of Califor,

of Cereal

allinger

,ican Socie

es, Pro- p.4.

:. Fourth

ber, Paper

Ion, Appei

* hesent address: Engineering Service Division, E.I. du Pont de Nemours & Co. Inc., Wilmington, Del. 19898, U.S.A.

Resource Recovery and Conservation, 2 (1976) 39-55 39 0 Elsevier Scientific Publishing Company, Amsterdam - Printed in The Netherlands

THE RECOVERY OF SOLUBLE COPPER FROM AN INDUSTRIAL CHEMICAL WASTE

K.B. KEATING and J.M. WILLIAMS*

Engineering Technology Laboratory, Engineering Research and Development Division, E.I. d u Pont d e Nemours & Co. Inc., Wilmington, Del. 19898 (U.S.A.)

(Received 7th January, 1976)

ABSTRACT

A novel electrochemical device was used to recover copper metal from an inorganic chemicals process effluent. In a sidestream pilot-plant demonstration, the copper content was continuously reduced from an average of 20 mg/l to less than 2 mg/l. The basic concepts, laboratory testing, and pilot-plant demonstration are described.

IY-L AODUCTION

General

Future water-pollution regulations may require effluent concentration limits for heavy metals more restrictive than the Public Health Service drinking water standards. Such regulations would severely limit the allowable levels of copper, lead, silver, chromium, mercury, and manganese in plant effluents. Parallel with this development is the increasing scarcity of the world's mineral r!" -:!rces including metals such as these.

The simplest form of any pollution abatement-resource recovery system would have two key features: reduction of the contaminant content of an effluent to a level meeting regulatory standards, and recovery of the contaminant in concentrated form suitable for recycle or sale.

The technologies of chemical precipitation, solvent extraction, ion exchange, cementation, and reverse osmosis have all been considered for treating specific

, effluents containing such metals. There are practical limitations, however, in i aP$ving these methods: high cost, inadequate heavy-metals rerroval to meet ' h. .ening standards, lack of information on acceptable ultimate disposal of 1 removed metals, or insufficient technical data on how a particular treatment

method is affected by other components in the effluent. Moreover, except 1 _for reverse osmosis, these methods generally require continued addition of

I

40

treatment chemicals, which increase the quantity of contaminants in both solid and aqueous discharges and complicate the problem of ultimate disposal.

Electrolytic reduction of metals to remove them from solution avoids the addition of treatment chemicals. This technique is not new and various kinds of devices have been reported. For example, small scale devices have been constructed utilizing beds of nonconducting particles [ 11 , porous carbon electrodes [2] , packed bed graphite electrodes [3] and fluidized bed electrodes [4]. A comparison of several electrochemical reactor types for copper removal has also been made [ 51 and more recently, a cell utilizing noncon- ducting fluidized beds and mesh electrodes has been described [6].

For several years a program has been underway to develop electrochemical technology for removing traces of certain heavy metals from chemical industry effluents. A major emphasis has been a concept called extended-surface electrolysis (ESE), which has the potential advantage of metallic-ion removal. with separation and recovery of the metal in a useful form. This new technology removes the contaminant metals by electroplating them onto a specially constructed flow-through electrode. The procedure is particularly applicable in the low concentration range of less than 100 mg/l of contaminant metal.

in the waste-treatment field because it is extremely inefficient at low metal concentrations. This limitation is illustrated by the dashed curve in Figure 1 for the reduction of copper ions at the cathode of an electrolytic cell. The plot, taken from data obtained in our laboratory, shows coulombic efficiency (fraction of current effective for metal deposition) as a function of copper-ion concentration in an acidic solution. The efficiency drops off sharply below 10,000 mg/l which is why conventional electroplating is generally conducted at concentrations above 20,000 mg/l. The concentration of heavy metals in industrial wastes, however, usually ranges from about 500 mg/l down to

I

Conventional electrolysis with planar electrodes has found no direct applicatic

Cathode: Cut' + 2 e - C ~ '

I I

---

PLATING

1 lo5

Copper Conc., mgl lit Fig. 1. Effect of concentration on electrical efficiency in metals reduction.

