I
ERIC GOLDBLATT Stilfontein Gold Mining Co. Ltd., Transvaal, Union of South Africa
Recovery of Cyanide from Waste Cyanide Solutions by Ion Exchange
THE
recovery of cyanide from gold barren cyanide solution, particularly the waste solution from gold reduction plants on the Witwatersrand, has long been desired, and numerous investigations using the conventional chemical methods have been carried out. The main difficulty in recovery in the most cases has been the low concentration in which the total available cyanide exists in the solution. At Stilfontein this is in the order of 0.0370 in terms of potassium cyanide, KCN. Added to this is the large number of metal cyanide complexes present, of which some are not amenable to treatment. The most common method for cyanide regeneration involves the acidifying of the waste cyanide solution with volatilization of the hydrogen cyanide in the Crowe-MeNill process. This process is i n use in various parts of the world where the recovery of gold by cyanidation is practiced ( 2 ) . The application of this method is not economical with solutions below 0.1 potassium cyanide (72). I n view of the low concentration of cyanide, ion exchange presented a possible method for the recovery of cyanide in the waste solution from this company's reduction works. Ion exchange had already been used for the concentration of gold and silver complexes present in gold pregnant cyanide solution ( 4 ) . Burstall and others
( 7 ) have demonstrated the ability of a strong base resin, Amberlite IRA-400, to adsorb the cyanide complexes usually found in Witwatersrand gold-bearing cyanide solution. However, no mention is made of the adsorption characteristics of the cyanide ion alone. Cyanide in the waste solution of other industries has been concentrated successfully from acidified solutions on strong base resins operating in the hydroxide cycle (9). Walker and Zabban (73) have carried out considerable experimentation with both cationic and anionic exchange, for the treatment of plating room waste cyanide solutions. Mention is also made by Kunin and Myers (8), where a hydrochloric-hydrocyanic acid mixture is passed through a column of strong baseresin exchange, where both acids are adsorbed with subsequent elution of pure hydrocyanic acid in the effluent. With a weak base resin and the same acid mixture hydrochloric acid only is adsorbed.
Hydrocyanic acid alone in solution has been found to be adsorbed on IRA.460' with no leakage (7). Weak base resins have a low capacity for the cyanide complexes (73). However, this article shows that the concentration of the complexes on the ion exchange governs the amoutlt, of free cyanide which can be recoveredc Therefore, full adsorption of the complexes must take place, as their leakage into the effluent will mean a loss of cyanide, which will have a deleterious effect on gold precipitation should the effluent be re-used in the gold plant. If a clean effluent can be obtained for re-use, it will be, with water such a precious commodity in South Africa, as important a n accomplishment as the recovery of cyanide. This investigation, therefore, was limited to studies with strong base-resin exchange for the adsorption of simple and complex anions from waste cyanide solutions.
I The staff-industry collaborative report preceding described recovery of gold by treatment of gold slime with cyanide solution, at the Daggafontein mine of Anglo American Corporation of South Africa. At Daggafontein the cyanide solution is recirculated after gold has been precipitated. This article, from another South African gold mining company, Stilfontein, tackles the difficult problem of recovering cyanide from the solution from which gold has been precipitated. By use of a strong-base ion exchange resin, Stilfontein has obtained a recovery promising enough to warrant additional work on a pilot plant scale.
REGENEW
1
IR- IM
TO GMI, PLANT MILLING ORE IN CYANIDE SOLUTION
I
CN:
CNS
r!3
'ri COLUMN
COLUMN
I
COMPLEXES
/
WITH ELUTION OF CNS-
WLUMN
I
F4oM PREVIOUS CYCLE
METAL CATIONS HCN
He%
FULL LINES 5" SCLLTONFLOR
Figure 1. Adsorption cycle, adsorption of anions from waste cyanide solution on IRA-400 in two columns in series Elution and second adsorption cycles
FULL UNES SHOW SOLUTION F W W
Figure 2. IRA-400
Elution of adsorbed anions from the first column
VOL. 48, NO. 12
DECEMBER 1956
2107
Table 1.
Typical Cyanide Gold Barren Solution" A
Influent Sample ,Yo. B C
D
Grams per liter
CNCNS-
codNickel Copper Iron Cobalt
complex complex complex complex
Qn(CN)a-2 as Ni(CN)4-3 as C U ( C N ) ~ - ~ as Fe(CN)6-4 as C O ( C N ) E - ~ Au(CN)z-
0.0460 0.0550 0.102 0.001 0.0473 0.0220 0.0300
Trace Trace 0.00009
0.0400 0.0704 0.108 0.001 0.0461 0.0195 0.0100 Trace Trace 0.00002
0.0350 0.0502 0.120 0.001 0.0541
0.0455 0.0560 0.102 0.001
0.0150 0.0005 Trace 0.00001
0.0475 0.0180 0.0150 Trace Trace 0.00001
11.15
11.05
0.0201
PH a
11.00 11.20 From Stilfontein Gold Mining C0.k gold reduction works.
