Removal Of Heavy Metals From Water By Electrodialysis …infohouse.p2ric.org/ref/21/20271.pdf · REMOVAL OF HEAVY METALS FROM WATER BY ELECTRODIALYSIS IN THE PRESENCE OF ACIDS AND

REMOVAL OF HEAVY METALS FROM WATER BY ELECTRODIALYSIS IN THE PRESENCE OF ACIDS AND SALTS

David A. Rockstraw, John F. Scamehorn, Department of Chemical Engineering, University of Oklahoma, Norman, Oklahoma 73019

Abstract

The use of electrodialysis to remove cadmium from water has been shown to be feasible for a wide range of conditions. As much as three orders of magnitude reduction in the concentration of cadmium was observed for the feed with the concentrate stream from the process containing cadmium concentrations greater than 100,OOO ppm. The effect of cadmium concentration, pH, and added sodium chloride concentration on the overall current efficiency, apparent stack resistance, and the osmotic water transfer is discussed. Under many conditions, linear mixing rules can describe important parameters when complex mixtures of electrolytes are being electrodialy zed.

I . Introduction

The treatment of plating rinse waters containing dissolved metals is a major environmental concern of the metal plating industry. This paper discusses the removal of metals from water in the presence of acids and/or salts by electrodialysis.

Electrodialysis is a membrane-based separation process in which the partial separation of the components of an ionic solution is induced by an electric current. Electrodialysis has found extensive industrial usage for such applications as desalination of brackish waters', recovery of acids2, demineralization of corn sugar solutions, preparation of photographic emulsions and concentration of radioactive solutions3, and recovery of heavy metals from plating rinse waters4 and mining mill process waters5.

II. Basic Operating Principles of Electrodialvsis

A schematic diagram for an electrodialysis stack is shown in Figure 1. The stack consists of a series of alternating cation-exchange and anion-exchange membranes, each separated by a spacer through which the solutions are allowed to flow. When an electrical potential is applied across the stack, cations in solution begin to migrate toward the cathode, anions toward the anode. When a cation encounters a cation- exchange membrane, it will pass freely to the other side. As the cation continues to migrate, it will subsequently encounter an anion-exchange membrane that resists the tendency of the cation to travel further. When the similar phenomenon is considered for the migrating anions, the overall effect is alternating compartments between consecutive membranes becoming increasingly enriched, and increasingly depleted in electrolyte .

As shown in Figure 1, the stack configuration used in this work consisted of two cell pairs (a cell pair being the sequence cation-exchange membrane, spacer, anion- exchange membrane, spacer) as well as an isolating compartment. Use of the isolating compartment forces the ion common to all streams in the process to be an anion. The presence of a common cation in the electrode stream would result in the metal cations being reduced at the electrode surface, (plating the electrode) leading to a reduction in the performance of the electrodialysis unit.

A ANION-EXCHANGE MEMBRANE

C CATION-EXCHANGE MEMBRANE

CATHODE )-.

Figure 1 - Schematic of Experimental Membrane Stack

a

Figure 2 is a schematic of the experimental unit used in this work. Three independent piping networks transport the concentrate (the stream into which the solute(s) are being transferred), the diluate (the stream from which the solute(s) are being removed), and the electrode streams (a solution that serves to protect the electrode surface) from their respective holding tanks, to the membrane "stack", and then returns them to their originating reservoir. The entire separation occurs in the membrane stack.

III. Electrodialysis Performa nce

When the economic viability of electrodialysis is being investigated for design purposes, there are three parameters of major concern: the overall current efficiency, the apparent stack resistance, and the osmotic water transfer.

The overall current efficiency is defined as the net transfer of chemical equivalents from the diluate to the concentrate, divided by the net passage of electrical equivalents across the membrane stack in a specified time interval.

The overall current efficiency is calculated by a material balance on the stack and can be based on the conditions that exist in the diluate or the concentrate. Overall current efficiencies calculated for this work are based on the diluate side of the stack according to the expression

TI1 = { F y D } A ;

where qI is the overall current efficiency, F is Faraday's constant, V, is the volume of diluate processed in the time interval At, AC, is the change in concentration of the diluate, n is the number of cell pairs present in the stack, and I is the number of electrical equivalents transferred during At.

