Hydrometallurgy, 24 (1990) 135-156 135 Elsevier Science Publishers B.V., Amsterdam - - Printed in The Netherlands

Studies on the S o l v e n t E x t r a c t i o n of Gold from Cyan ide Media

P.A. RIVEROS

Canada Centre for Mineral and Energy Technology (CANMET), Mineral Sciences Laboratories, 555 Booth Street, Ottawa, Ont. KIA OG1 (Canada)

(Received January 17, 1989; revised and accepted August 17, 1989 )

ABSTRACT

Riveros, P.A.,1990. Studies on the solvent extraction of gold ~omcyanidemedia. Hydrometal- lurgy, 24:135-156.

The recovery of gold from real cyanide solutions using commercial quaternary amines was stud- ied. The quaternary amines exhibited good selectivity for gold over base metals, fast reaction kinetics, high loading capacity and low water solubility. The use of an aromatic diluent, such as Solvesso 150, was required to prevent the formation of emulsions. The feasibility of running a continuous circuit was demonstrated with a pilot plant campaign. Gold can be eluted efficiently with acidic thiourea and air sparging or recovered in metallic form by incinerating the organic solvent. Solvent extraction appears to have a great potential for the recovery of gold from cyanide solutions.

INTRODUCTION

A project is underway at the Canada Centre for Mineral and Energy Tech- nology (CANMET) to develop a solvent-in-pulp (SIP) system for recovering gold from leaching pulps. The long-term objective is to bypass the costly solid/ liquid separation step and, thus, to develop an advantageous alternative to the currently used carbon-in-pulp technology. Some preliminary results about this project have been recently reported [1,2]. Initially, several extractants, in- cluding some recently investigated organophosphorus extractants [ 3,4 ], were tested to determine their potential and suitability for a SIP application. In the course of this testwork, it was found that quaternary amines exhibited an ex- cellent combination of chemical and physical properties. This result prompted a more in-depth investigation into various aspects of the extraction of gold with quaternary amines. This investigation revealed that commercial quater- nary amines possess a good selectivity for gold over base metals, a high loading

0304-386X/90/$03.50 © 1990 Elsevier Science Publishers B.V.

136 P.A. RIVEROS

capacity and fast kinetics. It was further found that gold can be either stripped with acidic thiourea and air sparging or recovered in metallic form by incin- erating the organic solvent. A small pilot plant was run to demonstrate the feasibility of a continuous extraction process. This paper reports the results and discusses the potential of solvent extraction in the extractive metallurgy of gold.

E X P E R I M E N T A L

The commercial extractant Aliquat 336 [tri- (Cs-Clo)alkyl methyl ammo- nium chloride] from Henkel Corporation was used in most experiments. In addition, two new products supplied by Sherex Chemical Company Inc. and designated as Adogen 481 (tri-octyl methyl ammonium chloride ) and Adogen 483 (tri-tridecyl methyl ammonium chloride) were also tested. The aromatic diluent Solvesso 150 (Exxon Corporation) was used in all cases. The organic solutions were washed several times with aqueous solutions of sodium chloride before using.

The concentration of gold and other metals in aqueous solutions was deter- mined by Inductively Coupled Plasma (ICP) spectrophotometry. The organic solutions were analyzed by X-ray fluorescence methods.

QUATERNARY AMINES

The extraction of gold with quaternary amines has been the subject of pre- vious studies [ 5 ]. It has been established that quaternary amines have a great affinity for the gold cyanide complex and that the exchange reaction proceeds as follows:

R4N+X - + A u ( C N ) ~ =R4N+Au(CN)~ + X - (1)

where R is a long hydrocarbon chain and X - is a common anion such as C1- or H S 0 4 .

Previous attempts to apply quaternary amines to the recovery of gold have been hampered by the formation of massive emulsions when the organic was contacted with alkaline cyanide solutions [6,7]. This phenomenon is attrib- utable to the surfactant properties of most organic amines in the hydroxide form. Previous researchers have tested several modifiers with the purpose of correcting the emulsification problem, but surprisingly only aliphatic diluents have been used so far.

Shivrin et al. [6] established that the smaller the size of the hydrocarbon chains bonded to the ammonium group, the higher the tendency to emulsifi- cation. This phenomenon was attributed to the relatively higher water solu- bility of the amines with shorter chains. Amines with long hydrocarbon chains exhibited a smaller tendency to emulsification, but also a lower loading capac-

SOLVENT EXTRACTION OF GOLD FROM CYANIDE 137

ity. Thus, the most appropriate chain lengths appear to be R = Cs for the amines of the type R4N + X - and R = Cs-C lO for the amines of the type R3 (CH3) N + X .

Groenewald [7] found that Di-Isobutyl ketone (DIBK) was fairly effective in reducing the degree of emulsification. Although the use of DIBK enabled Groenewald to gather important information on the properties of quaternary amines, the relatively high water solubility of DIBK would preclude its use on an industrial scale.

In the present work, it was confirmed that emulsions formed readily when an aliphatic diluent, such as Varsol 140, was used. It was also found that the aromatic diluent Solvesso 150 produced good phase disengagements provided that some simple precautions were taken. Since fresh commercial quaternary amines have a considerable amount of water-soluble impurities, the organic solutions should be washed thoroughly with an aqueous solution of sodium chloride before using. Washing with dilute sodium hydroxide produced a severe emulsification. Slow phase disengagement and haziness were observed when the organic was washed with pure water, probably due to the low ionic strength. On the other hand, the phase separation is greatly improved once the organic phase is partially loaded with cyano complexes. For this reason, the operation of a continuous circuit was found to be practically emulsion-free. On a bench scale, experiments using 20% vol/vol solutions of extractant in Solvesso 150 showed satisfactory phase disengagements. This represents a significant im- provement over Groenewald's DIBK/Kerosene system which could not toler- ate concentrations higher than 4% Aliquat 336 due to the formation of emul- sions [7].

