Hydrometallurgy 72 (2004) 225 – 234
The recovery of gold from ammoniacal thiosulfate solutions
containing copper using ion exchange resin columns
Hongguang Zhang 1, David B. Dreisinger *
Department of Metals and Materials Engineering, The University of British Columbia, 309-6350 Stores Road, Vancouver, BC, Canada V6T 1Z4
Received 7 October 2002; received in revised form 8 April 2003; accepted 2 July 2003
The loading of gold and copper, both individually and simultaneously, from thiosulfate solutions onto ion exchange resin
columns and the subsequent elution of these species have been investigated. In the presence of copper, effective loading with
good selectivity for gold can be achieved at pH 11, which balances the stability of the solution and minimizes the formation of
poisoning polythionates. The loaded metals can be eluted with high efficiencies using a number of composite eluant solutions.
These include sodium tetrathionate stabilized with sodium sulfite and ammonium thiosulfate (ATS), sodium sulfite coupled with
ammonia, and sodium chloride with the addition of ATS. The addition of thiosulfate to the eluant seems to have a negative
effect on the gold elution efficiency and should be avoided wherever it is possible. Of all the three eluants investigated, the
sodium sulfite/ammonia combination is found to be the most efficient, but the sodium chloride/ATS combination is likely the
D 2004 Elsevier B.V. All rights reserved.
Keywords: Gold recovery; Copper; Thiosulfate; Ion exchange
1. Introduction leaching reaction has been proposed as (Li et al.,
The leaching of gold from its ores using thiosulfate
solutions has been extensively studied in the last two Au þ CuðNH3 Þ2þ þ 4S2 O2À
decades (Zipperian and Raghavan, 1988; Tao et al.,
1993; Abbruzzese et al., 1995; Li et al., 1996). It is ! AuðS2 O3 Þ3À þ CuðS2 O3 Þ3À þ 4NH3
2 2 ð1Þ
found that the dissolution of gold in a thiosulfate
solution is usually very slow, unless catalyzed by
CuðS2 O3 Þ3À þ O2 þ 2H2 O þ 16NH3
copper and ammonia. A possible mechanism for the
! 4CuðNH3 Þ2þ þ 8S2 O2À þ 4OHÀ
4 3 ð2Þ
* Corresponding author. Fax: +1-604-822-3619.
E-mail address: email@example.com (D.B. Dreisinger). In most studies, ammonium thiosulfate (ATS), which
Now at A.J. Parker Cooperative Research Centre for Hydro- is primarily used as a fertilizer, is used as the leaching
metallurgy, Murdoch University, Murdoch, WA 6150, Australia. agent as it is inexpensive and already contains the
0304-386X/$ - see front matter D 2004 Elsevier B.V. All rights reserved.
226 H. Zhang, D.B. Dreisinger / Hydrometallurgy 72 (2004) 225–234
ammonia required to support the copper catalysis of has been shown (O’Malley and Nicol, 2001; Zhang
gold leaching. Copper is usually added to the leach and Dreisinger, 2002a) that tetrathionate (S4O6 À) and
solution as cupric sulfate. In many cases, sufficient its degradation product, trithionate (S3O6 ), strongly
copper may be naturally leached from the ore. The inhibit the adsorption of gold and copper onto the
efficiency of the leaching depends on many factors, resins. These results are consistent with the findings of
including pH, Eh, temperature, ammonia-to-thiosul- polythionates being poisonous to the resins used in the
fate ratio, solution stability, etc. However, satisfactory recovery of uranium (Nugent, 1956; Merritt, 1971).
gold recovery can be achieved under well-maintained Fortunately, it has also been shown that tetrathionate
conditions. Because of the advantages of low toxicity can be effectively removed by alkaline decomposition
and low reagent costs, thiosulfate leaching is proba- at ambient or slightly elevated temperatures (Zhang
bly, at the present time, the most promising method to and Dreisinger, 2002b)2. The experimental conditions
replace the conventional cyanidation process for gold for the current work could then be established.
extraction (Ritchie et al., 2001).