10 J

low

Bas I

Earl catk thrc accc effil si i rf ttie deci

E! figu pro-

con the imp low satir fror

con-

EXt# cell sep: Thi

tnu. of a Thi: 50 (

fro] at t 1

of c S'ql I- op and SUC! to F

stac

Ma I

ts in both nate disposal. i avoids the rarious kinds Lave been ds carbon bed electrodes

OPPer ig non-con- 3 3 . ctrochemical Lmical industry surface -ion removal. new technolog pecially y applicable nant metal. direct applicat

t low metal ! in Figure 1 c cell. The plot ‘ficiency of copper-ion

,rply below ly conducted ry metals in down to

1.

41

10 mg/l. The objective is to obtain an effluent concentration of 5 mg/l or lower, depending on the species.

&sic concept

Early in these studies it was recognized that the rate-limiting step in the catliodic reduction of cupric ions at an electrode was the diffusion of ions through the dilute solution to the electrode. This mass-transfer limitation accounts for the behavior shown in Figure 1. To extend high coulombic efficiency to lower concentrations the total metal migration to the electrode surface must be increased. There are two ways to accomplish this: increase the surface area of the electrode; or improve the mass-transfer coefficient by d t ,:reasing the thickness of the ionic diffusion boundary layer.

Both approaches were used. Starting with planar electrodes, several con- figurations were evolved to improve electrical efficiency and the most useful proved to be a packed-bed ESE cell. Figure 1 shows its improved performance compared with a planar electrode cell. The enhanced efficiency of the new configuration is due to the combined effects of hydrodynamic reduction of the diffusion boundary layer and increased cathode surface area. Still further improvement might be possible, but since the contaminant concentration is low, little must be removed. Thus, a 30 percent coulombic efficiency is sa: isfactory and attempts at further improvements did not appear warranted from an economic viewpoint.

Extended-surface electrolysis (ESE) cells. As shown in Figure 2, the ESE spiral cell is a sandwich construction containing a fixed, “fluffy” cathode, a porous separator layer, a screenlike material as the anode, and another separator layer. This sandwich is rolled up into a spiral “jelly roll” and inserted into a pipe. Such cells have an open structure with a void volume of from 93 to 95 percent, thus providing low resistance to fluid flow. The cathode is usually constructed of .i knitted stainless steel niesh with a filament size of from 0.05 to 0.13 mni. This type of structure affords a high surface area, in the range of from 30 to 50 cm2/cm3 of volume. These cells have been operated at fluid velocities of from 1 to 10 cm/s and apparent current densities of from 1 0 to 200 mA/cm2 at the separator. The basic ESE structure is patented [7].

of cells can be stacked as modules so that the fluid flows through them sequentially. Figure 3 shows a stack of five modules each having its own d.c. power supply, a pump for circulating fluid through the system, rotameter, a~ -1 other hardware. Under typical operating conditions of from 10 to 20 l/min, such a stack of modules can reduce the concentration of copper in an effluent to about 10 percent of that entering the unit. Additional cells added to the stack will further reduce the metal concentration in the effluent.

To remove a large fraction of contaminant metals from an effluent, a number

Mathematical model. Concurrent with the studies in the laboratory, a

Side

Fig. 2. Spiral ESE cell.

!.

43

Fig. 3. Stacked cells with power supplies and accessories.

44

mathematical model was develo chemical device. The model, wk :h includes competing reactions at both electrodes, mass-transfer, and electrokinetic effects, is quite complex in its complete form and requires machine computation. Neglecting the electrokinetj effects and assuming a mass-transfer-limiting case , a simplified equation results for interpretation of experimental data:

d to describe the behavior of such an electro.

r -zak i

Here Caut and Ci, are the concentrations of the contaminant metal exiting and entering the cell, z is the module length, a is the specific cathode surface area (per unit volume), k is the mass-transfer coefficient, and u is the super- ficial fluid velocity. The simplified model has the following implications :

The fraction of pollutant removed by an ESE module is independent of concentration levels; under mass-transfer-limiting conditions, such a device will operate just as effectively at 1 ppm as at 1000 ppm.

device and the available cathode area.

