic Raschig rings being used tofurther disperse and condition the flow prior to thecarbon/polyester packing layers.

THEORYIt is well known that the rate of removal of metalions from solution is often mass transport controlled,particularly in the case of low levels of dissolvedmetal.11 This situation may be characterised bythe limiting current for metal deposition, whichexpresses the maximum rate of reaction underconvective–diffusion control. The limiting currentmay be defined as:

IL = kmAzFCB (4)

where km is the mass transfer coefficient, A is thecathode area, z is the number of electrons involved

in the metal deposition reaction and CB is the bulkconcentration of metal ions in the electrolyte.

There are several routes to the determination of theperformance factor kmA from concentration data.11

Two methods have been used in this work.

(i) Measurements of steady state conversion over thereactor and the use of the design equation for asingle pass reactor under plug flow conditions.8

ln[

C(IN)

C(OUT)

]= − kmA

2.3Q(5)

(ii) The use of a batch reactor system to generate datain the form of metal ion concentration history,followed by the application of a plug flow reactorin batch recycle model, which involves the designequation:8,9

ln[

C(t)

C(0)

]= − t

τR

[1 − exp

(−kmA

Q

)](6)

where t is the batch time and τR is the nominalresidence time in the reactor. The term in the squarebrackets on the right hand side of this equation iseffectively the fractional conversion of reactant perpass through the reactor.

Under the experimental conditions, the residencetime in the reactor is very much smaller than that inthe reservoir (τR << τT). A simple batch model maythen be adopted12,13 and eqn (6) simplifies to:

ln[

C(t)

C(0)

]= −kmA

Vt (7)

where V is the volume of electrolyte within the batchsystem. According to this equation, a plot of thenormalised concentration of metal ions vs time willbe linear and a slope of kmA/V if the factors kmAand V are constant. It has been shown that the simplebatch model is a good approximation under conditionsof high flow rate and a low conversion per pass, ie whenthe fractional conversion per pass through the reactoris small.12,13 Under the experimental conditions, theapplication of eqn (7) is readily justified.

The BTTR is normally modelled as a plug flowreactor, but non-ideal flow is likely and deviationfrom the ideal model must be studied. Usingstimulus–response techniques, curves that describethe concentration–time function of a tracer in theexit stream of the vessel can be established. Theapplication of a dispersion model allows the calculationof a Peclet number that can be used to describeflow dispersion within the reactor.14,15 In this paper,the axially-dispersed plug flow model has been used.This model considers that experimentally observeddispersion takes place only in the direction of fluidflow and is described by the equation:

∂C∂θ

=[

DUL

∂2C

∂x2

]− ∂C

∂x(8)

956 J Chem Technol Biotechnol 79:954–960 (online: 2004)

Performance of a bipolar trickle tower reactor

The Peclet number can be calculated by solution ofthis partial differential equation but Levenspiel andSmith15 have described a simple strategy based on thefirst moment and second moment of the curves andthe Peclet number. The procedure used here followsthat used in other works.10,16

RESULTS AND DISCUSSIONStudies were conducted to determine the hydrody-namic behaviour of the reactor.1,12 The mean resi-dence time in the reactor may nominally be calculatedfrom the expression τR = VR/Q, but a more realisticvalue can be calculated from the impulse–responsetechnique10,16 which considers non-ideal flow. Suchmean residence time values were calculated from thefirst moment of the curves obtained from tracer exper-iments as a function of the film Reynolds number(Fig 2). The tracer technique leads to higher residencetime values since it considers tailing phenomena ratherthan the nominal, averaged values of reactor residencetime predicted by the calculation of VR/Q values whichpresume ideal, plug flow behaviour within the reactor.The incorrect assumption of ideal plug flow condi-tions does not consider the practical phenomena ofchannelling and internal circulation of flow within thereactor. The data in Fig 2, together with a knowl-edge of the reactor geometry, allowed the apparentdeadtime in the reactor to be calculated as 2.5 s at avolumetric flow rate of 50 cm3 s−1.