However, one of the problems associated with
thiosulfate leaching is the difficulty in recovering gold 2. Experimental
from pregnant solutions. For example, it has been
shown that the gold – thiosulfate complex does not Commercially available strongly basic anion ex-
effectively adsorb onto activated carbon (Gallagher et change resins were obtained in their wet forms and
al., 1990) and hence the carbon-in-pulp technology is used without treatment. The resins involved in this
not applicable to the thiosulfate system. Cementation work, including Dowex G51, Dowex 21K and Amber-
with zinc or copper (Berezosky and Sefton, 1979; lite IRA-410, were selected from a group of resins that
Perez and Galaviz, 1987; Guerra, 1997) has also been had been investigated previously (Zhang and Drei-
studied but this method suffers from high zinc con- singer, 2002a). These resins are all of gel type, with a
sumption or possibly passivation by cuprous sulfide polystyrene divinylbenzene matrix and quaternary
formation on the copper cement surface. Recent ammonium functional groups. Other properties are as
publications and patents (O’Malley and Nicol, 2001; given in Table 1. The ion exchange column was
Zhang and Dreisinger, 2002a; Fleming et al., 2002) constructed with a 25-mL burette with small glass
have shown the possibility of using ion exchange wool plugs both at the bottom and on the top of the
resins to recover gold from thiosulfate solutions and resin bed. The volume of the bed was approximately
leach pulps. Therefore the loading and elution behav- taken as 1.5 mL/g of wet resin. The solution was
ior of gold and copper has been investigated in the pumped in through a rubber stopper on the top of the
current work using ion exchange resin columns and burette using a peristaltic pump. The effluent was
the results are now presented. collected from the tip at the bottom.
Another problem associated with the use of thio- Synthetic solutions prepared with deionized water
sulfate is its oxidation by dissolved oxygen, which is were employed for the study. Ammonium thiosulfate
also strongly catalyzed by copper (Li et al., 1996). (ATS) obtained from Aldrich was 99% pure. All other
Given the high redox potential, the Cu(II) species can chemicals were of analytical grade. Gold was intro-
oxidize thiosulfate to tetrathionate. duced to thiosulfate solution in the form of AuCl4 , as
the commercial reference solution (1000 F 1 ppm Au
with 10% HCl), which was immediately reduced by
2CuðNH3 Þ2þ þ 6S2 O2À ! S4 O2À þ 2CuðS2 O3 Þ3À
4 3 6 2 thiosulfate to form Au(S2O3)2 À. Copper was added as
þ 8NH3 ð3Þ CuSO4 5H2O. All loading experiments were carried
out at the ambient temperature (23 – 25 jC) and pH 11.
The cupric species is regenerated by dissolved oxygen
through Eq. (2). The undesirable oxidation of thiosul-
fate not only remarkably increases reagent consump- 2
In initial laboratory experiments conducted at the University
tion and hence operating cost, but also undermines the of British Columbia, the destruction of trithionate appears to be
subsequent process for gold recovery. For example, it . more difficult than the destruction of tetrathionate.
H. Zhang, D.B. Dreisinger / Hydrometallurgy 72 (2004) 225–234 227
Table 1 arises due to the small change in copper concentration
Physical and chemical properties of the resins used between the feed and product solution. The change in
Resins Ionic pH Moisture Bead Capacity copper concentration is small relative to the precision
form range (%) size (meq/mL)
of the analysis procedure, introducing the possibility
of poor balances.
Dowex G51 ClÀ 0 – 14 43 – 48 0.3 – 0.9 1.4
Dowex 21K ClÀ 0 – 14 43 – 48 0.6 – 1.2 1.2
Amberlite IRA-410 ClÀ 0 – 14 42 0.48 1.35
3. Results and discussion
Such a pH was found to be optimal for the following 3.1. Loading of gold or copper only
reasons. Firstly, the Cu – ATS solution tends to decom-
pose at lower pHs, forming copper sulfides, while at The results for gold loading from 0.1 M ATS
higher pHs the chance for copper hydroxide precipi- solutions onto the columns of Dowex G51 and Dowex
tation increases. Secondly, the resins can be prevented 21K resins are shown in Fig. 1. The effluent was
from poisoning at pH 11 since tetrathionate, which collected in fractions at different times in order to
may be formed through Eq. (3) prior to and during the demonstrate the kinetics of the loading process. On
loading, is not stable in strongly alkaline solutions and G51, 300 bed volume (BV) loading solution contain-
quickly decomposes to thiosulfate and sulfite (Zhang ing 20 ppm Au was pumped through the column at a
and Dreisinger, 2002b). The loading solution was rate of 13 BV/h and virtually no gold was detected in
prepared in such a way that the ATS was dissolved the effluent. This indicates that the ion exchange
first with the addition of a suitable amount of NaOH to reaction was fast since the retention time of the
raise the pH to slightly above 11 before CuSO4 was solution in the resin bed was less than 5 min. On
added. The solution was then adjusted to pH 11 by 21K, 630 BV solution was used and almost exactly
NaOH or H2SO4 and used immediately for loading the same loading as on G51 was obtained for the first
experiments. The elution of gold and copper from 300 BV, but with the effluent gold concentration being
loaded resins was carried out using different eluant slightly higher. As more solution passed through the
solutions with their compositions described in the text. column, gold loading increased almost linearly, reach-
An automatic fraction collector was used to take ing about 18 kg/t, in correspondence to 0.7 ppm gold
solution samples during either loading or elution if the in the effluent. Both columns were not saturated.