Overall current efficiencies of less than unity are due to a number of contributing factors; (i.) the membranes may not have a permselectivity of unity, (ii.) parallel currents may exist across the membrane stack manifold, (iii.) current flow across the membrane may be accompanied by a large transfer of water, and (iv.) at high current densities or low solute concentrations, hydrogen and hydroxide ions present in the aqueous media begin to participate in the current carrying process.

The apparent stack resistance is the sum of all electrical resistances present across the membrane stack. Contributions to the apparent stack resistance can be attributed to the membranes, the solutions in the stack, and the electrode reactions that occur. Individual components of the overall stack resistance are additive in the same way that the resistances of a series connection of resistors are additive in a DC circuit.

Solution contributions to the apparent stack resistance are dependent on the flowrates, the resistivities of the solutions and the spacings of the membranes. Resistivity is inversely related to concentration, therefore diluate contributions are usually much higher than those due to the concentrate streams.

Area specific apparent stack resistances were calculated by the expression

where A is the effective cell pair area. The transfer of water across an ion-exchange membrane (osmotic water transfer)

acts to limit the amount of diluate that can be produced and at the same time limits the concentration that can be obtained in the concentrate stream. Osmotic water transfer occurs due to electrical, temperature, concentration, and mechanical pressure gradients that exist across the membrane, as well as due to hydration of the ions being

3

Figure 2 - Schematic of Experimental Unit

W B A L L V A L V E W N t t D L f V A L V E

@ rntssunf WAGE

T t U P t R A T U R E a INDICATION

C.W. COOLING WATER

4

transported. Gering and Scamehord have shown the the amount of osmotic water transfer that occurs under normal operating conditions is almost entirely due to hydration of the solute and electroosmosis.

The forementioned design variables each have an impact on the costs associated with the operation of an electrodialysis unit. Capital costs are approximately proportional to the required effective cell pair area per unit flowrate (a);

where V, is the applied stack voltage. Operating costs are dominated by the costs associated with the required energy load per unit volume of feed processed (p);

From these expressions the relationship of the design variables to the costs of installing and operating an electrodialysis unit become apparent. Both a and p are inversely related to the overall current efficiency, while a is proportional to the apparent stack resistance. It is therefore desirable to operate at high current efficiency and low stack resistance. Since the applied voltage (V,) is inversely proportional to the effective cell pair area and directly proportional to the energy demand, the applied voltage can be used as an optimization variable between these costs.

Since osmotic water transfer has the effect of increasing the amount of concentrate produced and decreasing its concentration, the effect that water transference has on the costs of operation is to increase pumping costs if the concentrate is to be recycled or to increase disposal costs if the concentrate is a waste stream. If the diluate is being reused, high osmotic water transfer also corresponds to increased make-up water demand.

IV. Results

The results to be presented are for multicomponent systems containing HC1, NaCl and CdCl,. All references made to the "percent" or "fraction" of a species present in a particular solution should be considered to be on a basis of chemical equivalents. All concentrations are reported in units of normality. The notation xi should be interpreted as the "chemical equivalent fraction" of species i in the indicated solution;

chemical equivalents of species i total chemical equivalents of all species

xi =

When considering the figures that present the overall current efficiency and the apparent stack resistance, it should be kept in mind that the results were obtained from a batch process. The applied stack voltage was consistent between experimental runs at 4 volts per cell pair, and the flowrate of the solutions through the stack was 0.245 liters per minute.

Results are presented in a format to provide a direct comparison between the design variables for systems of one species and those of multiple species. Curves on the figures that lack data points are from the results obtained by Shah and Scamehorn2 for the x,=l and xNa=l curves, and from Gering and Scamehod for the xcd=l curves. These one-component curves are included when available at or near the indicated run conditions given for the multicomponent system being considered. This data is included to facilitate a comparison between the "pure" and ?nixed" systems.

5

The data of Gering and Scamehord is not for a completely pure system in that it was necessary in that study to have a small amount of acid present in order to lower the solubility of the metal to prevent the formation of insoluble metal salts on the membrane surface where concentrations tend to be high.