The figures of merit derived from the meanresidence time, such as the space-time or spacevelocity,1 should be calculated by considering theexistence of stagnant regions and other non-ideal flowpatterns evaluated from tracer experiments. The filmReynolds number, which characterises fluid flow downthe Raschig rings, can be defined16 as:

(Re)f = Q2πnr(ro + ri)υ

(9)

15 20 25 30 35 40 45 508

10

12

14

16

Res

iden

ce T

ime,

τ / m

in

Reynolds Number, Re

, actual values, nominal values

Figure 2. Mean residence time in the reactor as a function of filmReynolds number, showing actual values calculated from analysis ofthe first moment of KCl concentration vs time curves and nominalvalues calculated from the relationship: τR = VR/Q.

which assumes that the electrolyte film thickness alongthe wetted perimeter of the Raschig ring is equal tothe film thickness along the inner wetted perimeter.

The Peclet number, which characterises axial flowdispersion, can be defined as:

Pe = ULD

(10)

A non-linear relationship between the Peclet numberand film Reynolds number was found in the presentwork. Peclet numbers of 14.2, 14.9, 17.4 and 20.3were found at film Reynolds numbers of 29.3, 33.2,41.1 and 46.0, respectively. The higher Pe valuesexperienced at higher Re indicate a lower degree ofdispersion at the higher flow rates.

In order to describe the mixing and dispersionover a single layer, the Bodenstein number maybe considered:

Bo = PeLp

L(11)

where Lp is the length of the electrode layer (ie theheight of a Raschig ring) and L is the overall length ofthe reactor. Bo values of 0.23 and 0.33 are obtainedhere, which are similar to those found by Fleischmannand Ibrisagic16 for a reactor volume approximately fourtimes smaller than the present one. The Bodensteinnumber, Bo, represents mixing over the length of asingle Raschig ring. Despite the differences in thereactor geometry between this work and previousliterature,16 there is no effect of reactor length whenthe Bodenstein numbers are compared. On the otherhand, the Peclet number describes dispersion over thereactor and differences are apparent between the twosets of reactors.

When the reactor was used to remove Cu(II) ionsfrom a solution, three regions can be considered in theconcentration vs time curves (Fig 3):

(i) an exponential region (representing a first orderdecay) which occurs at short times,

(ii) an intermediate region, where the curve is nolonger exponential and its slope decreases withtime and

(iii) a region, at longer times, where the concentrationdecay is slight and the metal ion concentrationtended to remain at a constant value, usually inthe order of 1 ppm, at sufficiently long times.

This behaviour can be attributed to a progressivechange in the rate-determining process and toredissolution of copper. It is well established thatlonger batch times (and hence lower reactantconcentrations) lead to a decrease in the side activearea of the cathodic zones of the Raschig rings.3 Inorder to optimise the reactor and establish the bestconditions for copper removal, several experimentswere carried out under diverse operational conditions.Figure 4 shows concentration decay curves for variousinitial Cu(II) ion concentrations. The rate of decayis seen to decrease at lower initial Cu(II) ion levels,

J Chem Technol Biotechnol 79:954–960 (online: 2004) 957

P Trinidad et al

Time, t / min

0 40 80 120 160 200

ln (

Con

cent

ratio

n, C

(t) /

ppm

)

0.5

1.5

2.5

3.5

4.5

Figure 3. Concentration decay curve for removal of Cu(II) ions fromsolution, showing the three typical zones observed.

0 5 10 15 20 25 30 35 40 450

50

100

150

200

Con

cent

ratio

n, C

(t) /

ppm

Time, t / min

, 50 ppm, 100 ppm, 150 ppm, 200 ppm

Figure 4. Concentration decay curves for various initial Cu(II) ionconcentrations.

although the final Cu(II) levels are similar. Semi-logarithmic plots of the normalised concentration vstime were used to calculate the product kmA calculatedvia eqn (4) and it is seen in Table 1 that the valuesdepend upon the initial Cu(II) concentration.

The overall current efficiency for copper deposition(φ), ie the product yield based upon the electricalcharge passed during an electrochemical reaction, wascalculated from:

φ = �CVzF∫NI.dt

(12)

Table 1. Calculated kmA values for various operational conditions

(Q = 33.3 cm3 s−1)

Initial Cu(II)concentration,C(0) (ppm)

Voltage per layer(V) pH of electrolyte

kmA(cm3 s−1)