kinetics were to be followed. In other experiments, the These results agree well with those obtained previ-
loading or elution effluents were collected as a whole
only for the final results. Gold and copper in solution
were analyzed by commercial analytical laboratories
using fire assay and ICP, respectively. The loading Q,
presented as the amount of a metal on unit mass of
wet resin, as well as the elution efficiency, could then
be calculated. Yet, more reliable loading values could
be determined from elution, i.e. from the total amount
of metals eluted and remained, the latter being found
out by resin assay. Accordingly, the elution efficien-
cies could be calculated on the same basis. As found
both previously (Zhang and Dreisinger, 2002a) and in
the current work, the mass balance results for gold
loading and elution from both calculation procedures
are generally well consistent. However, the results for
copper calculated from solution analysis are not Fig. 1. Loading of gold onto columns of Dowex G51 (flow rate 14
reliable or consistent with the loading and elution BV/h) and Dowex 21K (flow rate 13 BV/h) from 0.1 M ATS
values calculated by resin assay. This inconsistency containing 20 ppm Au at pH 11.
228 H. Zhang, D.B. Dreisinger / Hydrometallurgy 72 (2004) 225–234
ously from batch studies (Zhang and Dreisinger, saturated. Accordingly the copper loading reached a
2002a) and indicate that gold can be removed from maximum. In the last stage, Cu concentration in the
pure solutions rapidly and loaded on resins to very effluent increased slowly exceeding the initial level
high concentrations. and Cu loading declined slightly, indicating that some
In real leach solutions, however, copper is almost of the loaded copper returned to the solution. This
always present, usually in much higher concentrations may be attributed to the presence of a small amount of
than gold, since it is added as a catalyst for gold tetrathionate or trithionate in the original loading
dissolution. Therefore, the interaction between copper solution, which may have not completely decomposed
and the resins was investigated. Typical kinetic curves under the given conditions. The deleterious species
for the loading of copper on Dowex 21K resin from accumulated on the resin as more solution passed
0.1 M ATS containing 500 ppm Cu are presented in through the column, replacing some of the already-
Fig. 2. As can be seen, the change in effluent Cu loaded copper. Similar curves were recorded with
concentration underwent three stages as a total of 125 Dowex G51 resin, but with somewhat higher loading
BV solution passed through the column. In the first capacity. The maximum Cu loading obtained on 21K
stage (0– 20 BV), Cu concentration was maintained at was about 23 kg/t while it was about 30 kg/t on G51.
about 120 ppm. This probably reflected the presence
of some Cu(II) in the original loading solution since 3.2. Simultaneous loading of Au and Cu
the cationic Cu(II) – ammine complex would not be
expected to load with the anion exchange resin. The kinetics for the co-adsorption of gold and
However, this level varied in different experimental copper on Dowex 21K resin from a solution contain-
runs possibly because the ratio of Cu(II)/Cu(I) was not ing both metals is demonstrated in Fig. 3. Different
steady due to the reaction in Eq. (3). It was believed flow rates were used but a similar trend to the single
that virtually all the anionic Cu(I) – thiosulfate com- metal tests was observed. It can be seen from Fig. 3a
plex in the original solution had been adsorbed by the that the kinetic curves for gold loading in the presence
resin in the first stage. As a result, the amount of of copper are closely similar to those obtained using
copper loaded on the resin increased rapidly with pure solutions for the initial 200 BV. After this point,
increasing solution volume. In the second stage gold broke through the column with its concentration
(20 – 80 BV), the breakthrough of Cu(I) occurred in the effluent increasing sharply, indicating the resin
and the total Cu concentration in the effluent rose was near saturation. Apparently, the capacity for gold
gradually to the same level as in the initial loading on the resin was much lower as a result of competitive
solution (dashed line), suggesting the column was adsorption of copper. It can also be seen that the flow
rate at which the solution passed the column affected
gold loading appreciably. With the higher flow rate,
the breakthrough occurred earlier but the saturation
was achieved later. This is because at the higher flow
rate, the solution retention time was shorter, the ion
exchange reaction was less complete and so more
solution was required to saturate the column.