Scope
of the Investigation
I t was considered of prime importance whether the waste cyanide solution should be acid or alkaline for passage through a n ion exchange resin bed. Because gold barren waste cyanide solution from the reduction works is alkaline (pH l l . O ) , it was preferable that the process of ion exchange take place with the solution in this condition. To neutralize this solution would require a large quantity of acid, the expense of which would render the process too costly to operate. Therefore, the ion exchange using the original alkaline solution was undertaken for initial study. In addition to the above. it was necessary to give consideration to availability. cost, and suitability of possible eluting reagents. Therefore, this study of cyanide recovery by ion exchange was limited to an investigation with the most economical reagents produced locally-e.g., sulfuric acid, sodium chloride, and sodium carbonate. A cursory search of literature shows a legion of both the cyanide and thiocyah
CN-C N s-I .5 I .4
-
nate complexes, many of which must exist in the solutions from this plant. A number of these have been identified already, and the main anions present in Stilfontein waste cyanide solution are given in Table I. Experimental
For the sorption cycle on this investigation the conventional column procedure was used, with two or more identical glass columns in series containing a n anion exchanger, Figures 1, 2, and 10. Each column was 150 cm. in length, and 4.0 cm. in diameter. Commercial Amberlite IRA-400 resin, with no prior treatment, was loaded to a depth of 2 feet giving a volume of 720 ml. of settled resin. Before use, this resin was regenerated with a 5% solution of sulfuric acid. and rinsed with water until no trace of acid remained. Each day fresh cyanide waste solution was obtained directly from the gold reduction plant, loaded into 40-gallon drums, from where it was fed by gravity to the columns operating
by the usual downflow method. The flow rate for the sorption cycle was maintained at 0.27 bed volume per minute (2.0 U. S.gallons per cubic feet of wet settled resin per minute) and measured with an accurately calibrated rotameter. Regeneration of the loaded columns was carried out also by down flow, and, except when otherwise stated, was maintained at a rate of 0.067 bed volume per minute (0.5 U. S. gallon per cubic feet of wet settled resin per minute). For the regeneration of Amberlite IRA-400 acid elutriant was chosen. Since this brought about decomposition of the adsorbed complexes, it was necessary to adsorb these released metal cations on a further column containing a cation exchange; prior to returning the eluent to the gold plant circuit. A cation resin bed of the same dimensions was therefore incorporated in series with the IRA-400 column while eluting. In this case, cation resin bed adsorption and elution flow rates were the same (0.067 bed volume per minute). The only mechanical difficulty experienced during the experiments was the formation of a magnesium sulfate floc, and, on occasions, a precipitate of hydrated zinc oxide formed in the cyanide solution ( 3 ) which resulted in fouling of the top of the resin bed. These precipitates are unlikely to occur on a plant scale if the cyanide solution is taken directly after gold precipitation. However, downflow rinsing with two bed volumes of tap water, and backwashing with three bed volumes at a flow rate of 0.067 bed volume per minute was sufficiently effective in removing the floc that had accumulated during the adsorption cycle. The anions encountered in this investigation are discussed under separate headings-in the sorption cycle. and the regeneration cycle.
- -_ _
FLOW R A T E 0 . 2 7 B E D V O L U M E / M I N U T E ( 3 0 B E D V O L U M E S OF 0 2 N H2S04 USED FOR E A C H E L U T I O N I
,
1.3
9
t 1.2 K
c 1.1 z 1.0
- 09
: 08 z
;a7 k06 w
;O
5
V
30 4 a3 02 0.1 0
100
20 0
300
40 0
500
600
BED VOLUMES
Figure 3. Adsorption curve of cyanide and thiocyanate ions on Amberlite IRA-400 for first six consecutive cycles (first column only)
2 108
INDUSTRIAL AND ENGINEERING CHEMISTRY
Cyanide Anion, Other workers on cyanide ion exchange with IRA-400 resin report partial or complete adsorption of the cyanide radical. A negligible amount of adsorption was apparent with Stilfontein cyanide solution using fresh resin. As the adsorption and acid elution cycles proceeded, the cyanide ion was adsorbed in increasing amounts, Figure 3. This increase was due to the precipitation in the resin bead of cuprous cyanide, which was formed by decomposition of the copper complexes that remainea on the resin bed in increasing quantities after each acid elution. I t appeared that, on a new adsorption cycle, the free cyanide ion combined with the cuprous cyanide to form the complex with the immediate adsorption of the latter on the resin. When cyanide ion breakthrough did occur, it indicated that all the cuprous cyanide had been complexed, and that the cyanide ions were passing directly through the resin column with no adsorption. The second column in series received the first column’s effluent free from nearly all the complexes and leakage of the cyanide radical was readily apparent. It appears then that, in competition with the other simple anions present in the first column’s effluent, other than the hydroxyl ion, the cyanide ion has the weakest affinity for IRA-400. The concentration curves after six sorption cycles (first column stabilized) for the cyanide ion in the first and second columns are illustrated by Figures 4 and 5, respectively. During the adsorption cycle of the first column, the effluentinfluent ratio (C/CiJ on reaching 1.0 remained so, indicating that the cyanide ion had been completely adsorbed as the complex. Otherwise a curve similar to the thiocyanate with a high effluentinfluent ratio would have been evident where the thiocyanate was initially adsorbed, and under the prevailing conditions eluted off by the complexes in the effluent. With the second column where cuprous cyanide was absent, a little cyanide ion adsorption had taken place (Figure 5 ) only to be subsequently eluted with the thiocyanate and carbonate ions in the influent (effluent of first column). Thiocyanate Anion. With the passage of the waste cyanide solution through the first column, the adsorption of the thiocyanate ion on the IRA-400 resin bed was complete. However, its affinity for the resin exchange sites was not so great as the complexes, and during the adsorption cycle the incoming complexes readily displace the thiocyanate ions that were adsorbed on the resin bed during the earlier part of the cycle. This is shown in Figure 4 for the first column where the thiocyanate effluent influent ratio increases above 1.0. Figure 3 also illustrates the preferential adsorption of the complexes to that of the thiocyanate.