Figures 3 to 8 provide results for the overall current efficiency, Figure 9 for the apparent stack resistance, and Figure 10 is for the osmotic water transfer. Tables 1-3 present the conditions prevalent in the concentrate stream for each experimental run, while the equivalent fractions of species present in the diluate are provided directly on the figures near the data point for which it applies. Such data points for which this information is included are solid.

V. Discussion

From Figures 3 and 4, high current efficiencies are observed for pure cadmium chloride or sodium chloride removal. Current efficiencies are substantially lower for the removal of pure hydrochloric acid.

Figure 3 gives the results obtained for the saldmetal with only enough acid present to prevent fouling of the membranes caused by metal precipitates as previously mentioned. This run initially contained approximately an equal number of salt and metal equivalents in both the diluate and the concentrate. By the position of the curve for the salt and metal mixture midway between the curves representing the pure component systems, it can be concluded that the current efficiency of the mixed system is roughly a linear combination of the current efficiencies of the respective pure systems. It can also be concluded from the relatively unchanging nature of the equivalent fractions of both species observed during the run that the rate of change of heq&&tcmcemat ion of a specie in the diluate is dependent on the equivalent concentration of that specie present in the diluate and the concentrate. This also is of no surprise if one considers the laws of mass transfer that describe such a phenomena that includes both electrically induced migration and diffusion due to the presence of a concentration gradient7.

Figures 4 and 5 also present results obtained for the saldmetal system. The diluate in Figure 5 starts at over ninety percent salt concentrating into a mixture of half salt and half metal, while that in Figure 4 starts at greater than ninety percent metal concentrating into a stream of pure salt. The position of these curves relative to the curves of the pure component systems indicates that at high current efficiencies observed at relatively high diluate concentrations, the current efficiency is dominated by the composition of the diluate stream, while at low diluate concentrations and current efficiencies, the current efficiency is dictated by the relative amounts of each specie in the concentrate.

This effect can be explained by an understanding of the phenomena that act to limit the current efficiency. When the overall current efficiency is large, the concentration gradient across a membrane is relatively small8. Under these circumstances, the major cause of reduction in the current efficiency is due to the intrusion of the co-ion on the membrane from the diluate side, hence the composition of the diluate dictates the relative position of the current efficiency curve between the respective single component curves. When the concentration gradient across the membrane becomes large, small current efficiencies are observed. This is due to diffusion of the counter-ion being driven by the concentration gradient from the concentrate back into the diluate. As indicated by the results, the composition of the concentrate influences the observed current efficiency for the system in this region of the figure.

6

,

11 II

II

t-m

o

S8

8

00

0

\

II II

II

\ -\

xx

0

0

0

a0 u3

jc

(%) A3N313Md3 LN388In3 T

IW3A

O

7

0 0

.o

0

d I

0 0

0

0

d . 0

0 0

0

d

+ 0

0 0

0.

N

O

100

h

R u 80 * u x B

E w

i

OQ 60 w & & 3 u

uo 0

20

-

$--+ xCd = 0.905 XNa = 0.093 x, =om2

/, J J B xCd = 0.697 xNa = 0.302 XH =0.001

xCd = 0.395 xNa = 0.605 XH =o.OOo Y

I I

a , o o o i 0,0010 0 .0100

DILUATE CONCENTRATION (N)

FIGURE 4 - OVERALL CURRENT EFFICIENCY VS. DILUATE CONCENTRATION FOR MIXED SYSTEM OF ACID, SALT AND METAL.

8o i i I

60 i i I t i I

i

LU

/ xNa XH = = 0.929 0.011 7. N

/ /

/' /

0

FIGURE 5 - OVERALL CURRENT EFFICIENCY VS. DILUA'I% CONCENTRATION FOR MIXED SYSTEM OF ACID, SALT AND METAL.

I I I I I

100

r

80

60

110

20

C

0.963

FIGURE 6 - OVERALL CURRENT EFFICIENCY VS. DILUATE CONCENTRATION FOR MIXED SYSTEM OF ACID AND SALT.

xCd = 0.749

xCd = 0.714

0.0001 0.0010 0 .0100 0.1000

DILUATE CONCENTRATION (N)

FIGURE 7 - OVERALL CURRENT EFFICIENCY VS. DILUATE CONCENTRATION FOR MIXED SYSTEM OF ACID AND METAL.