The kinetic curves for copper loading in the pres-
ence of gold are shown in Fig. 3b. As it is illustrated,
copper loading increased very rapidly to the saturation
level within about 50 BVand then dropped gradually as
more solution was provided. No effect of flow rate was
seen, probably because the kinetics of the process was
so fast that the change from 13.3 to 6.9 BV/h did not
Fig. 2. Loading of copper onto a column of Dowex 21K resin from make an appreciable difference in copper loading. A
0.1 M ATS containing 500 ppm Cu at a flow rate of 15 BV/h and complete picture for the co-loading of gold and copper
pH 11. was then obtained by combining Fig. 3a and b. At the
H. Zhang, D.B. Dreisinger / Hydrometallurgy 72 (2004) 225–234 229
due to the fact that the difference between the Cu
concentrations in the initial and final solutions was not
sufficiently large compared with the error in copper
analysis. For comparison, a sample of loaded resin
was analyzed giving gold and copper loadings to be
9.17 and 9.75 kg/t, respectively. The copper loading
from resin analysis was then used as a reference in the
The concentration of gold in practical leach solu-
tions may well be below 20 ppm, the level used in the
above experiments. Therefore, it is necessary to dem-
onstrate that gold can be loaded effectively on resins
from solutions containing very low levels of gold. Fig.
4 shows the results of gold loading on Dowex 21K
resin from solutions comprising of 0.1 M ATS, 1 ppm
Au and various amounts of Cu. As can be seen, gold
loading is very much dependent on the copper level in
solution. In the case of 100 ppm Cu, high gold loading
(5.5 kg/t) was achieved when sufficient amount of
solution was pumped through the column. As the Cu
concentration increased to 200 and 500 ppm, the
maximum gold loading decreased to 3.3 and 1.6 kg/
t, respectively. However, copper loading was found by
resin assay to remain similar to the previous values,
Fig. 3. Co-loading of gold (a) and copper (b) onto Dowex 21K resin regardless of the feed Cu concentration in solution.
columns at different flow rates from 0.1 M ATS containing 20 ppm
Au and 500 ppm Cu (pH 11). 3.3. Elution
beginning, the column was quickly saturated with In the elution experiments, gold and copper were
copper due to its high concentration in the loading pre-loaded onto the resin columns from a solution
solution. As more solution passed, however, gold containing 0.1 M ATS, 20 ppm Au and 500 ppm Cu
accumulated on the resin, taking the place of copper.
Therefore the amount of copper on the resin decreased
as that of gold increased. Considering the fact that the
effluent gold concentration was much lower than
copper, it can be concluded that the adsorption of gold
on the resin is much stronger than copper. This result
favors the selective loading of gold over copper.
The co-loading of gold and copper was repeated a
number of times under the same conditions as de-
scribed in Fig. 3. However, no fraction samples were
taken but, instead, the total amount of effluent was
collected over the entire period of the experiments.
Reproducible final gold loading values ranging from
8.70 to 9.21 kg/t were obtained. These numbers were
calculated from solution analysis. On the other hand, Fig. 4. Loading of gold from 0.1 M ATS solutions containing 1 ppm
the calculated numbers for copper loading were found Au and various concentrations of Cu at a flow rate of 30 BV/h and
to vary significantly between 5 and 25 kg/t. This was pH 11.