CYANIDE LEACH SOLUTIONS

Most experiments were done with a clarified cyanide solution from Camp- bell Red Lake Mines (Ontario). Since this solution had a high concentration of gold and base metals (Table 1 ), it was particularly useful to study the selec- tivity trends of quaternary amines. However, since most gold operations pro- duce solutions having lower concentrations of both gold and base metals, a second solution was obtained from Teck-Corona Hemlo Mines (Ontario) to compare the results. As shown in Table 2, this solution was more dilute and less contaminated with base metals. Both solutions had a pH of 10.5 and were used without modifications, except that NaCN was added as required to corn-

TABLE 1

Composition of the Campbell Red Lake solution (mg 1 1 )

Au Cu Fe Ni Zn Ag Co As Sb Ca

10 85 31 43 34 1.2 2.1 3.3 1.8 27

138

TABLE 2

Composition of the Teck-Corona Hemlo solution (mg 1-1 )

P.A. R1VEROS

Au Cu Fe Ni Zn Ag Co As Sb Ca

6 5 14 '2 0.2 0.2 0.1 0.3 2.3 115

3500 j

1 i2500 l

o 2000 1

1500 - ~ - o 1000 -

~ 500

o G,

o o o o o 5~ Aliquat 336/Campbel l~ oooao 10~ Aliquat 336/Campbelq AAA~ 20~ Aliquat 336/Campbel~ ~-~¢-¢ 25~ Aliquat 336/Hemlo !

5~ Adogen 481/Campbell ***** 5~g Adogen 483/Campbell j

W/

Y

0 2 4 6 8 10 12 14 16 A q u e o u s C o n c e n t r a t i o n of Gold, m g / L

Fig. 1. (]old ex t r ac t i on i so t he rms .

pensate for the gradual loss of free cyanide over time. No speciation studies were conducted as it is generally accepted that metals are present in their lower oxidation states and forming cyanide complexes anions, e.g., Au(CN)~-, Cu(CN)~- , Cu(CN)~- , Zn(CN)~- , N i (CN)~- and Fe(CN) 4- [8].

I,OADING CAPACITY

Figure 1 presents equilibrium isotherms for the extraction of gold under var- ious conditions. The gold loadings vary widely with the compositions of both the organic and aqueous phases. Aliquat 336 exhibited a higher loading capac- ity than both Adogen 481 and Adogen 483 at the same volume percent concen- tration. As expected, increasing the concentration of Aliquat 336 from 5% to 10% and 20% resulted in progressively higher gold loadings. The gold loadings from the less contaminated Teck-Corona Hemlo were quite high and thus a much lower concentration of Aliquat 336 (2.5%) was used to obtain compa- rable gold loadings. As will be discussed later, this is because the presence of zinc cyanide and other base metals has some influence on the gold loading.

SOLVENT EXTRACTION OF GOLD FROM CYANIDE 139

'S

- 4

0

L.)

0 , 4

,13

1000 ~ ~ _ _ ~ _ _ ~ _ ~ v

lOO

1 0 J i , i i i , i i I i , ~ i J i , i i I , , r r i , r ~ i

0 2000

ooooo 5~ Aliquat 336/Campbell ] aaaDD 10~ Aliquat 336/Campbel l i ~aaaa 30s~ Aliquat 336/Campbel l ] ~9999 2.Ssg Aliquat 336/Hemlo i

5~ Adogen 481/Campbell ***** 5~ Adogen 483/Campbell

I 000 3000 E q u i l i b r i u m G o l d L o a d i n g , m g / L

Fig. 2. Gold d i s t r i bu t ion coeff ic ients ve r sus gold loading.

Figure 2 presents the gold distribution coefficients, DAu, as estimated from the isotherms versus the equilibrium gold loadings. It can be seen that with an adequate concentration of extractant the DAu can reach values close to 1000. These gold loadings and distribution coefficients compare very favourably with those reported for other liquid extractants. For example, the DAu values ob- tained with 100% tributyl phosphate (TBP) and a 1 : 1 mixture of TBP/Alam- ine 336 were 8.4 and 30.1 respectively with the Campbell Red Lake Solution [9]. Miller et al. obtained DAu values ranging from 17 to 155 for di-butyl butyl phosphonate (DBBP) and from 6 to 55 for TBP using synthetic gold cyanide solutions [3 ]. Groenewald reported a maximum gold loading of only 0.52 g l- 1 for Aliquat 336 in kerosene/DIBK, also with a synthetic solution [ 7 ].

S E I , E C T I V I T Y

The selectivity of quaternary amines was studied by saturating a given vol- ume of the organic phase with a cyanide solution in a series of successive stage- wise contacts. The stagewise saturation procedure was as follows: One aliquot of organic (initially 50 ml) was contacted for 10 min at room temperature with a cyanide solution at an aqueous to organic (A: O ) ratio of 40:1. After complete phase disengagement both the organic and the aqueous phases were sampled and analyzed. This procedure was repeated several times using the same or- ganic and fresh aqueous solution.