50 3.0 2.0 4.6100 3.0 2.0 4.7150 3.0 2.0 5.4200 3.0 2.0 5.4100 1.5 2.0 3.8100 2.0 2.0 4.2100 2.5 2.0 4.5100 3.0 2.0 5.3100 3.0 1.0 4.6100 3.0 1.3 4.9100 3.0 7.0 4.9

where N is the number of electrode layers whichexperience the passage of current, I. Table 2 showsthat φ falls with time and is diminished at lower initialconcentrations of Cu(II). The current efficiency valuesare not very high, due to parallel reaction, especiallyhydrogen evolution and redissolution of copper. Ata high level of dissolved copper, a thicker depositis found at the cathode that can increase the masstransport conditions due to a greater hydrodynamicshear at the roughened electrode surface, togetherwith an increase in the active surface area.18

The influence of flow rate on concentration decaywas studied. The kmA value increases at higherflow rates with a complex flow rate dependence.Typically, kmA/V values of 1.5 × 10−3 s−1 wereachieved at a volumetric flow rate of 8.3 cm3 s−1,with the performance rising to kmA/V values of4.5 × 10−3 s−1 at a flow rate of 50 cm3 s−1. Theuse of an excessive flow rate can lead to lowcurrent efficiency values due to increased bypassing ofcurrent between neighbouring electrode layers underelectrolyte conditions which tend to flooded flow ofthe reactor packing1 (Table 2).

The normalised space velocity describes the capacityof the reactor, being dependent upon the productof mass transport coefficient and electrode per unitreactor volume. The normalised space velocity wascalculated from the expression:1,12

sn = kmA2.3Vn

(13)

where Vn is the normalised volume of electrolyte.12

This produces values of 1.7 cm3 cm−3 h−1 and

Table 2. Calculated percentage current efficiency for copper deposition

Time,t (min)

C(0) = 50 ppmQ = 33.3 cm3 s−1

C(0) = 100 ppmQ = 33.3 cm3 s−1

C(0) = 200 ppmQ = 33.3 cm3 s−1

C(0) = 100 ppmQ = 8.3 cm3 s−1

C(0) = 100 ppmQ = 50.0 cm3 s−1

10 3.0 7.9 9.5 14.6 13.520 2.1 4.7 7.0 9.5 7.430 1.8 3.3 5.1 6.4 5.540 1.6 2.8 4.1 5.6 4.250 1.4 2.3 3.4 4.3 3.2

958 J Chem Technol Biotechnol 79:954–960 (online: 2004)

Performance of a bipolar trickle tower reactor

5.0 cm3 cm−3 h−1 for flow rates of 8.3 cm3 s−1 and50 cm3 s−1, respectively. For a batch undivided reactorutilising a rotating cylinder electrode for the removalof cadmium (500 ppm Cd(II) at pH 7) the normalisedspace velocity was 3.9 cm3 cm−3 h−118 in a dividedlaboratory reactor utilising 5.0 cm × 5.0 cm × 1.2 cmreticulated vitreous carbon (RVC) cathode for theremoval of 10 ppm Cu(II) ion level down to levels of0.1 ppm in 0.5 M Na2SO4 at pH 2 had a normalisedspace velocity in the range of 50–450 cm3 cm−3 h−1.18

By comparing the values of the normalised space veloc-ity for the BTTR, it is shown that the RVC valuesare considerably higher and the value for cadmiumremoval is comparable.

The effect of applied voltage was studied for sin-gle pass and batch recirculation modes of operation(Fig 5). The kmA value was calculated from the differ-ence between the inlet and outlet concentrations12 andfrom the slopes of the two linear decays; the differencein kmA values was less than 10% in comparison withthe batch recycle model.19 The performance of thereactor was studied for applied voltages in the range1.5, 2, 2.5, 3 volts per layer, corresponding to theincreased kmA values shown in Table 1. In general,fair agreement (error <10%) was obtained for thekmA values extracted from the results of single passand batch recycle experiments. Batch recycle data wereused to determine the reported kmA values due to thelower errors involved. The increase of the performanceof the reactor for high values of applied voltage is dueto an increase in the active electrochemical area andmass transport. It is known that the potential distribu-tion over a single layer depends upon applied voltage.1

A larger applied voltage will reduce the inactive zonebetween cathode and anode, allowing a larger activearea. At higher reactor voltages, however, there wasan increased tendency for degradation of the carbonpacking material. The resultant loss of particulate car-bon from the anodic areas of Raschig rings tended tocause partial blocking of the spray nozzle at the topof the tower, leading (in the absence of a filter) tosome problems of non-uniform fluid dispersion at thereactor inlet.