230 H. Zhang, D.B. Dreisinger / Hydrometallurgy 72 (2004) 225–234
using a flow rate of about 15 BV/h. The elution was
conducted immediately after the completion of the
A number of eluant solutions were used to strip the
gold and copper from the resins. First of all, tetrathi-
onate was selected as an eluant since it strongly
competes with the metals to adsorb onto the resins
(Zhang and Dreisinger, 2002a). Ammonium thiosul-
fate was added to the stripping solution to stabilize the
eluted gold and copper. A low concentration of
tetrathionate (0.02 M) was initially employed but the
elution was not complete. An attempt to raise the
tetrathionate concentration, however, led to the for-
mation of a white precipitate. This is most likely due Fig. 5. Elution of gold from columns of Dowex G51 (8.92 kg/t Au)
to the reaction between tetrathionate and thiosulfate to and Dowex 21K (18.83 kg/t Au) with an eluant composed of 0.25
form pentathionate (S5O6 À), which soon decomposes
2 M Na2S4O6, 0.5 M ATS and 0.25 M Na2SO3 with the pH adjusted
to 9 (flow rate 2 BV/h).
to precipitate elemental sulfur (Lyons and Nickless,
shown in Fig. 6. It is clear from the comparison with
S4 O2À þ S2 O2À WS5 O2À þ SO2À
6 3 6 3 ð4Þ
Fig. 5 that the elution of copper was faster than that of
gold, with the maximum concentration in the eluate
To prevent the formation of sulfur, sulfite (SO3 À) was
appearing at about 5 BV. However, the elution curve
added according to the above reaction, and this proved
also shows a tail similar to those in Fig. 5, indicating
effective. However, the following reaction is also
the slow kinetics for the removal of the last 20%
likely to occur upon the addition of sulfite (Lyons
copper from the resin. A copper recovery of 96.5%
and Nickless, 1968) to form trithionate (S3O6 À):
was obtained at about 30 BV.
S4 O2À þ SO2À WS3 O2À þ S2 O2À
6 3 6 3 ð5Þ Fig. 7 shows the elution of gold and copper
together with the tetrathionate eluant from a resin
Therefore the tetrathionate eluant solution may well loaded with both metals. The results agree very well
be a mixture of tetrathionate, trithionate, thiosulfate with those from Figs. 5 and 6 in that copper was
and sulfite. In terms of elution, trithionate is expected removed from the column faster than gold in the
to be as effective as tetrathionate, since it is also a initial stage of elution. In the later stage (about 20%
poison to the resins (O’Malley and Nicol, 2001). The metal remaining), the elution of both gold and copper
mixture is referred to as the tetrathionate eluant in this became much slower. Nearly complete elution of both
paper usually with a composition of 0.25 M Na2S4O6, metals was achieved at about 50 BV. Consistent
0.25 M Na2SO3 and 0.5 M ATS though other compo- elution efficiencies (99.7% Au and 99.3% Cu) were
sitions may also be used. obtained from separate experiments where the effluent
Fig. 5 shows the kinetic curves for gold elution was collected as a whole with all 50 BV eluant,
from resins loaded with gold only using the tetrathi- generating an eluate solution of 120 ppm Au and
onate eluant. As can be seen, the initial elution rate 130 ppm Cu. It should be mentioned that similar
was fast with the maxima gold concentration in the elution results could be achieved using the tetrathio-
eluate reached at 10 – 15 BV. In the later stage, nate eluant with a different composition, e.g. 0.1 M
however, the elution rate slowed down as indicated Na2S4O6, 1 M Na2SO3 and 0.1 M ATS.
by a long ‘‘tail’’ following the maxima on the kinetic Sodium sulfite alone was also found to be very
curves. The total elution efficiencies at about 55 BV effective for gold and copper elution. However, cop-
were 95.3% (G51) and 98.7% (21K), respectively. per hydroxide precipitated in the eluate solution and
The kinetic curve for the elution of copper from a this was prevented by the addition of ammonia. A
copper loaded column using the tetrathionate eluant is separate experiment showed that ammonia itself was
H. Zhang, D.B. Dreisinger / Hydrometallurgy 72 (2004) 225–234 231
Fig. 6. Elution of copper from a column of Dowex G51 (24.7 kg/t Cu) Fig. 8. Elution of gold and copper from a column of Dowex 21K
using the same eluant as for Fig. 5 (flow rate 1.3 BV/h). (8.87 kg/t Au and 7.66 kg/t Cu) using an eluant composed of 2 M
Na2SO3 and 1 M NH3 with the pH adjusted to 11 (flow rate 2
not able to strip the metals. Interestingly, gold was
stable in the eluate solution and may have been copper as compared to the tetrathionate eluant. As a
present as Au(SO3)2 À, which is known to exist and
3 result, complete elution of gold and copper required
has found ample applications in gold plating baths only about 25 BV of the eluant solution.