Figure 3 presents the stagewise saturation of 5% Aliquat 336 with the Camp- bell Red Lake solution. It was found that there are marked differences in the

140 P.A. RIVEROS

2500

~2ooo. 0

1500

~I000 ~D

500

0 0

o o o o o Gold I ¢ ¢ ¢ 0 9 Nickel I ~oo~ C o p p e r I I~ . . . . I r o n j

1 2 3 4 5 6 7 N u m b e r of S t a g e w i s e C o n t a c t s

Fig. 3. Stagewise saturation of 5% Aliquat 336 in Solvesso 150 with the Campbell Red Lake so- lution. The aqueous to organic ratio was 40 : 1 at each stage.

0

0

,.0

1 0 0 0

100"

1 0

0.1

0

a-_~_~ 2 ×

ooooo Gold A a a a a Zinc O ~ Nickel ooooo Copper x x x x x I r o n

~ ; ~ ~ ; ~ 8 ; 1'0 N u m b e r of S t a g e w i s e C o n t a c t s

Fig. 4. Variation of the distribution coefficients during the stagewise saturation of 5% Aliquat 336 with the Campbell Red Lake solution.

extractability of each metal cyanide complex. While gold, zinc and nickel are extracted to a large extent, copper and iron are practically not extracted. Among the metals that are extracted, gold is the only one whose loading increased constantly as the organic became saturated. A comparison between the relative organic concentrations and the feed concentrations (Table 1 ) shows that gold is being concentrated in the organic phase with respect to all the other metals. Figure 4 presents the variation of the individual distribution coefficients throughout the stagewise saturation. It can be seen that Aliquat 336 exhibits the following order of affinity for the metal cyanide complexes:

S O L V E N T E X T R A C T I O N O F G O L D F R O M C Y A N I D E 141

Au > Zn > Ni > Cu,Fe (2)

Figure 5 presents the separation factors of gold over each metal as a function of the gold loading. The separation factors were calculated by dividing the individual distribution coefficients. Thus the separation factor of metal A over metal B is:

SA/~ =DA/DB (3)

It can be seen that the selectivity of Aliquat 336 for gold over copper and iron is very high. The selectivity for gold over nickel and zinc is less pronounced but nonetheless significant. There is a slight tendency for the separation fac- tors to increase as the gold loading increases, which suggests that a multistage countercurrent process would enhance the degree of separation.

Similar results were obtained with other extractants. Figure 6 presents the stagewise saturation of 5% Adogen 483. This extractant exhibited a slightly higher selectivity for gold but lower capacity as compared with Aliquat 336. Adogen 481 behaved similarly to Aliquat 336, except that a third phase formed at high loadings, possibly due to its shorter hydrocarbon chains. The third phase formation was readily cured, however, by using 2% isodecanol as a phase modifier.

Figure 7 presents the stagewise saturation of 2.5% Aliquat 336 with the Teck- Corona Hemlo solution. Gold is clearly extracted preferentially and in high

!000

r / ? ;

_ 100=

O

~ 1 0 -

0

0 . 1 ,

×

0 e--4

Of

~ SAu/Ni × × SAu/Fe

SAu/Zn

, , , , , , , , I , , , , , , , , , ] , , , ,

1000 2000 Equilibrium Gold Loading, mg/L

Fig. 5. Variation of the separation factors during the stagewise saturation of 5 J% Aliquat 336 with the Campbell Red Lake solution.

142 P.A. RIVEROS

~ 2000 ,

~ d 1500~

1 0 0 0 o

0

o . ~ 500

o 0

ooooo Gold 4 ~ A A Z i n c < ~ ¢ 9 0 N i c k e l o ~ a Copper × ~ ~×x Iron

/

J

~ s ~ s 6 7 N u m b e r o f S t a g e w i s e C o n t a c t s

Fig. 6. Stagewise saturation of 5% Adogen 483 in Solvesso 150 with the Campbell Red Lake so- lution. The aqueous to organic ratio was 40 : 1 at each stage.

.•2000 oO

~ 1 5 0 0 -

~ 1 0 0 0 - 0

0 ro

.~ 500

o

0

ooooo G o l d ! ~ ¢ 0 0 N i c k e l ] /p m m o D o Copper × × × × × I r o n / ~

1

/ /

/ /

v u ~ m

N u m b e r o f S t a g e w i s e C o n t a c t s

Fig. 7. Stagewise saturation of 2.5% Aliquat 336 in Solvesso 150 with the Teck-Corona Hemlo solution. The aqueous to organic ratio was 40 : 1 at each stage.

loadings from this dilute and less contaminated solution. This is attributable to the absence of Zn (CN)~- ions, which also results in slighter higher extrac- tions of nickel, copper and iron.

The separation of gold from copper and iron is particularly important. These two metals, which are usually associated with gold in many ores, give rise to several contamination problems in the refining of gold. Figure 8 summarizes the gold/copper separation factors, SAu/Cu, obtained from various stagewise saturation experiments. With the Campbell Red Lake solution, the SAu/Cu val- ues ranged from 300 to 700 whereas with the Teck-Corona Hemlo solution a

S O L V E N T E X T R A C T I O N O F G O L D F R O M C Y A N I D E 143

1 0 0 0 o ~ -

I00:

ESD

0 1 0 r..)

o r~

¢,

o o o o o 5 ~ A l i q u a t 3 3 6 / C a m p b e l l c c c [ ] a 1 0 ~ A l i q u a t 3 3 6 / C a m p b e l l ~ A A A 2 0 ~ A l i q u a t 3 3 6 / C a m p b e l l 0 ¢ ¢ ¢ 0 2 . 5 ~ A l i q u a t 3 3 6 / H e m l o × × × × × 5 ~ A d o g e n 4 8 1 ~ C a m p b e l l * * * * * 5 ~ A d o g e n 4 8 3 / C a m p b e l l

] , , J , i i , , i i i , T r T J i i i i , , , , f f , , J o 1 ooo 2000 5ooo

E q u i l i b r i u m Gold Loading, m g / L

Fig. 8. Separation factors of gold over copper.