0 2 4 6 8 10 12-2.0

-1.5

-1.0

-0.5

0.0

0.5

C(out)

C(in)

ln [

C(t

)/C(o

)]

Time, t / min

Figure 5. Natural logarithm of normalised inlet and outlet Cu(II)concentrations vs time in the batch recycle mode of operation(Q = 33.3 cm3 s−1. C(0) = 100 ppm).

At low values of potential difference per electrodelayer, eg at <1.3 V per layer, there is an insufficientdriving force for electrochemical reaction and nodepletion of Cu(II) ions in the solution is observed.Higher values of potential applied gave rise to aprogressive decrease in current efficiency (Fig 6) dueto an increase in the rate of anodic oxygen evolutionallowing chemical oxidation of the deposit and atendency for hydrogen evolution to physically dislodgepowdery deposits of metal from the cathode zones,followed by dissolution of the free metal particles.

The influence of pH was studied in the rangepH 7 to pH 1 via addition of sulfuric acid to theelectrolyte, leading to a change in the conductivityfrom 0.77 mS cm−1 to 32 mS cm−1, respectively. Animproved performance is obtained in neutral solutionin terms of a slightly higher rate of concentrationdecay (Table 1) and current efficiency (Fig 7). It mustbe realised, however, that the use of pH values greaterthan approximately pH 4 can result in less pure copperdeposits, due to chemical precipitation of copper ions(as, eg copper (II) hydroxide) occurring in additionto cathodic deposition of copper metal. Low valuesof pH produce low current efficiency due to currentby-pass through the electrolyte between neighbouringlayers.

0 10 20 30 40 50 600

5

10

15

20

25

30

35

40

% C

urre

nt E

ffic

ienc

y

Time, t / min

, 2.0 V per layer, 2.5 V per layer

, 3.0 V per layer

Figure 6. Current efficiency for Cu(II) ion removal vs time for severalapplied voltages (C(0) = 100 ppm. Q = 33.3 cm3 s−1).

0

2

4

6

8

10

12

14

16

18

20

% C

urre

nt E

ffic

ienc

y

0 10 20 30 40 50 60 70

Time, t / min

, pH 7.0, 0.44 mS cm-1

, pH 2.0, 7.1 mS cm-1

, pH 1.3, 26.4 mS cm-1

, pH 1.0, 32 mS cm-1

Figure 7. Current efficiency for Cu(II) ion removal vs time, forsolutions having various pH and electrolytic conductivity values(C(0) = 100 ppm).

J Chem Technol Biotechnol 79:954–960 (online: 2004) 959

P Trinidad et al

CONCLUSIONS1. Stagnant regions in the bipolar trickle tower reactor

resulting in a lowering of the mean residence time.2. The Peclet number has been calculated over a range

of film Reynolds numbers and a linear relationshipbetween Pe and Re has been found, in agreementwith previous work on a smaller scale.16

3. It has been shown that the BTTR can be used formetal recovery from solutions with a final low levelconcentration in the order of 1 ppm under suitableconditions.

4. The reactor’s performance is better for high levelsof copper such as 200 ppm, for this concentrationa high current efficiency was also found.

5. The use of a moderate flow rate of electrolyte helpsto secure a reasonable rate of concentration decaybut high flow rates can produce flooding in thetower, resulting in appreciable bypass currents anda loss of current efficiency for metal deposition.

6. An increase in the applied voltage resulted in afaster concentration decay but a significant penaltyin energy consumption was incurred.

7. Under constant reactant concentration and flowconditions, the applied voltage can increase themass transport coefficient, km, due to improvedlocal turbulence arising from gas agitation; anincrease in the electroactive area is also expected.The factor kmA and related parameters such as thenormalised space velocity are useful performanceindicators since the independent evaluation of thekm and A is difficult in three-dimensional electrodereactors.

8. The application of 2.0 V per layer appears tobe a reasonable compromise under experimentalconditions. The use of a higher cell voltage provideda modest decrease in batch electrolysis time butincurred a power appreciable penalty and a lowercurrent efficiency due to higher rates of hydrogenevolution.