(Zur and Ariel, 1982). Alternately, the gold complex Further experimental results given in Table 2 con-
in solution may have been stabilized by the small firm that sodium sulfite is more efficient than the
amount of thiosulfate present. Fig. 8 shows the kinetic tetrathionate eluant for the elution of gold and copper.
curves for the elution of gold and copper using the With 33 and 37 BV eluant comprising of 2 M Na2SO3
sulfite eluant. It is evident from Fig. 8 that both gold and 1 M NH3, gold and copper were completely
and copper were stripped rapidly from the resin removed from the resin, generating eluate solutions
column, with the maximum concentrations achieved with at least 170 ppm Au. Table 2 also shows that the
within 10 BV. More significantly, no obvious tails addition of 0.1 M ATS to the eluant solution greatly
were observed on the elution curves for both gold and reduces the elution efficiency, particularly for gold.
With a total of 53 BV eluant containing ATS, only
85.5% Au was stripped. This result may explain the
‘‘tail’’ effect of the tetrathionate eluant that contains
ATS. The presence of thiosulfate might have hindered
the elution of gold, and possibly copper as well, in
some way which is not currently understood.
Another effective eluant studied was sodium chlo-
ride. The addition of not only ammonia, but also
Elution of Au and Cu from Dowex 21K resin with the sodium
sulfite eluant added with ammonia and ATS
Eluant composition Total Au elution Cu elution
BV (%) (%)
2 M Na2SO3, 1 M NH3 33 99.9 98.0
2 M Na2SO3, 1 M NH3 37 99.9 100.0
Fig. 7. Elution of gold and copper from a column of Dowex 21K
2 M Na2SO3, 1 M NH3, 38 74.4 98.6
(9.12 kg/t Au and 8.19 kg/t Cu) using the same eluant as for Fig. 5
0.1 M ATS 53 85.8 98.7
(flow rate 2 BV/h).
232 H. Zhang, D.B. Dreisinger / Hydrometallurgy 72 (2004) 225–234
thiosulfate, was necessary to stabilize the eluted Table 4
metals in the eluate solution. Table 3 shows the results Results of loading/elution cycling experiments from Dowex G51
resin (2.00 g) using Eluant A
of gold and copper elution with sodium chloride
Cycle Metals loaded Metals eluted Recovery
solutions. With 2 M NaCl and 1 M NH3 (27 BV)
gold elution was only 16.4%, while copper was Au/mg Cu/mg Au/mg Cu/mg Au (%) Cu (%)
completely removed. In accordance with the low gold 1 19.66 18.96 17.87 16.24 90.9 85.6
elution, heavy gold deposit was observed on the resin 2 19.30 41.80 19.23 18.12 99.7 43.3
3 19.48 20.00 19.20 18.02 98.6 90.1
surface. Obviously, gold was first stripped from the
4 19.16 25.16 18.91 17.68 98.7 70.3
resin and then decomposed due to the lack of com- 5 19.10 9.00 19.13 18.60 100.2 206.7
plexing agent. The addition of 0.1 M ATS prevented Overall 96.68 114.9 94.32 88.66 98.5a 99.0a
gold deposition. However, gold elution was not satis- a
Total metals remained on resin: Au 1.43 mg and Cu 0.86 mg.
factory using 2 M NaCl, with gold recoveries being
only 71.8% and 82.1% at 37 and 52 BV, respectively. eluant solution at 2 BV/h. The compositions of the
Copper elution was also somewhat reduced. These eluants used were:
results are comparable to those obtained with the
sodium sulfite eluant containing ATS (Table 2). Again, (A) 0.25 M Na2S4O6, 0.5 M ATS and 0.25 M Na2S2O3
the presence of thiosulfate may also be responsible for (pH 9.0);
the low elution efficiencies. In further experiments, the (B) 2 M Na2SO3 and 1 M NH3 (pH 11.0); and
concentration of sodium chloride was increased to 4 M (C) 4 M NaCl and 0.1 M ATS (pH 9.2).
to compensate the negative effect of thiosulfate. Am-
monia was no longer necessary because ATS was used. Two different resins were tested for each of the
As can be seen from Table 3, with the eluant contain- eluants. After elution with Eluant A, the column,
ing 4 M NaCl and 0.1 M ATS, copper was completely being saturated with tetrathionate (or trithionate as it
removed at 30 BV and gold at 42 BV. might appear in the solution), was reconditioned
before it was reused for loading in the next cycle.