200
300
500
400
700
600
BED
800
1000
900
1100
VOLUMES
Figure 4. Typical sorption curve for first column of Amberlite IRA-400 resin bed containing cuprous cyanide (after stabilization)
In this case where the resin column contains cuprous cyanide, which increases after each cycle (up to the sixth adsorption cycle) and the resulting cupric cyanide from the combination of the cuprous cyanide and free cyanide takes u p further exchange sites, the thiocyanate breakthrough is advanced, indicating a reduction in the available exchange sites for the thiocyanate ions. Calculation also showed that the amount by which the thiocyanate effluent-influent ratio exceeds 1.O, was compatable with the number of exchange sites taken over by the complexes. Regarding the second column with the absence of complexes in the effluent from the first column, almost complete adsorption of the thiocyanate ion was attained. The trace quantities that appeared in the
latter part of the final effluent was due to the high rate of flow-through of the solution (Figure 5). Fortunately, a small amount of thiocyanate does not have any harmful effects on gold precipitation should it be returned to the gold plant (73). Nevertheless, it is not desirable to have thiocyanate building up in the plant circuit if it possibly can be avoided. Carbonate Anion. The carbonate ion was adsorbed on IRA-400 in preference to the cyanide ion, but not that of the thiocyanate. However, preferential carbonate ion adsorption did not take place when cyanide ions combined with cuprous cyanide already existing on the resin bed. Leakage of carbonate to the gold plant with the final effluent is of no conseauence as it has no detrimental effect on gold plant operation.
1.21 1.1
-
-
1.0
OB 0.70.605 0.40.3-
0.9
0
0
Co=INFLUENT (FIRST COLUMN’S EFFLUENT) C=EFFLUENT FLOW RATE 027 BED VOLUME / MINUTE OCNOCNS-
02 -
ai -
I
4
200
300
c!
400
500
2
! =
600
I
-
-
700
I -
800
900
1000
BED VOLUMES(THR0UGH BOTH COLUMNS) Figure 5. Absorption curve for second column (in series with first; of fresh IRA400 resin) VOL. 48, NO. 12
DECEMBER 1956
2109
Thiosulfate Anion. Thiosulfate ions varied in concentration from trace to 0.001 gram per liter in the cyanide waste solution. Most of this anion was adsorbed on the first column of IRA-400. Nevertheless, there was a consistent trace leakage through both columns. Although thiosulfate affects the precipitation efficiency of the gold by the zinc, this trace in the final effluent to the gold plant can be ignored. Auric Cyanide. I n the first three to four sorption cycles the auric cyanide was completely adsorbed on the first column in preference to all other anions except one type of cobalt cyanide. This, however, was not maintained after the fourth cycle when leakage became evident. This behavior of auric cyanide was also repeated in the case of the second column, but in this instance it was anticipated that the auric cyanide should build up on the resin bed with successive cycles; the application of a suitable preferential elutriant for auric cycanide would then have been used to recover the gold as a concentrate. However, analysis of portions of the resin bed after each cycle showed that the complex reached a maximum of only 0.15 meq. per ml, of resin. In this case, the auric cyanide concentration in solution was extremely low and thiocyanate exhibits elution characteristics with the complex (7) ; this may have been responsible for the leakage. Cobalt Cyanide Complex. Various types of cobalt cyanide and thiocyanate complexes are listed in literature. I n this investigation there was evidence of more than one cobalt complex present in the cyanide solution. With only a trace existing in the influent it was difficult, without further considerable study, to determine what type was actually present. Whenever one of these complexes is adsorbed on the first column of IRA400 resin it became extremely difficult to remove, more so than suggested by other investigators. Like auric cyanide the cobalt complex builds up fairly rapidly in the first three to four adsorption cycles to approximately 0.03 meq.as Co(CN)G+-per milliliter of resin. Thereafter this rate of increase was dependent to a certain extent on the nature of the cyanide solution itself. The complex responsible for the poisoning of the resin has been suggested to be the hydrated form possibly as the C O ( C N ) ~ ( H ~ O(6, ) - ~77). Nickel Cyanide Complex. Nickel cyanide, or any other nickel complex or salt present in the gold pregnant cyanide solution prior to gold precipitation, has a deleterious effect since the nickel is preferentially precipitated out on the zinclead couple in place of the gold. Therefore, any leakage of nickel or nickel compound must be kept a t a minimum, should the effluent from the ion exchange unit be returned to the gold plant. Un2 17 0