1 0 0 0 o r.,3

o

1 0 0 I-,

I3,.,

cy]

o

"cJ

0

u 0 0

* * x x

¢

o o o o o 5 ~ A l i q u a t 3 3 6 / C a m p b e [ l mODDm 1 0 ~ A l i q u a t 3 3 6 / C a m p b e l l a a a a a 2 0 ~ A l i q u a t 3 3 6 / C a m p b e l l 0 0 0 0 0 2 . 5 ~ A l i q u a t 3 3 6 / H e m l o × × × × × 5 ~ A d o g e n 4 8 1 / C a m p b e l l * * * * * 5 ~ A d o g e n 4 8 3 / C a m p b e l l

' ' ' ' ' ' ' ' ' i , , , , , , , , , i , , , , , , , ( T

1 0 0 0 2 0 0 0 5 0 0 0

Equilibrium. Gold Loading, mg/L

Fig. 9. Separation factors of gold over iron.

lower degree of selectivity was observed. Similarly, Fig. 9 shows the corre- sponding separation factors for gold over iron, SAu/Ye. In this case, the SAu/Ee values ranged from 150 to 900 with the Campbell Red Lake and from 35 to 90 with the Teck-Corona Hemlo solution. The reason for the difference in the degree of selectivity is probably related to the concentration of zinc and to the ionic strength. Since the Campbell Red Lake solution contains more ions than

144 P.A. RIVEROS

the Teck-Corona Hemlo, its higher ionic strength appears to exert a relatively stronger influence on the extraction of copper and iron than on the extraction of gold.

Despite these fluctuations, it is apparent that quaternary amines exhibit a marked selectivity for gold over copper and iron. This result agrees with pre- vious work indicating that quaternary amines have a higher affinity for uni- valent ions, such as Au(CN)~ , than for multivalent ions, such as Fe(CN) 4- , Cu (CN) 43-, and Cu (CN) ~- [ 10 ]. However, this trend contrasts strongly with the behaviour of strong-base resins which, despite having a similar active group, exhibit no selectivity for gold [11]. This suggests that the size of the hydro- carbon chains attached to the ammonium group bears a strong effect on the selectivity exhibited by that group. A similar conclusion was stated by Shivrin et al. [6] who found that quaternary amines of the type R4N+X - and R~(CH3)N+X - exhibited a higher selectivity for gold than the quaternary amines of the type R2(CH3)2N+X - or R(CH3)3N+X -. The fact that the amines used in the present work belong to the first category of amines, i.e., R~ (CH3)N+X -, agrees with the above observation. Furthermore, most strong- base resins have the configuration R (CH3)3N + X - and are, therefore, related to the less selective family of quaternary amines.

C O N T I N U O U S C I R C U I T

It has been recognized that running a continuous solvent extraction circuit for gold presents a big difficulty because of the extremely high aqueous-to- organic ratio that is required [ 12 ]. Depending on the gold concentration in the feed solution and the desired loading, the A: 0 ratio could be in the range of 100-1000: 1. This would exceed by far the typical ratios found in practice. Moreover, such a high A: O ratio dictates that the phase mixing should be done with an organic-continuous dispersion in order to minimize the losses of or- ganic solvent.

After considering the alternatives, and bearing in mind that a small organic inventory is required to keep the volume of the cyanide solution within reason- able limits, it was decided to use a series of agitated glass columns specially designed and built for this experiment. A diagram of these columns is shown in Fig. 10. During the operation, the aqueous solution flowed continuously through a series of columns by gravity. Each column contained a given volume of organic, which was pumped periodically into the next up-stream stage. The columns were fitted with mechanical impellers of the pump/mix type in order to have the aqueous solution dispersed into the organic phase. By careful ad- justment of the impeller speed, weir controls, and aqueous flow rate, it was possible to operate the columns with a stable organic-continuous (water-in- oil) dispersion. When this condition was attained, two liquid zones were clearly discernible within the columns: Settled aqueous raffinate occupied the lower

SOLVENT EXTRACTION OF GOLD FROM CYANIDE 145

PHA 5'E I:.: I

AQUEOUS PHASE OUT

PUMP/MIX STIRRER

O 0RGANIC-CONTINUOUS DISPERSION

I@SETTLED AQUEOUS P~S~ CONTROL

Fig. 10. Extraction column for continuous countercurrent solvent ext rac t ion of gold,

Au

10.0

7.0 84 32 44 36 755

2.6 82 31 43 37 ~ 252

1.2 81 31 43 37 ~ 90 ---....~._.. /

8.2 79 30 40 19 ~ 0

AQUEOUS, mg/L ORGANIC, mg/L

Cu Fe Ni Zn Au Cu Fe Ni Zn

85 31 43 34 2300 76 0 557 3820

62 0 482 3470

62 O 480 3460

60 0 507 3300

8 0 0 0

Fig. 1l. Extraction profile for 10% Aliquat 336 and the Campbell Red Lake solution. Aqueous to organic flow ratio = 200 : 1.

three quarters of the column while the top quarter was filled with the organic- continuous dispersion.

The McCabe-Thiele procedure was applied to the distribution isotherms of Fig. 1 to choose appropriate combinations of flow rates, number of stages, and extractant concentration. The aqueous flow rate was set at about 1 Bed Vol- ume/min (50-60 ml min-1), which gave a residence time of 2-3 min in each column.