its validity was demonstrated in the laboratory. Figure 4 shows representative operating data obtained with an ESE module on H2S0,-acidified CuS04 solution (pH = 2). Concentration does indeed decrease exponentially with the length of the device, or the number of modules in series, as predicted by the model. For instance, at a low flow rate, the copper concentration, determined by atomic absorption, can be reduced from 20 mg/l to 0.5 mg/l with only

The effluent concentration decreases exponentially with the length of the

Before the model was used to estimate optimum size for plant-scale systems,

20

10 8

6

- 4 c .- -

v * 8.2 c m

v = 4.9

I sec

v = 1 . 6 "

I I I I

5

\ 0.6

0.4 0 1 2 3 4

M o d u l e N u m b e r Fig. 4. Copper removal as a function of ESE unit length (number of modules).

fou. timg

7 sulf tre; poi

1 levi

TA

'ILt

hlo No

-

- 0 1 2 3 1

SC fc tl n.

a U

45

four modules. Figure 4 also indicates the effect of fluid velocity, or holdup time, on effluent copper concentration.

These laboratory data, obtained with synthetic waste (acidified copper sulfate solution), were then compared with data for two actual process wastes treated similarly in an ESE system. Table 1 lists the results. The important points are:

level of 1 mg/l. Effluent copper concentration can readily be reduced to below a regulatory

ESE performance is basically the same for both real and synthetic wastes.

TABLE 1

'lkeatment of wastes __ Module No.

Cu concentration in wastes, mg/l

Synthetic Plant A Plant B _____ ~ _ _

0 20.0 45.5 15.5 1 8.2 15.5 5.6 2 3.4 5.4 2.8 3 1.3 2.1 1.7

0.9 0.7 -_- 0.6

Scale-up considerations. Figure 5 shows a practical equipment arrangement for the ESE concept. Waste is fed continuously to an equalization tank; from there it is pumped through the ESE unit, then to the outfall after the contami- nant metals have been removed. About once a day, waste flow t o the ESE unit is interrupted for an hour or so, and the accumulated metals in the cells are stripped out by circulation of acidic leachate through the cell. This

Outfall

Metal For Sa'e D. C. Power supply

3 1 ,.L- , + [TT,

E S E Unit

Waste ( Intermittent I

\ - Recovery Unit Waste Hold - u p

Leachate Tank Fig. 5. Practical scheme for metals removal.

46

leachate, with a high concentration of metal ions, is then suitable for metals recovery by conventional electrolysis.

the size of ESE systems for treating specific plant effluents under idealized conditions of constant inlet composition and absence of suspended solids. The required sizes are quite small, particularly when compared with those for other technologies, such as chemical precipitation which requires clarifiers many times the size of the ESE units.

requiring at 1 O O : l concentration reduction, the electrical energy to drive the rake on a clarifier for alkaline precipitation treatment would be equivalent to that for ESE removal of the copper. The clarifier in this example would be 35 f t (10.7 m) in diameter and 16 f t (4.9 m) deep and would require a 5 HP motor for the rake, utilizing approximately 4 kW. This is about the same power as required for removal of copper by electrolysis at 40 percent electrical efficiency.

Using the mathematical model, the laboratory data was scaled up to estimate

It was calculated that, for a 190 l/min waste containing 50 mg/l copper and

SITE DEMONSTRATION

Following the laboratory studies, a movable 11 to 12 l/min demonstration unit was assembled to develop information on materials of construction and long-term operating characteristics with an actual waste. The particular waste selected subjected the demonstration unit to considerable variability in important parameters, such as copper concentration, complexing agents, organics, etc. The waste was characterized by taking grab samples over a period of time. The ranges of the critical parameters are listed in Table 2.

TABLE 2

Parameters of chemical process waste

Flow rate 3000 l/min Soluble copper 1-55 mg/l Suspended solids 40-2600 mg/l PH 1-9.6 Chloride ion 600-3500 mgfl Other inorganics Organics

NH,, SO:-, A13+, S, traces of heavy metals Water-immiscible solvents, surfactants, organic nitrogen compounds, organic acids

I_

Besides illustrating the variability of the waste (batch process), these findings suggest a complex chemistry for the copper ion, since chloride, ammonia, and organic acids can complex with copper. The high chloride ion concentration and low pH combination indicated potential materials problems, and, of course, the high suspended solids loading suggests the necessity of some solids removal as a pretreatment.