I I

100

80 n K + u v

G

xcd = 0.3( xNa = 0.61 XH = 0.01

1 Y

xCd = 0.485 xNa = 0.491

xcd = 0.474 xNa = 0.461

/ / xH =0.065 \

XCd = 0.431 xNa = 0.536 XH = 0.033

/’

0.0001 0.0010 0.0100 0,1000

DILUATE CONCENTFUTION (N)

FIGURE 8 - OVERALL CURRENT EFFICIENCY VS. DILUATE CONCENTRATION FOR MIXED SYSTEMS OF ACID, SALT AND METAL.

10000

1000

too] 0.0001

0.0010 0.0100

0.1000

DILUATE CONCENTRATIO

N (N)

FIGURE 9 - A

PPAR

ENT STACK RESISTANCE V

S. DILUATE CONCENTRATIO

N - 13-

FOR VARIOUS SY

STEMS OF ACID, SALT AND M

ETAL.

0.8

0.7

0.0 0.5 1 .0 1.5 2.0 2.3

CONCENTRATE CONCENTRATION, AVE. (N) -

- 19- FIGURE 10 - OSMOTIC WATER TRANSFER VS. CONCENTRATE CONCENTFU'MON

FOR VARIOUS MIXED SYSTEMS OF ACID, SALT AND METAL.

Figure 3 4 5 6 7 8 8

Symbol [Concentrate] (N) 'Cd 'Na 'H

0.76 ,499 .4% .005 0 1.5 .036 .953 .011

2.5 .48 1 .527 .002 0 1.3 .Ooo .997 .003 0 1.1 ,992 .Ooo .008 0 2.5 .513 .477 .010 A 2.4 .516 .479 .003

1.4 .005 1 .983 .012 I .005 I 1.1 I .999 .Ooo .001 I .004

Symbol [Concentrate] (N) 'Cd %a xH I Xp(t=O)

0 0 0

* 1.1 .957 .042 .001 .002 1.1 .524 .476 .Ooo .001 1.1 S42 .457 .001 .002 2.5 .513 .482 .005 .106 2.4 .516 .479 .003 .030

Symbol 0

15

'Cd %a XH .998 .Ooo -002

I A 0

.542 .457 .oo 1

.524 .476 .Ooo

.233 .763 .004 0 *

.958 .042 .Ooo

.499 .496 .005

Figure 6 for the acidsalt system indicates that the overall current efficiency for removal of the salt is greatly reduced when ten percent of the diluate is acid. This is the expected result when the overall current efficiency curves for the pure salt (xNa=l) and the pure acid (xH'1) systems are considered. As shown in Figure 7, the same effect is observed for the acidmetal system as was observed for the acidsalt combination.

A direct comparison can be observed between the acidsalt and the acidmetal systems at the diluate concentration of 0.01 N where each system contains approximately 90 percent by chemical equivalents of salt and metal respectively. The overall current efficiency in the acidsalt system has been reduced by approximately 25 percent of the difference in the current efficiencies of the respective pure component systems, whereas that in the acidmetal system has been reduced by only about 15 percent of the difference between the respective pure metal and pure acid systems.

Also in the acidmetal system at a diluate concentration of just over 0.02 N and xH P 0.3, the overall current efficiency has been reduced from the pure metal system by an amount of almost 55 percent of the difference between the pure metal and pure acid systems. This indicates that the current efficiency of a system containing acid is very sensitive to the amount of acid present.

Figure 8 presents results obtained for the tertiary system composed of acid, salt and metal. In both cases presented in this figure, the concentrate was roughly equal fractions (equivalent basis) of salt and metal. The trends observed for the acidsalt and acidmetal systems are once again observed for the three component system. In each of the two runs presented in this figure the amount of metal in the diluate was about fifty percent. When the amount of acid relative to salt was increased, the overall current efficieq-was observed to decrease.

Figure 9 shows the results obtained for the apparent stack resistance for a number of runs. All non-solution contributions to the apparent stack were kept consistent between experimental runs (including the contributions due to the electrode stream concentration) and concentrate concentrations seldom varied by more than several percent during a run, allowing changes in the apparent stack resistances to be directly attributed to changes in the diluate concentration. Figure 9 suggests that the apparent stack resistance is virtually unaffected by the composition of the diluate and concentrate streams but follows a strong correlation with the total concentrations of both streams when expressed on a chemical equivalent basis. Apparent stack resistance is seen to increase for decreases in either the diluate or concentrate concentration.