fortunately, this was only partially achieved, and a resultant leakage of approximately 257, of the nickel became apparent in the first column’s effluent. Figure 4. This was also the case with the second column of resin in series. Work by Hancock and Thomas (5) indicates that the nickel complex is partially dissociated by the high concentration of calcium and hydroxyl ions that exists in the waste cyanide solution, with the subsequent release of the nickel as a cation, which is unaffected by the anionic exchanger. -4lternatively, adsorption of one type of nickel complex was complete and the leakage was due to other cyanide or thiocyanate complexes which have less affinity for the resin exchange sites in the presence of the many other anions in the solution. This was viewed empirically and the fact that more nickel was adsorbed than escaped into the return effluent was sufficient to indicate that no serious increase of nickel concentration would occur in the gold plant. Iron Cyanide Complexes. As there was only a minute trace of total iron found in the cyanide solution, it was assumed to be in the form of ferrous cyanide as the ferric cyanide is not generally found in gold cyanide solution. There were still trace quantities of iron after the cyanide solution had passed through both columns. Since the initial concentration was so low, little could be done to determine what reaction, if any, could have taken place.
Zinc Cyanide and Copper Complexes. Of all the complexes in the cyanide influent, zinc and copper cyanide complexes were the only ones to be adsorbed completely on IRA-400 in the first column with no leakage. As the normal copper complex is trivalent, and has potentially higher exchange ability, a higher concentration of this complex would be expected a t the surface of the resin bed. Moreover, if sorption is carried out with greater concentrations of copper in the influent, the divalent zinc cyanide would tend to be eluted. However, analysis of portions of a fresh resin bed, newly exhausted, shows that the zinc-copper complex concentration ratio throughout the bed was the same as in the influent. With a resin bed containing cuprous cyanide, the quantity of cyanide ions adsorbed indicated the formation of the Cu(CT\;)4-3 anion. This conversion to the complex was visible on sorption. The resin bed after elution was left a dark color, and, as sorption proceeded, there was a lighter demarcation of the boundary where the cuprous cyanide complexed. While the resin column was on sorption, the volume of the bed decreased by approximately 15%, the number of anions present obscuring the cause of this phenomenon. Runs were, there-
INDUSTRIAL AND ENGINEERING CHEMISTRY
fore, carried out on fresh resin columns with influents containing one of the simple or complex anions only. Results showed that the adsorption of thiocyanateand/or copper complex (or formation of the copper complex) on the resin bed was responsible for the major decrease in volume (Figure 2). The anions remaining on the first column resin bed after adsorption and suitable washing, and providing the bed. has been through at least six cycles, are listed as follows in order of decreasing concentration: C U ( C N ) ~ - Zn(CN)d ~, -z, (these first two were in approximately equal concentrations) ; Ni(CK)3-2, S203-2, Au(CN)*-, C O ( C N ) ~ - ~and , Fe(CN)6-4, with the first three complexes being the major source of recovered cyanide. As this process was, initiated to regenerate cyanide only, n o attempt was made to elute the above anions selectively. Before elution by acid was attempted, neutral and alkaline solutions of 10% sodium chloride, 10% sodium carbonate, and 5y0 caustic soda were tried as, elutriants, but proved to be unsuccessful economically. An economical elutriant for cyanide recovery can be one where the majority of the anions are removed by passage through the resin bed of, at the most, six bed volumes of 57, sodium hydroxide or 10% sodium carbonate. Besides being inefficient regenerants these salts were selective to individual anions, only. However. a 0.2;V hydrochloric acid proved very efficient for cyanide recovery, but for economical reasons, sulfuric acid would be preferable as a n acid elutriant. A systematic investigation of the efficiency of varying strengths of sulfuric acid as an elutriant was then carried out. Miith any acid strength the elution of the complexes, resulted in the release of hydrocyanic acid with precipitation of the metals a s simple cyanides-e.g., cupric cyanide, zinc cyanide, and nickel cyanide. These in turn dissolved with further decomposition as the acid passed through the bed. One aspect which governed the strength of the acid elutriant was the rate by which the cation resin column, which was in series with the strong base-resin column, while on elution, adsorbed the metals now existing as cations in the first column’s eluate. The most suitable acid strength for elution, yet not containing too high a metal concentration for the given cation resin bed dimensions and flow rate (0.067 bed volume per minute), was sulfuric acid of 0.2N strength. Zinc and Nickel Cyanide. Zinc cyanide with the nickel assumed to be nickel cyanide was the main source of recovered cyanide. Elution with 0.2N sulfuric acid regenerant results in the decomposition of these complexes. According to the following equations, the resultant hydrocyanic acid and respec-
100.0
a 75 0 0 W
m
zcn W
a f
500 iL Q W
0
3
d Zn
o\"
o Ni 0 Cu FROM Cu (CN),-3
t-
-I W
CN-
ABSORBED DURING THE 6'hSORP CYCLE
25 0
Q
s,o,-2
FLOW RATE 0.067 BED VOLUME 1 MINUTE
I 10,o
5.0
0
15.0
I 25.0
20.0
1 30.0
BED V O L U M E S
Figure 6.