47

Pilot unit

Figure 6 shows a simplified flow sheet of the pilot facility. Operation was as follows: the waste was continuously pumped into the pH tank where the PH was adjusted to 3 for optimum operation; the waste then passed through a wire-wound stainless steel screen filter, a sand filter with a recirculation loop, a “cell filter’.’ (a deenergized ESE cell), and the ESE column where the offending metal was plated out. When the pressure drop across the ESE column

E ’HI SE Column

Outfal l

5d ’ H2S04 50 % NaOH

h u Cell Filter

Feed Pump Screen filter ” Sand Fi l te r

Fig. 6. Flowsheet for 11-12 l/min ESE plant demonstration.

ible for metals

h those for othe

jemonstration struction and articular waste

rogen compounds,

of some solids

increased to a preset value because of deposited copper in the cells, the column was leached with 20 percent HN03. After several cycles, when the concentration of the copper in the leachate reached an appropriate level, copper was re- covered by conventional electrolysis in the recovery cell. Figure 7 shows the pilot unit in place.

Copper removal from waste

Figure 8 shows plots of copper concentrations versus column length at several different times during the 600-h demonstration a t 11 to 12 l/min and at pH = 3. There was little variation in the rate of copper removal with time, although, occasionally, a line of smaller slope (lower rate) was obtained. Also shown, for comparison, is the line for “ideal” solution (CuS04 in H,S04, pH = 3). It can be seen that the copper removal rate from this latter environment is greater than in the actual process waste. Runs using “ideal” solution were also made with the ESE column before and after the on-site operation. There w:’ no observable deterioration of performance over this period as determined bh the rate of copper removal or cell electrical characteristics.

Figure 9 shows operating data with the actual process waste at various pH values, again with the “ideal” CuS04 solution shown for comparison. The pH and hence the extent and kind o f copper complexation have a great effect on the copper removal rate. For this particular waste, a pH of 3 seems about

c

49

lo E

CU SO4 ( IDEAL I

I I I I I

ESE Column Length, cm.

I I 50 75 0 25 . 1

Fig. 9. Effect of pH on copper removal rate.

optimum. At this pH value, 90 percent removal of the copper by the 90 cm ESE column is assured. Lower pH values did not enhance the copper removal rate.

Two possible explanations arise for the effect of pH on the copper removal rate:

(1) A kinetic resistance (slow step other than electron transfer) which d + pends on the actual species depositing on the electrode surface, may control the rate of the process. The nature of the species, in complex industrial wastes, would be dependent o n the pH.

(2) The diffusivity of the diffusing species may be reduced if the copper is strongly complexed, as it may be in this waste (ammonia, organic acids present). Although an exhaustive discussion is beyond the scope of this paper, a starting point for this argument is the well-known j factor for mass transfer [8].

j , = (Sh) (Re)-’ (SC) -”~ = - = f (Re) v PD

where Sh = Sherwood number Re = Reynolds number Sc = Schmidt number k = Mass transfer coefficient of species, cm/s V = Superficial velocity of fluid, cm/s D = Diffusivity of species, cm2/s

p = Density of fluid, g/cm3 p = Viscosity of fluid, g/cm s

The Reynolds number is constant for a given geometry and operating con- dition so that the mass-transfer coefficient h is proportional to the 2/3 power of the diffusivity, D, of the copper species. The magnitude of D depends strongly on the species in which the copper is present, Le., the complex in which the copper finds itself. Thus the mass-transfer rate of copper to the cathode will also depend on the pH of the solution when complexing species like ammonia and organic acids are present.

Leaching copper from ESE cell

An important factor in favorable economics for ESE operation is the ratio of duration of operation to duration of leaching. The duration of operation is determined by the rate of increase of pressure drop across the ESE column. This rate depends on the amount of copper collected and its morphology, which is in turn affected by copper concentration, waste chemistry, pH, flow rate, and current distribution. The leach duration depends on the amount of copper collected, its morphology, and on the rate of dissolution, Le., the corrosiveness of the leachate. For this process waste the operation duration was 8 to 16 h, depending upon the copper concentration in the waste during the period. Leaching the copper from the ESE column took about 35 min, using 20 wt percent HN03.