Figures 11 and 12 present extraction profiles at steady state for the Campbell

146 P.A. RIVEROS

AQUEOUS, mg/L ORGANIC, mg/L

Au Cu Fe Ni Zn Au Cu Fe Ni Zn

6.0 5 14 2 02 2580 82 0 733 9fl

2.6 5 14 1 0 ~ 568 161 0 364 21

0.6 5 14 0 0 ~ 94 232 0 70 0 ~ . ~ .....~

0.3 5 14 O 0 ~ 0 0 0 0 0

Fig. 12. Extraction profile for 5% Aliquat 336 and the Teck-Corona Hemlo solution. Aqueous to organic flow ratio = 400:1.

Red Lake and the Teck-Corona Hemlo solutions, respectively. As expected from the bench-scale data, gold was extracted with a good degree of selectivity, especially over copper and iron. Gold recoveries were estimated at 98% with the Campbell Red Lake solution and 95% with the Teck-Corona Hemlo solu- tion. These values could be readily increased by adding more stages to the circuit.

The loaded organic showed a 10-20% volume reduction with respect to the fresh organic. Although it is possible that some organic was lost through ad- sorption on the walls and tubing, most of this volume reduction is believed to be due to a shrinkage that takes place when a quaternary amine becomes loaded with gold. As discussed below, the water solubility of quaternary amines is quite low and could not account for the volume reduction. The amount of gold collected in the organic fractions agrees closely with the amount of gold ex- tracted from the feed solution. This fact gives support to the assumption that the organic is not lost but it shrinks when it becomes loaded with gold.

X-ray fluorescence analyses of the loaded organic showed that the extraction of minor metals was negligible. With the purpose of verifying this result, the raffinate was periodically sampled throughout the campaign and analyzed for silver, cobalt, arsenic, antimony, and calcium. The analyses, which are pre- sented in Table 3, underline the high selectivity of Aliquat 336 for gold over most elements. Apparently, solvent extraction, unlike activated carbon, pro- vides an effective way to separate gold from many common impurities. In a processing plant, this fact could reduce or eliminate the subsequent gold refin- ing steps.

The concentrations of free cyanide, thiocyanate, and the solubility of the

SOl,VENT EXTRACTION OF GOLD FROM CYANIDE

TABLE 3

Continuous circuit. Analysis of minor elements

147

Solution Concentrations (mg1-1)

Ag Co As Sb Ca

Campbell feed 1.2 2.1 3.3 1.8 27 Campbell raffinate:

Sample 1 1.1 2.1 3.0 1.7 27 Sample 2 1.1 2.1 2.8 1.6 07 Sample 3 1.2 2.1 3.4 2.0 22 Sample 4 1.2 2.1 3.1 1.6 21

Hemlo feed 0.2 0.1 0.3 2.3 115 Hemlo raffinate:

Sample 1 0.2 0.1 0.3 2.3 115 Sample 2 0.1 0.1 0.0 2.3 130 Sample 3 0.1 0.1 0.0 2.3 127

TABLE 4

Continuous circuit. Concentration of free cyanide, thiocyanate, and solubility of quaternary amines

Solution Free CN- SCN- Dissolved quaternary (mgl 1) (mgl -~) amine (mgl 1)

Campbell feed 449 351 < 1 Campbell raffinate:

Sample 1 426 321 12 Sample 2 390 396 9 Sample 3 386 304 11 Sample 4 385 302 4 Sample 5 380 289 3

Hemlo feed 270 52 3 Hemlo raffinate:

Sample 1 260 50 8 Sample 2 280 51 8 Sample 3 270 50 8

Aliquat 336 in the aqueous phase are important factors in a gold recovery pro- cess. Samples of the raffinate were taken periodically during the campaign and the analyses are presented in Table 4. The extraction of both cyanide and thi- ocyanate ion was small. As expected from reported values [5], Aliquat 336 exhibited a remarkably low water solubility which represents an important advantage from economical and environmental standpoints. In addition, the phase separation was excellent throughout the campaign. Only minor amounts

148 P.A. RIVEROS

of crud were occasionally observed with highly loaded organics, but this did not accumulate or interfere with the extraction. The physical characteristics of quaternary amines appear to be more suitable for a large-scale application as none of the problems associated with some organophosphorus extractants [ 13,14 ] were encountered.

STRIPPING

Because of their high affinity for the gold cyanide complex, quaternary amines are not readily stripped. Among the several procedures that were tested, only two showed promise and were selected for a more detailed study: (1) stripping with acidic thiourea solutions and air sparging, and (2) incineration of the organic solvent.

Acidic thiourea

This reagent is reportedly used in the U.S.S.R. for the elution of gold from strong-base resins [15]. In the presence of a mineral acid and thiourea, the gold cyanide complex decomposes and a gold-thiourea complex, Au[CS (NH2)2] + , forms. Since this complex is positively charged, it is not held by ammonium groups.

R4N+Au(CN)~ +2HX+2CS(NH2)2

= R 4 N + X - +Au[CS(NH2)2]2 X - + 2HCN (4)

where X - is usually C1- or H S 0 4 . The attractive features of this procedure are: (1) the extractant is simulta-

neously stripped and regenerated, and (2) an important amount of free cya- nide can be recovered and recycled. In the case of strong-base resins, some operational problems have been reported, such as [ 15 ]: (1) breakage of the resin beads due to osmotic shock, (2) precipitation of cobalti- and ferrocyanide compounds in the resin matrix, and (3) decomposition of thiourea during the gold electrowinning. In addition, the generation of the toxic gas hydrocyanic acid during the stripping requires strict safety precautions.