Elution
of first column with 0.2N sulfuric acid (after sixth sorption cycle)
tive metal cations appear in the eluate, Figure 6. R:M(CN)d HzS04 -+ M ( C N ) z 2HCN R:SO4 (1) M(CN)2 HzS04 MS04 2HCN
+
+
+
+
-+
+
(2)
where ( R = resin, and M = metal cation). Both complexes decomposed a t approximately the same rate as given by reaction (Equation 1). Thereafier the rate for decomposition by Equation 2 differs for the two cyanides with the nickel cation only appearing in strength when more than half the amount of zinc had been eluted from the resin bed, due to nickel cyanide being less soluble in sulfuric acid than zinc cyanide. Therefore, it was necessary to continue elution with acid for as long as conveniently possible, but not less than 30 bed volumes, which removed between 70 to 80% of the nickel cyanide. Of the 30 bed volumes used, about 18 of the last portion of the eluate was available for re use as a n elutriant. The remaining nickel salt was effectively removed by the water rinse and backwashing. Copper Cyanide Complexes. Elution with 0 . 2 N sulfuric acid of the copper complexes from IRA-400 resin was similar to that of zinc and nickel cyanides, where there was first a decomposition of the cupric cyanide to the cuprous cyanide. Thereafter the rate of ionization of the cuprous cyanide was very slow. Practically, results indicated that the rate of elution was dependent on the
volume of acid passed through the bed, but independent of acid concentration in the range of 0.2 to 1.ON sulfuric acid. The cuprous cyanide is only slightly soluble in sulfuric acid, which accounts for the noneffectiveness of higher acid concentration. I t can therefore be assumed that the cuprous cyanide precipitated within the resin bed would be extremely difficult to remove, but that on or near the bed surface far more easily depending on the quantity of elutriant flowing past. With the passage of 30 bed volumes of acid eluate for each elution, equilibrium conditions were established where the amount of cuprous cyanide remaining on the resin bed did not exceed 1,370 of wet-settled resin. Washing and backwashing, even when excessive, did not lower this figure. During the investigation, occasionally insufficient elutriant was used, so that cuprous cyanide began to foul the resin bed. This was remedied by passing 2N potassium cyanide through the resin bed, and six bed volumes eluted approximately 50% of the cuprous cyanide from the column. Cobalt Cyanide Complex. Some of the cobalt was eluted with the 0.2N sulfuric acid and reached a peak value of 0.001 gram per liter at approximately seven bed volumes. At this low concentration it was difficult to determine how much was complex or cation. During successive cycles a cobalt cyanide begins to build u p on the resin and might eventually cause serious
fouling. Elution with 2 N potassium thiocyanate as suggested by Burstall and colleagues proved unsuccessful. Several tests were carried out with the use of up to 30 bed volumes of this elutriant, but only a maximum of 10% of adsorbed cobalt complex was removed per cycle. Experiments with other eluting reagents such as 10% sodium chloride, sodium carbonate, and sodium hydroxide proved
60 0
50 0
400 m
Y
z Y v)
E
g ~
300
Y
?,
" 8
20 0
10 0
0
I 50
I I00
I 150
BED V O L U M E S
Figure 7. Elution of second column loaded with thiocyanate VOL. 48, NO. 12
DECEMBER 1956
21 1 1
90 0
e -
1070 sodium carbonate-at a flow rate of 0.067 bed volume per minute. A lower elution flow rate with 10% sodium chloride did not show any further improvement. Therefore. regenerating of a strong base-resin bed would be almost equivalent in cost to the free cyanide recovered. An alternative, which was adoptrd for general running of two columns in series, was to allow the thiocyanate-loaded second column to take the place of the first column for the following adsorption cycle, Figure 2. The metal cyanide complexes in the influent were then used to elute the thiocyanate from the resin. This unfortunately meant a loss of the free cyanide during elution of the first column. As soon as the elution was accomplished, the effluent from the second column (now on adsorption) was diverted to the first with the subsequent adsorption of the thiocyanate, again allowing a cyanide solution free of del?terious substances to return to the gold plant. As some cuprous cyanide remained behind on the resin of the first column after acid elution, ionic cyanide, on passage through the resin bed was complexed and adsorbed, the capacity of the resin for thiocyanate subsequently lowered. However, the column a t this lower capacity was able to adsorb the thiocyanate in the effluent from the column adsorbing the complexes (Figure 5), and a column of fresh resin did not attain full thiocyanate saturation. With this procedure of alternating the columns for adsorbing the complexes, the total recovery of available cyanide was of the order of 85%.