Recovery of copper

Recovery of copper by “converltional” flat plate electrolysis was made at good current efficiencies and reasonable current densities both in the labora

crylate) 10 X 4-7/8 X 6 in. (25.4 X 12.4 X 15.2 cm) container through whic the leachate was circulated. In a oneday recovery operation an overall coulombic efficiency of 80 percent was obtained at a current density of approximately 20 mA/cmZ with two 4 X 4 in. (10.2 X 10.2 cm) Pt/Ti anodes (Warston Excelsior, United Kingdom) and three parallelconnected 4 X 4 in.

A semiquantitative analysis by emission spectroscopy was made on the recovered copper with the results shown in Table 3. This table also lists a column for specifications for fire-refined tough pitch copper as defined by ASTM standard B216-72 [9]. This is a much more useful guideline than that for electrolytic tough pitch copper (only chemical requirement as per ASTM standard B5-74 is 99.90 percent Cu minimum 191 ). Recent publications [ lo ]

3erating con- he 213 power depends omplex in per t o the exing species

In is the ratio If operation ESE column.

,rphology, ;try, pH, flow le amount of L, i.e., the on duration waste during )ut 35 min,

was made at in the laboratcq to build up the >very unit for l ymethylmetha- through which overall ensity of ) Pt/Ti anodes :ted 4 X 4 in. as reduced from was good -one

ade on the also lists a ; defined by ?line than that t as per ASTM blications [ 101

t

51

Fig. 10. Slab copper recovered from plant process waste. Scale: inches.

have suggested far more stringent specifications for electrolytic tough pitch copper, e.g., Pb < 0.0015 percent. It can be seen that the purity of the re- covered copper is within the requirements of fire-refined touch pitch copper except for the antimony content.

determined the electrical efficiency of copper recovery from this leachate. Filtering to remove any solids present had no appreciable effect on plating efficiency, as shown in Figure 11. Plating efficiency in real leachate was gc erally lower than for corresponding copper concentrations in synthetic leachate (Cu-HNO,). This may be explained by copper-chelating materials, reducible organics, etc., initially present in the waste and then transferred to the leachate. The possible variability in the concentration of such materials may also account for the observed spread of efficiencies.

Laboratory work indicated that the absolute concentration of soluble copper

The data of Figure 11 also support earlier observations that, below a con-

!

52

TABLE 3

Comparison of analysis of recovered copper with ASTM standard for fire-refined tough pitch (FRTP) copper

Element Concentration, % FRTP Std., %

5% - 0.030 0.003 Mo - 0.030 Mg Si < 0.0015 AI < 0.0015

< 0.00015 Ag Ti < 0.00015 Zr < 0.0015 Fe < 0.0015 Pb < 0.0015 0.004 Sn < 0.0015 Ni < 0.0015 0.050 Bi < 0.0015 0.003 As - 0.015 0.012 Te - 0.015 0.025 (Se + Te)

~-

_ - - - _ - - 0.00015 _ - _ _ _ _ Considered same as Cu _ _ - - - - _ - -

_ _ _

100

I 90

- 8 0

2 70 5 g60 .u 50 z 540 - 3 0

sp

._

B

n

20

10

0 0 5. ooo 10, Mx) 15, OOO

Soluble Copper Concentration, mgl l it

Fig. 11. Coulombic efficiency of copper recovery.

centration of 10,000 mg/l soluble copper, coulombic efficiencies as low as 10 percent may be expected (see Figure 1). Electrically efficient operation of the recovery unit can be assured by keeping the leachate copper con cent ratio^ above 10,000 mg/l.

centrated solution. Alternatives to collection of the slab copper are sale or recycle of the con-

h

-

-

as on of tratio

con-

53

2oooX 5000 x Fig. 12. Scanning electron micrographs of 316 S/S cathodes after exposure to process waste. Reduction on reproduction X %.