Quaternary amines are free from most of the above problems. Since they do not have a rigid matrix and extract little or no iron and cobalt, they are prac- tically immune to both osmotic shock and the precipitation of cobalti- and ferrocyanide compounds. More importantly, since quaternary amines exhibit fast reaction kinetics and high loading capacity for gold, the inventory of loaded organic in a processing plant would be relatively small. This fact would greatly facilitate the safe handling of hydrocyanic acid.

Groenewald [ 7 ] reported that the stripping of Aliquat 336 with thiourea in hydrochloric acid solutions was inefficient, producing a maximum concentra-

SO[NENT EXTRACTION OF GOLD FROM CYANIDE 149

tion of only 0.68 g Au 1-1 in the strip liquor. The same researcher obtained better efficiency with mixtures of thiourea and perchloric acid. However, no information was given about the possible oxidation effect of the perchloric acid on the quaternary amine or about the need to remove the perchlorate ion before recycling the extractant.

In the present work it was confirmed that solutions of thiourea in hydro- chloric or sulphuric acid produced poor strippings or gold. Further tests showed that the presence of hydrocyanic acid (HCN), which is generated during the stripping, has a strong influence on the stripping efficiency. Hydrocyanic acid is a gas at room temperature, but it is evolved from aqueous solutions at a slow rate because of its high solubility in water [ 16 ]. Unless it is removed at a fast rate, hydrocyanic acid tends to concentrate in the aqueous phase where it af- fects negatively the efficiency of the gold stripping as shown by eq. 4.

It was found experimentally that air sparging is an effective way to remove hydrocyanic acid from the strip liquors. The equipment used for the thiourea stripping with air sparging is shown in Fig. 13. The sparging of air accom- plishes both the continuous removal of HCN and the mixing of the two liquid phases. On leaving the stripping chamber, the air stream is forced through two traps filled with sodium hydroxide which capture the HCN and transform it into NaCN for recycling.

Figure 14 shows the HCN evolution (measured with Draeger tubes) when a sample of loaded organic was contacted with three different acid solutions by means of air sparging. The air flow rate was set at about 0.5 Bed Volumes per second. The results show that, despite the vigorous air sparging, HCN was evolved slowly since it took between 1-2 h to eliminate it completely from the

AIR IN

AIR FLOW ~--~

®

~ -----~AIR OUT

O ORGANIC/AQUEOUS/AIR MIXTURE

O NaOH TRAPS FOR HCN

~VALVE

Fig. 13. Equipment for stripping gold with acidic thiourea and air sparging.

150 P.A. RIVEROS

-~.200

< 150

o ~ 100

o 50

z

0

i oooo 0.5 Hydrochloric Acid ~na~ 1.0 NN Sulphuric Acid OGDD 1.0 N Sulphuric Acid +

0.5 M Thiourea

60 120 TIME, m i n u t e s

180

Fig. 14. Concentration of hydrocyanic acid in the stripping chamber during the acid stripping.

~ 2000-

o

1500 I

0

lOO0

o

o o 500

o

o o o

ooooo Acid washed Organic

: ~ - ~ Untreated Organic:

\ \ -<

\

i , , i , , , , , , , , , i

20 40 TIME, m i n u t e s

60

Fig. 15. The rate of gold stripping with acidic thiourea and air sparging.

stripping chamber. The elimination of HCN was much faster when thiourea was added to the acid solution. This may be the result of a faster decomposition of the cyanide complexes by the action of thiourea. In the absence of thiourea, the evolution of HCN proceeds at a slower rate, regardless of the type and the concentration of the acid.

It was found that, in the absence of thiourea, solutions of 0.5-2 N hydro- chloric or sulphuric acid were inefficient for stripping gold, even when applying air sparging. However, some of the less stable cyanide complexes were decom- posed and this fact provided a practical way to remove selectively some of the

SOINENT EXTRACTION OF GOLD FROM CYANIDE 151

impurities from the loaded organic. Washing with 1 N sulphuric acid was ef- fective for removing nickel and zinc, but some small amounts of gold were also stripped. On the other hand, washing with 1 N hydrochloric acid resulted in the removal of nickel with no stripping of gold, but zinc also stayed in the organic phase. The most satisfactory results were obtained with a mixture of 1 N H2SO4 and 0.05 N HC1, which removed most of the zinc and nickel without stripping any gold.

As shown in Fig. 15, washing the organic solution with acid had an important effect on the subsequent thiourea stripping of Aliquat 336. The stripping of gold was much faster with the acid-washed organic than it was with the un- treated organic. This is probably because a considerable amount of HCN is evolved and removed during the acid wash. As a result, lesser amounts of HCN are evolved during the stripping. This finding underlines the effect that the concentration of aqueous HCN has on the efficiency of the gold stripping.

The stripping procedure was tested on several samples of loaded organic. Two typical experiments are described below:

Experiment 1 A 50-ml sample of 10% Aliquat 336 solution, loaded with the Campbell Red

Lake solution in the continuous circuit, was put into the stripping chamber (see Fig. 13 ). The air sparging was started and 25 ml of a mixture of I N H 2 S 0 4

and 0.05 N HC1 were added. After two hours, the HCN concentration in the chamber, as measured with Draeger tubes, was within safe levels. The acid solution was allowed to settle and removed from the stripping chamber. A so- lution containing 0.5 M thiourea and 1 N sulphuric acid was introduced into

TABLE 5

Stripping of 10% Aliquat 336 with 0.5 M thiourea and 1 N H2SO4

Solution Initial Concentrations volume (mg l -~ ) (ml)

Au Cu Zn Ni Fe

Acid wash 25 1 9 6297 578 0 Strip liquor:

Fraction 1 20 3215 110 418 858 0 Fraction 2 20 1783 5 265 363 0 Fraction 3 20 575 0 180 6 0 Fraction 4 20 215 0 140 0 0 Fraction 5 20 68 0 99 0 0 Fraction 6 20 22 0 74 0 0

Loaded organic 50 2080 83 3860 578 0 Stripped organic 0 0 218 0 0

152 P.A. RIVEROS

the chamber. After 10 min of mixing, the spent thiourea solution was removed and a fresh aliquot of acid thiourea solution was added. The latter step was repeated several times until no gold was left in the organic phase. The air sparging was maintained constant during the experiment at a flow rate of 0.5 Bed Volumes per second. The analyses of the solutions are presented in Table 5.