Traditional column packing materials such asRaschig rings have been used for many years inlaboratory trickle tower reactors3 but more attractive,higher electrode area, materials include porous, three-dimensional meshes, perforated plates, fibre mats,foams and particulate beds, which are consideredelsewhere.1,5,17 The tower described in this paper hasseen semi-continuous, laboratory use for over 22 yearswith a typical duty of 500 h per year. The reactor stillperforms to within 4% of its original kmA value forcopper ion removal and has experienced an overallcarbon weight loss of approximately 4% (based ondry column packing weight). The top (anodic feeder)has been remanufactured three times and a carbonfilter has been incorporated into the flow circuit. Forimproved safety, a central (back to back) anodic feederhas been developed, the top and bottom feeders of thetower being taken to ground potential.6

ACKNOWLEDGEMENTSThese studies have been supported, in part, by anEPSRC award (to SAS) and an EU grant (to PT).The authors are grateful to Mr Tom Young for hisskilful construction of the dc power supply.

REFERENCES1 Walsh FC, A First Course in Electrochemical Engineering. The

Electrochemical Consultancy, Romsey, pp 272 (1993).2 Lin M, Wang Y and Wan C, A comparative study of electro-

chemical reactor configurations for the composition of coppercyanide effluent. J Appl Electrochem 22:1197–1200 (1992).

3 Ehdaie S, Fleischmann M and Jansson REW, Application ofthe trickle tower to problems of pollution control. I. Thescavenging of metal ions. J Appl Electrochem 12:59–67 (1982).

4 Pletcher D, Whyte I, Walsh FC and Millington JP, Reticulatedvitreous carbon cathodes for the metal ion removal fromprocess streams. Part 1: Mass transport studies. J ApplElectrochem 21:659–666 (1991).

5 Robinson D and Walsh FC, The performance of a 500 amprotating cylinder electrode reactor. Part 1: Current–potentialdata and single pass studies. Hydrometallurgy 26:93–114(1991).

6 Walsh FC and Reade GW, Electrochemical techniques for thetreatment of dilute metal ion solutions, in EnvironmentalOriented Electrochemistry, ed by Sequeira CAC. Elsevier,Amsterdam, pp 3–44 (1994).

7 Walsh FC, Design and performance of electrochemical reactorsunder mass transport controlled conditions. Bulletin ofElectrochemistry 6:283–293 (1990).

8 Levenspiel O, Chemical Reaction Engineering, 2nd edn. JohnWiley, New York (1972).

9 Harrell JE and Perona JJ, Mixing of fluids in tanks of largelength-to-diameter ratio by recirculation. Industrial ProcessDesign and Development 7:464–468 (1968).

10 Trinidad P and Walsh FC, Hydrodynamic behaviour of theFM01-LC reactor. Electrochim Acta 41:493–502 (1996).

11 Ford WPJ, Walsh FC and Whyte I, Simplified batch reactormodels for the removal of metal ions from solution. I Chem ESymp Ser 127:111–126 (1992).

12 Walsh FC, Determination of the normalised space velocity forcontinuous stirred tank electrochemical reactors. ElectrochimActa 38:465–468 (1993).

13 Walker ATS and Wragg AA, The modelling of concentra-tion–time relationships in recirculating electrochemical reac-tor systems. Electrochim Acta 22:1129–1134 (1977).

14 Levenspiel O and Bischof K, Reaction rate constant may modifythe effects of backmixing. Ind Eng Chem 53:313–314 (1961).

15 Levenspiel O and Smith W, Notes on the diffusion-type modelfor the longitudinal mixing of fluids in flow. Chem Eng Sci6:227–233 (1957).

16 Fleischmann M and Ibrisagic Z, Performance of bipolar tricklereactors. J Appl Electrochem 10:169–172 (1980).

17 Pletcher D and Walsh FC, Porous, three-dimensional elec-trodes, in Electrochemistry for a Cleaner Environment, ed byGenders D and Weinberg N. Electrosynthesis Co Inc, NewYork, pp 51–100 (1992).

18 Grau JM and Bisang JM, Electrochemical removal of cadmiumusing a batch undivided reactor with a rotating cylinderelectrode. J Chem Technol Biotechnol 76:161–168 (2001).

19 Walsh FC, Pletcher D, Whyte I and Millington JP, Electrolyticremoval of cupric ions from dilute liquors using reticulated vit-reous carbon cathodes. J Chem Technol Biotechnol 55:147–155(1992).

960 J Chem Technol Biotechnol 79:954–960 (online: 2004)