adsorbed on the IRA-400 resin. However, as the adsorption equilibrium G f the complex was attained at a low concentration on the resin, and did not seriously interfere with the capacity for the major complexes, its removal was therefore unnecessary by the use of other reagents. Thiosulfate Anion. The first six bed volumes of 0.2N sulfuric acid eluted almost all the thiosulfate (Figure 6). At this period of the elution cycle the effluent was only acid. but: as more elutriant was passed through and the whole was collected in a single container. the acidity of the bulk eluate increased so that the thiosulfate was eventually decomposed to the sulfite. The latter if returned to the gold plant has no deleterious effect on gold precipitation.
100 0
80 0
700
LL 0 I
600 W 2
w
_I
ap 500
40 0
30 0
20 0
Elution of Adsorbed Anions from Second I R A - 4 0 0 Resin Column
10 0
0
20
I C
30
4 0
50
60
After the first sorption cycle with a fresh resin bed, the main anions adsorbed on the second column were the thiocyanate, with a little nickel cyanide, auric cyanide, and a trace of cobalt cyanide. Hereafter, conditions differed as copper complexes were introduced in subsequent cycles. At the present time, recovered thiocyanate is a waste product, and any regeneration must be extremely efficient and inexpensive. Unfortunately, this proved otherwise. Elution of thiocyanate required a considerable volume of elutriant at a high concentration in comparison with a 0.2;L' sulfuric acid solution. The latter proved totally inadequate for thiocyanate elution and would require hundreds of bed volumes to regenerate a fully loaded IRA-400 resin bed of thiocyanate. Figure 7 illustrates the low elution efficiency of thiocyanate b>- the locally available inexpensive reagents-1 0% sulfuric acid, 10% sodium chloride, and
70
B E D VOLUMES
Figure 8. Elution of IR-120 cation resin column with 10% sodium chloride
even less effective. Eventually it was found that only by removing the contaminated resin and treating it batchwise with 21%' potassium thiocyanate at 70' C. was there any real success. Fifty per cent of the complex was removed after 1 hour a t this temperature. Encouraging results have also been obtained by allowing the first column poisoned with cobalt cyanide to replace the second column during the following sorption cycle, since this column will receive only the thiocyanate and cyanide ions. The elution of the complex by this method was rather slow, but indicates a method where the process can be adapted to clean up its own fouled resin. Auric Cyanide. Sulfuric acid elutriant, 0.ZA%',had no effect on auric cyanide
Adsorption of Cations from Eluate of First Column (0.2N HzS0.J on IR-120 Resin Before passing eluate of the first column back to the gold plant, it is essen-
100 0
100 0
7 75 0
75 0 w 0
m
z -
0
rn w
v)
e w
z w
a:
I
p
50.0
0
0
w
iL a
/
c 3 c
$
500
u.
I
250
c 3
OCN-
i w
CCNS-
0 2
'
F L O W RATE 0 0 6 7 BED V O L U M E / M I N U T E
250
I O
2 0
30
49 50 60 BED VOLUMES
70
80
90
100
110
Figure 9. Elution of cuprous cyanide conditioned IRA-400 resin column with 0.2N sulfuric acid
21 12
INDUSTRIAL AND ENGINEERING CHEMISTRY
c2s04
F L O W RATE 0 0 6 7 BED VOLUME / MlhUTE
I 0
0 2 N H2S04103%HCN
e 02N
0
50
I 10 0
I 150
1 20 0
BCD VOLJMCS
Figure 10. Elution with first column with sulfuric acidhydrocyanic acid elutriant
tial that the metal cations be removed. This has been accomplished by use of the strongly acidic IR-120 resin used in the sodium form. Other resins were tried, such as Permutit Z.K. 225, which adsorbed the zinc, but only a fraction of the nickel. IRC-50, also in the sodium form, was ineffective, adsorbing hydrogen ions only from the acid eluate. Work by McGarvey and colleagues (70) showed that zinc and copper had preferential exchange potentials to those of calcium and magnesium with TR-120. This is important as Stilfontein make-up water contains considerable quantities of these alkaline metal cations. Only nickel in this case was unknown with regard to its affinity for a cation resin in the presence of these other ions. However, adsorption runs with IR-120 in series with the anion resin column on elution, showed that nickel was preferentially adsorbed to that of calcium and magnesium. Therefore, the order of decreasing exchange potentials with IR-120 is copper, zinc, nickel, calcium, and magnesium. The effluent of sulfuric-hydrocyanic acid can then be fed to a lime neutralizer, the suspension allowed to settle, and the clear alkaline cyanide concentrate sent to the gold plant. Regeneration of the IR-120 resin was effected with 10% sodium chloride acidified to p H 2 to 3 with sulfuric acid. The regeneration curve has been plotted in Figure 8, showing simultaneous breakthrough of all three metal ions. Recovery of the sodium chloride regenerant was then carried out by adding caustic soda (or sodium carbonate) to the eluate to a neutral p H whereby the metallic hydroxides were precipitated out, filtered off, and the filtrate suitably adjusted for further use as an elutriant.