Experiment 2 The stripping procedure was repeated with a 50-ml aliquot of 5% Aliquat

336, loaded with the Teck-Corona Hemlo solution in the continuous circuit. The analysis of the solutions is presented in Table 6.

These results demonstrate that the stripping of gold from quaternary amines with acidic thiourea and air sparging is quite effective. The organic solution was completely depleted of gold after the stripping procedure. The acid wash was effective in removing significant amounts of impurities such as zinc and nickel. Therefore, the strip liquors, which on recycling could attain several grams of gold per liter, are not excessively contaminated and should produce a high-purity final product.

The recovery of gold from thiourea solutions has been studied before [17]. It is believed that gold could be readily recovered by electrowinning or precip- itation. Due to the high concentrations of gold in the strip liquor the possible losses of thiourea during the electrowinning or precipitation would be only a minor factor on the overall economics of the process.

TABLE 6

Str ipping of 5% Aliquat 336 with 0.5 M thiourea and 1 N H2SO4

Solution Init ial Concentra t ions volume (mg 1-1 ) (ml)

Au Cu Zn Ni Fe

Acid wash 25 1 13 12 1615 0 Strip liquor:

Fract ion 1 20 5228 110 6 17 0 Fract ion 2 20 1192 5 8 1 0 Fract ion 3 20 160 0 6 0 0 Fract ion 4 20 24 0 7 0 0 Fract ion 5 20 4 0 6 0 0 Fract ion 6 20 2 0 7 0 0

Loaded organic 50 2228 55 137 734 0 Str ipped organic - 0 0 111 0 0

SOLVENT EXTRACTION OF GOLD FROM CYANIDE 153

Incineration of the organic solvent

As shown above, gold loadings in the range of 2-3 g Au per liter of organic solution can be readily obtained in a countercurrent operation. It is interesting to note that the major component of the organic phase is the diluent, whose volume concentration ranges from 80 to 97.5%. If this diluent was removed by distillation, the residual extractant would have a high concentration of gold. In Fig. 16 the isotherms of Fig. 1 have been modified to show the gold loadings per volume of extractant. It can be seen that for the more dilute organic solu- tions, e.g. 2.5% and 5% Aliquat 336, the gold loading per volume of extractant can reach very high values. The gold concentration in these cases suggests that incineration may be a viable way to recover the contained gold.

Incineration is practiced occasionally to recover gold from activated carbon. However, activated carbon burns slowly and at high temperatures, which re- quires the use of an additional fuel. Therefore, elution methods are preferred for activated carbon. By contrast, quaternary amines, being liquid hydrocar- bon compounds, burn readily at comparatively low temperatures. Since base metals are unlikely to be reduced to the metallic state under these conditions, the gold recovered by incineration would be of high purity.

Several samples of loaded organic were subjected to distillation and incin- eration to demonstrate the technical feasibility of the procedure. Solvesso 150 distilled readily below 100 °C under vacuum and the separation from Aliquat 336 was complete. The liquid residue (loaded Aliquat 336 ) burned gently with- out the need for additional fuel. No losses of gold were detected during the

~ 6 0 .

4o

~ " 2 0 -

"tJ

2

o

ooooo 5~ Aliquat 336/Campbell ooooo 10~ Aliquat 336/Campbell ~ a 20~ Aliquat 336/Campbell $0¢¢¢ 2.5~ Aliquat 336//Hemlo ××××× 5~ Adogen 481/Campbell *_**** 5~ Adogen 483/Carnpbel]

©

4, ~ & ~'o Aqueous C o n c e n t r a t i o n of

1~2 1'4 Gold, m g / L

Fig. 16. Equilibrium isotherms modified to show the gold loading per volume of extractant.

154 P.A. RIVEROS

procedure. X-ray diffraction and scanning electron microscopy analyses of the incineration solid residue indicated the presence of metallic gold, nickel oxide, zinc oxide, and silica. Hot dilute nitric acid removed readily the zinc and the nickel from the solid residue. The remainder was melted in a furnace at 1100 ° C to produce clean round particles of high-purity gold.

Although replacing the burned extractant on a continuous basis may appear expensive in principle, the estimated reagent cost would be within the accept- able limits established by a recent study [18]. Furthermore, the incineration process offers additional advantages, such as: (1) elimination of electrowin- ning or precipitation, (2) faster gold output, leading to a reduction of the gold inventory, (3) the production of high-purity gold, requiring no further refining steps. One possible way to reduce the reagent costs would be to use quaternary amines of a technical grade. Whether these technical grade quaternary amines would exhibit the same chemical and physical characteristics would have to be evaluated with additional testwork.

C O N C L U S I O N S

The solvent extraction of gold from cyanide media with quaternary amines, such as Aliquat 336, is a feasible and advantageous process. Commercial qua- ternary amines present several favourable characteristics, including good se- lectivity for gold, fast kinetics, high loading capacity, low water solubility, and good phase separation.