Electrode materials

From the onset of work on the ESE concept, there was concern that anodes at $200 to $300/ft2 ($2150 to $3225/m2) (for material 0.025 to thick) could not be used economically on full-scale equipment. Furthermore, their long-term stability in chloride-containing wastes was questionable. An electrode materials testing program eventually led to selection of the “DSA” chlorine-evolving electrode (Electrode Corporation, Chardon, Ohio) as the preferred material. The following observations have been made on the use of these electrodes after 350 h of operation at 10 A/ft2 (approx. 10 mA/cm2):

No performance deterioration No loss of precious-metal content, as determined by the vendor No morphological changes, as determined by electron micrography of

replicated surfaces. The 31 6 stainless steel cathode showed some pitting, but not enough to

threaten the physical integrity of the electrodes. Figure 12 shows scanning electron micrographs of individual wire filaments in the cathode mesh.

LATEST DEVELOPMENTS

100 ga lh in ESE cell

During the past year a cell and system capable of handling 100 gal/min (378.5 l/min), with concentrations of up to 100 mg/l of contaminant metals, were designed and assembled. Tests of this ESE unit confirmed the removal of copper from a simulated aqueous waste at a commercial scale. A single ESE cell with a “working” volume of less than 1 ft3 (0.0283 m3) consistently removed from 75 to 90 percent of the copper from a lOO-gal/min (378.5 l/mii synthetic waste stream containing 5 to 100 mg/l copper. This degree of removal equaled or exceeded that predicted by data from smaller, laboratory- size, ESE cells. Higher degrees of removal could have been obtained either by reducing the flow rate or by using several cells in series. Larger flows could be handled by parallel units.

In extended operation cell performance did not noticeable decline. In fact, the performance, as measured by lower effluent concentrations, actually improved as accumulated copper increased the ESE cell’s active surface area during a given test. When enough copper had been electroplated to cause a substantial increase in pressure drop (approximately 30 psi, 0.2 MPa), the cell was leached with a dilute mixture of HISO, and HzOz to dissolve the copper and generate a highly concentrated copper sulfate solution. The leachii or regeneration, step of the cycle took less than 30 min compared with 10 to 30 h for the deposition portion of the cycle.

ACKNOWLEDGEMENTS

It is a pleasure to acknowledge the contributions of R.J. Galli and V.D.

55

Sutlic of Du Pont’s Engineering Development Laboratory for design and fabrication of the pilot plant and to J.L. Fitzjohn of Du Pont’s Engineering crechnology Laboratory for his work on mathematical modelling of the ESE cell.

REFERENCES

1 Anon., 1970. Electrochemical unit treats plating waste, American Machinist. 2 Bennion, D.N. and Newman, J., 1972. Electrochemical removal of copper ions from very

3 Chu, A.K.P., Fleischman, M. and Hills, G.J., 1974. Packed bed electrodes. I. The eltctro- & h i e solutions. Journal of Applied Electrochemistry, 2: 113.

chemical extraction of copper ions from dilute aqueous solutions. Journal of Applied Electrochemistry, 4: 323.

f Fleischman, M., Oldfield, J.W. and Tennakoon, L., 1971. Fluidized bed electrodes. Part IV. Electrodeposition of copper in a fluidized bed of copper-coated spheres. Journal of Applied Electrochemistry, 1 : 103.

chemical reactor designs in the treatment of dilute solutions. Electrochimica Acta, 19: 733.

6 Lopez-Cacicedo, C.L., 1975. Recovery of metals from rinse waters in ‘Chemelec’ electrolytic cells. Transactions of the Institute of Metal Finishers, 53 (2): 74.

7 Williams, J.M. (E.I. duPont de Nemours & Co.), 1975. Apparatus for Electrochemical Processing, U.S. Patent 3,859,195.

8 Bird, R.B., Stewart, W.E. and Lightfoot, E.N., 1966. Transport Phenomena, Wiley, New York, p. 646.

: ASTM 1974.1974 Annual Book of ASTM Standards, Part 6.

pitch copper. Journal of the Institution of Mining and Metallurgy, C75: 269.

5 Kuhn, A.T. and Houghton, R.W., 1974. A comparison of the performance of electro-

10 Mackay, K.E. and Armstrong-Smith, G., 1966. Quality control of electrolytic tough