The effect of acid content on the apparent stack resistance is seen in Figure 9 at the higher diluate concentrations present in the initial stages of the 2.5 N experimental runs where the acid had not yet been purged from the diluate. The extremely small size of the hydrogen ion allows it much easier passage through the membrane than the much larger salt and metal species. This greater mobility of the hydrogen ion results in a much lower electrical resistance as observed in the figure.

Water transference results are depicted in Figure 10. The experimental technique employed was designed for measurement of the overall current efficiency and the apparent stack resistance. Water transference investigations should use a single experimental run, from which the amount of water transferred can easily be determined. The nature of the runs performed involved a large amount of purge and sample withdrawn, making solution loss inevitable. However, if care is used in minimizing this loss, osmotic water transfer data can be estimated. In this calculation, solution loss would result in an underestimated osmotic water transfer. The interpretation of this data follows analogously to that provided when discussing the

16

overall current efficiency results. For the run that initially contained an equivalent amount of salt and metal in the concentrate with a negligible amount of acid, the amount of water transfer observed is almost a linear combination of the amount observed for the systems representing only the pure species.

The curve indicated as 1/C on Figure 10 is the contour that defines the point at which the solution passing through the membrane is of the same concentration as the concentrate stream. The point at which the xi curves intersect the 1/C contour represents the maximum attainable concentrate concentration for species i. These concentrations are 2.1 N, 2.3 N, and 1.8 N for HCl, NaCl and CdC1, respectively.

VI. Conclusions

The results of this work show that the design variables for a multicomponent system comprised of an acid, a salt and a heavy metal can be interpreted and therefore approximated from the design parameters obtained for the respective pure systems at the same operating conditions of stack voltage, flowrate and the total (all species) normal concentration of the diluate and the concentrate. Although further experimental work will be necessary to quantify much of the observed effects, this preliminary study indicates that;

3 J

4.)

Metals can be efficiently removed from aqueous solutions of moderate pH by electrodialysis. As the pH of the solution is decreased, the removal of the metal species becomes less efficient. The current efficiency for the salvmetal system can be described in terms of linear combinations of the current efficiencies for the associated single specie systems. The effect of acid on the current efficiency is highly nonlinear, with relatively small amounts of acid causing large reductions on the current efficiency.

Apparent stack resistances are independent of the nature of the solute(s) and tend to follow only the total normal concentrations of both the diluate and concentrate streams when expressed in Normality. Acid causes reductions in the apparent stack resistance due to its low mobility and thus its ability to carry charge across the stack much more quickly than the salt or the metal.

The osmotic water transfer obtained for a mixed solute system may also be predictable for combinations of salt and metal as linear combinations of the respective pure component systems if the acid content is relatively low.

VII. Acknowledneme nts

Financial support for this work was provided by the Bureau of Mines Grant No. G1125132-4001 and the Oklahoma Mining and Minerals Resources Research Institute.

17

VIII. References

(1.) W. A. McRae, Desalination Technology, Developments and Practices, Andrew Porteous, ed., Applied Science, Chapter 8 (1983).

(2.) P. M. Shah and J. F. Scamehom, Ind. Eng. Chem. Res., 26,269 (1987).

(3.) J. B. Farrell and R. N. Smith, Industrial and Engineering Chemistry, 54, 29 (1962).

(4.) M. J. Sweeny, Summer National AICHE Meeting, Paper No. l a , (1985).

(5.) T. Bhagat, Water and Pollution Control, 118, 11 (1980).

(6.) K. L. Gering and J. F. Scamehom, Sep. Sci. Technol., 23, 1231 (1988).

(7.) A. J. Baird and L. R. Faulkner, Electrochemical Methods, Fundamentals and Applications, Wiley & Sons, Inc., New York (1980).

(8.) L. H. Shaffer and M. S. Mink, Principles of Desalination, 2"d ed., K. S. Spiegler and A. D. K. Laird, editors, Academic Press, New York, Chapter 6 (1980).

18