Compared to conventional technology, solvent extraction offers a significant reduction of the gold inventory, elimination or simplication of the refining steps, and simpler transportation requirements, all of which would lead to im- portant economical savings. These attractive features should encourage the current attempts to develop a solvent-in-pulp process, which would have a great impact on the extractive metallurgy of gold.

In addition, solvent extraction could be used in those operations that pro- duce clarified solutions, such as heap leaching. This increasingly popular tech- nique would benefit from the reduction in solution volume resulting from the faster kinetics of solvent extraction.

The stripping of gold from quaternary amines can be readily done with acidic solutions of thiourea and air sparging. Because of their fast kinetics and high capacity, the stripping of quaternary amines can done in a small circuit, which facilitates the handling of the generated HCN.

In those cases where high loadings of gold are obtained, the recovery of gold by distilling the diluent and incinerating the residue appears attractive. An aspect that should be explored is the use of technical grade quaternary amines with the purpose of lowering the reagent cost.

SOINENT EXTRACTION OF GOLD FROM CYANIDE 155

ACKNOWLEDGEMENTS

The au tho r is grateful to the Na tu ra l Sciences and Engineer ing Research Counci l of C a n a d a for provid ing f inancial support . T h a n k s are ex tended to Mr. G. Ri tcey for m a n y valuable suggest ions and comments . The help received f rom Dr. R. Molnar , Mr. X. Gao and Mr. P. P r u d ' h o m m e dur ing the develop- m e n t of the project is grateful ly acknowledged. I would also like to t h a n k Dr.

D. Muir for several useful discussions.

REFERENCES

1 Ritcey, G.M., Molnar, R., Riveros, P., Wong, E.W., He, P.J., Gao, X., Prud'homme, P. and Koegler, C., 1988. Solvent extraction of gold from cyanide leach slurries. In: Int. Solvent Extraction Conf. {ISEC'88), July 1988, Moscow.

2 Gao, X. and Molnar, R., 1987. Solvent extraction of gold from leach slurry. CANMET Rep. MSL 87-140 (TR), October 1987.

3 Miller, J.D., Wan, R.Y., Mooiman, M.B. and Sibrell, P.L., 1987. Selective solvation extrac- tion of gold from alkaline cyanide solution by alkyl phosphorous esters. Sep. Sci. Technol., 22(2,3): 487-502.

4 Mooiman, M.B. and Miller, J.D., 1986. The chemistry of gold solvent extraction from cyanide solution using modified amines. Hydrometallurgy, 16: 245-261.

5 Ritcey, G.M. and Ashbrook, A.W., 1979. Solvent Extraction: Principles and Applications to Process Metallurgy. Part II. Elsevier, New York, N.Y., pp. 380-385.

6 Shivrin, G.N., Basov, A.S., Laskorin, B.N. and Shivrina, E.M., 1966. Extracting noble metals from cyanide solutions with quaternary ammonium compounds. Tsvet. Metal., 39: 19-23.

7 Groenewald, T., 1974. A modified diluent for the solvent extraction of Gold (I) cyanide from alkaline solutions. In: Proc. Int. Solvent Extraction Conf. (ISEC'74). Soc. Chem. Ind., Lon- don, pp. 2715-2732.

8 Osseo-Assare, K., Xue, T. and Ciminelli, S.T., 1984. Solution chemistry of cyanide systems. In: Precious Metals, Proc. Int. Symp. Metall. Soc. AIME, Los Angeles, Calif., Febr. 27-29, 1984.

9 He, P.J., 1987. The solvent extraction of aurocyanide with TBP and TBP-Alamine 336. CANMET Rep. MSL 87-108 (OP&J).

10 Irving, H.M.N.H. and Damodaran, A.D., 1971. The extraction of complex cyanides by liquid ion exchangers. Anal. Chim. Acta, 53: 267-275.

11 Fleming, C.A. and Cromberge, G., 1984. The extraction of gold from cyanide solutions by strong- and weak-base resins. J.S. Afr. Inst. Min. Metall., 84(5): 125-137.

12 Mooiman, M.B., Miller, J.D., Hiskey, J.B. and Hendriksz, A.R., 1984. Comparison of process alternatives for gold recovery from alkaline cyanide solutions, In: J.B. Hiskey (Editor), Heap and Dump Leaching Practice. SME/AIME, Salt Lake City, Utah, pp. 93-108.

13 Sibrell, P.L. and Miller, J.D., 1986. Soluble losses in the solvent extraction of gold from alkaline cyanide solutions by modified amines. In: Int. Symp. Solvent Extraction (ISEC'86), Vol. II, Munich, 11-16 Sept., 1986.

14 Sibrell, P.L., 1985. Phase disengagement and organic losses in the solvent extraction of gold. Thesis. Univ. Utah, Dep. Metall. Eng.

15 Fleming, C.A. and Cromberge, G., 1984. The elution of aurocyanide from strong- and weak- base resins. J.S. Afr. Inst. Min. Metall., 84(9): 269-280.

156 P.A. R1VEROS

16 McNamara, V.M., 1985. Canmet acidification volatilization reneutralization process fl~r cy- anide recovery and environmental control. CANMET Rep. MSL 85-44 (IR).

17 Desch~nes, G., Prud'homme, P.J. and Wong, E.W., 1986. Recovery of gold from thiourea solutions: Investigation of cementation with metal powder, activated carbon, ion exchangers, and solvent extraction. CANMET Rep. MSL 86-26 (TR).

18 Muir, D.M., 1988. Prospects for the recovery of gold from cyanide liquors and pulps by solvent extraction and by ion exchange. CANMET Rep. MSL 88-78 (IR).