P-l
2
To W e
9
STEP 4.
I7-l
' - -
To waste
LEGEND.
Adsaption of MeM CyMidcbrrqerOs
Further Experiments and Suggestions for Operational Process
The process so far presents a two-column procedure for adsorption, where the first column removes the cyanide influent of the complexes and the second column the thiocyanate, Figure 1. With this arrangement and the usual flow rate, the effluent contained only cyanide (except during the elution period of the columns) which can also be used in the gold plant providing the plant mills its ore in cyanide solution. A double purpose is served in the case of milling in cyanide-the use of all the free cyanide and the re-use of water, which is so important to such a n arid country as South Africa. The available cyanide from the decomposition of the complexes can then be added to the ore pulp as a concentrate. However, Stilfontein Gold Mining Co.'s reduction plant uses cyanide only as a concentrate and this is added after ore milling. Therefore, it is necessary that all re-
Figure 11.
DiRctimoftbv.
Flow sheet for cyanide recovery
covered cyanide should be as a concentrate. This was achieved by taking advantage of the ability of free cyanide to be adsorbed completely on IRA-400 resin fully impregnated with cuprous cyanide, derived from the decomposition of the copper complex adsorbed from the waste cyanide solution or from a synthetic cupric cyanide. The uptake of ionic cyanide under these conditions was approximately equivalent to 18 grams per liter of wet-settled resin a t the break-
through. By allowing elution with 0.2N sulfuric acid to proceed for only 6 to 7 bed volumes as illustrated in Figure 9, no cuprous cyanide was lost and the capacity for cyanide ions was maintained for the following cycles, Table 11. TWO ways in which the cuprous cyanide conditioned columns may be incorporated in the ion exchange unit for the concentration of the free cyanide arewhether full recovery of the cyanide, but not water, is required ; or if the demand is VOL. 48, NO. 12
DECEMBER 1956
21 13
Table II. Adsorption of Cyanide Ions on Cuprous Cyanide Conditioned IRA-400 Resin Sorption Cycle No. 1 2
C S - Adsorbed on Wet Settled Resin,
% 1.89 1.86
3
1.84
4
1.83
for a partial water recovery and only an 85y0 cyanide recovery. The latter of these two requirements is accomplished by attaching the conditioned columns to the effluent outlet of the previously described unit (Figures l and 2). When both columns are in series, no thiocyanate will enter the conditioned columns. For the first requirement for full cyanide recovery, and as the cyanide ion is complexed with the cuprous cyanide in the presence of thiocyanate, the effluent from a complex loading column may be fed directly to a conditioned column (Figure 11) ; thereafter the effluent from this column is sent directly to waste as it is contaminated with the thiocyanate. A cuprous cyanide impregnated (or conditioned) resin column accounts for approximately one third to one half the cyanide ions when in series with a cyanide complex loading column. Therefore to have a n uninterrupted exhaustion run of the complex loading column, two condidoned columns were operated alternatively (Figure l l). When one conditioned column became exhausted, the other replaced it, and the former cupric cyanide loaded column eluted with the usual elutriant of 0.2N sulfuric acid. Practically 10Oy0 removal of the cyanide ions that had recombined with the cuprous cyanide, was obtained as shown in the elution cycle (Figure 9). The eluate from the conditioned columns contains some eluted thiocyanate that had occupied the remainder of the resin exchange sites during the columns first sorption cycle (Figure 9). However, the total amount of thiocyanate eluted is not very great and as the cyanide concentrate will be eventually diluted on addition to the gold ore pulp, the quantity reintroduced into the gold plant will be of little consequence. In order to interfere as little as possible with the present process of gold extraction-Le., solids to solution ratio-a higher concentration of cyanide is preferred. Therefore, to increase the present cyanide concentration so far obtained, the eluates from the conditioned columns, after readjustment with fresh sulfuric acid to the former 0.2h’strength, were re-used as an elutriant for the complex loaded columns. The efficiency of the sulfuric acid-hydrocyanic acid elutriant is plotted in Figure 10, giving a
2 1 14
comparison with a normal sulfuric acid elution. With the double use of the elutriant the final strength of cyanide concentrate for return to the gold plant was about 1.5 to Z.OOj, of potassium cyanide. The flow sheet for cyanide recovery is illustrated in a step by step operational cycle by Figure 11, and a brief description of what takes place during each step is given below; normal operational procedure such as water washing and backwashing are omitted. Step 1. Column A is adsorbing the complexes and initially the thiocyanate. The free cyanide ions in the effluent are complexed with the cuprous cyanide on D (which is in series with A ) and adsorbed. Step 2. Column D has been loaded with the cyanide complexed as cupric cyanide and is eluted with sulfuric acid. Column A is still adsorbing complexes and its effluent is now entering C. Step 3. Column A has been exhausted and now is eluted with reacidified used eluate. The eluate from A passes directly into E (sodium form) where the metal cations are adsorbed. The effluent from E after liming is now available for gold extraction. Column B now receives waste cyanide solution influent, with column D in series. Step 4. Column B adsorbs the coniplexes, D has become exhausted and is eluted with acid. Column C now receives the effluent from B. Column E, loaded with metal cations is regenerated with acidified 1OyOsodium chloride solution. The salt eluate is then neutralized with a n alkali to precipitate the metals for recovery. The clear solution after removal of the precipitates is returned after suitable salt and acid adjustment for re-use. Step 5 . The third adsorption cycle and columns A and D are in series as in Step 1. C is on eluation with the eluate passing to the used eluate tank for reacidification. Column B is eluted with the cation column E adsorbing the metal cations from B eluate. The procedure now repeats the former steps. Conclusion
It has been demonstrated that cyanide recovery from waste cyanide solution is possible by the use of the strong base 1RA-400 ion exchange (74). The process has been developed under restricted conditions imposed by the limited source of available economical reagents. However, a reasonable efficiency has been obtained which is an encouragement for further investigations under full pilot plant conditions. Although a considerable portion of this article deals with the sorption side of the process, there are a number of anomalies that are still left to be dealt with. hforeover, with the large number of simple and complex anions present, considerable physico-chemical study is required to elucidate more clearly the equilibria
INDUSTRIAL AND ENGINEERING CHEMISTRY
that exist between the anions and the resin bed. Elution with acid has proved fairly straightforward, and results of two dozen or so cycles have been remarkably consistent. Where deviations have occurred, these have been accounted for. The presence of the thiocyanate anion and its elution difficulties have complicated the process, and sacrifice either of water or full recovery of cyanide must be made. The choice, of course, does not apply to countries where there is an abundance of water. The conditioning of a strong baseresin bed with cuprous cyanide for sorption of ionic cyanide in the presence of thiocyanate ions is believed to be new to ion exchange. The ease by which the complex is decomposed and the cyanide recovered might possibly lead to further application, such as the cleansing of city and town effluents of the objectionable presence of cyanide. Acknowledgment
The author wishes to express his thanks to the management of Stilfontein Gold Mining Co. Ltd. for permission to publish this article. Appreciation is gratefully acknowledged to the metallurgical staff of the Strathmore Mines for their constructive criticism in the preparation of this manuscript and to Francis X. McGarvey, Rohm & Haas Co., Philadelphia, who kindly undertook to forward this article for publication in the United States. Literature Cited (1) Burstall, F. H., Forrest, P. J., Kember, N. F., Wells, R. A , , IKD. ENG. CHEM. 45,1648 (1953). ( 2 ) Can. Inst. Mining M e t . Trans. 44, 130 (1946). ( 3 ) Dorr, J. V. N., Bosque, F. L., “Cyanidation and Concentration of Gold and Silver Ores,” p. 188, McGraw-Hill, New York, 1950. (4) Ibid.,p. 265. (5) Hancock, H. A., Thomas, G., Can. Mining Met. Bull. 47, 539 (August 1954). ( 6 ) Kirk, P., Rohm & Haas Co., Philadelphia, Pa., private communication. Kunin, R., McGarvey, F. X., IKD. EX. CHEM.41, 1265 (1949). Kunin, R., Myers, R. J., “Ion Exchange Resins,” p. 66, Wiley, New York, 1951. McGarvey, F. X., Chem. Age (London) 69 iAun. 21, 1953). (10) McGarv&, F: X., ’Tenhoor, R. E., Nevers, R. P., IND. ENG. CHEM. 44, 534 (1952). (11) Sedgewick, N. V., “Chemical Elements and Their Compounds,” vol. 2, p. 1381, Clarendon Press, New York, 1950. (12) Stilfontein Gold Mining Co. Ltd., unpublished research reports. (13) Walger, C. A., Zabban, W:,Plating 40, 165-8. 165-8, 269-79 (1953). (14) Union of i f SSouth o u t h ’ Africa, Afrida, Patent No. 2636A/55 (Aug. 23, 1955). RECEIVED for review January 5, 1956 ACCEPTED June 2, 1956