3.4. Loading/elution cycling This was done by washing the column with 10 BV
0.05 M NaOH at 5 BV/h, followed by a wash with
In practical operations, it is required that the col- deionized water. In the case of Eluant B and Eluant C,
umns be used repeatedly. Therefore the regeneration however, no column reconditioning was necessary.
behavior of different resins was tested by repeating the The concentrations of Au and Cu in the initial
loading and elution for five cycles. In each loading loading solutions, loading effluents and eluates were
cycle, a total volume of 1 L (333 BV) standard loading analyzed so that the loadings of the metals and the
solution, composed of 0.1 M ATS, 20 ppm Au and 500 elution efficiencies could be determined for the indi-
ppm Cu, was passed through the column at 14– 16 BV/ vidual cycles. At the end of the experiments, the resins
h. The loaded column, after washing with 3– 5 BV
deionized water, was then eluted with 40 –50 BV of an
Results of loading/elution cycling experiments from Dowex 21K
Table 3 resin (2.00 g) using Eluant B
Elution of Au and Cu from Dowex 21K resin with the sodium Cycle Metals loaded Metals eluted Recovery
chloride eluant added with ammonia and ATS Au/mg Cu/mg Au/mg Cu/mg Au (%) Cu (%)
Eluant composition Total Au elution Cu elution
1 18.80 21.00 18.81 20.00 100.0 95.2
BV (%) (%)
2 18.56 18.26 18.48 19.80 99.6 108.5
2 M NaCl, 1 M NH3 27 16.4 99.9 3 18.50 13.56 18.36 16.76 99.3 123.7
2 M NaCl, 1 M NH3, 37 71.8 95.7 4 18.46 13.50 18.55 17.02 100.5 126.1
0.1 M ATS 52 82.1 95.8 5 18.60 20.36 18.39 17.58 98.9 86.4
4 M NaCl, 30 94.5 99.6 Overall 92.90 86.66 92.58 91.16 100.0a 100.0a
0.1 M ATS 42 99.5 100.0 a
Total metals remained on resin: Au 0.01 mg and Cu 0.02 mg.
H. Zhang, D.B. Dreisinger / Hydrometallurgy 72 (2004) 225–234 233
were also assayed for the remaining metals and the is probably one of the disadvantages of Eluant A. The
overall elution efficiency for a particular metal was same problem is unlikely to occur with Eluant B and
calculated based on the total amount of the metal Eluant C because no column reconditioning is neces-
eluted and remained. sary. Even if incomplete elution happened, the
Some of the results from the cycling experiments remaining gold would not be expected to precipitate
are presented in Tables 4 – 6. As can be seen, regard- and therefore could be eluted in the following cycles.
less of the resins, gold loading was 9– 10 kg/t, which
is closely consistent with the results obtained from the
single-run experiments. Gold elution efficiency for 4. Conclusions
individual cycles was close to 100% based on the
solution analysis, which agrees well with the overall In the absence of copper, gold can be loaded onto
elution efficiency. Resin assay showed that almost no strongly basic ion exchange resin from thiosulfate
gold was left on the resin after the final elution, solutions rapidly and to high loading concentrations.
particularly when Eluant B and Eluant C were used. In the presence of copper, effective ion exchange
Similar results were obtained for the loading and operation can only be performed under limited con-
elution of copper. There are large errors in copper ditions due to the instability of the thiosulfate solution
loadings and elution efficiencies determined from and the possible formation of poisoning polythionates.
solution analysis, as pointed out in Section 2. How- For a typical solution of 0.1 M ATS with 500 ppm
ever, the loading values estimated from elution were copper, the optimum operation pH is about 11 and
around 9 kg/t. This is, again, consistent with previous there is not much allowance for changes. Under such
results under the same conditions. The overall elution conditions, the adsorption of gold onto the resins is
efficiencies indicate that Cu elution was always com- much stronger than copper. With a large excess of
plete at the end of the experiments. For all the resins, copper in the loading solution, the resin column is
almost identical results were obtained for each indi- usually first loaded with copper, some of which is
vidual cycle and no appreciable deterioration in load- then gradually replaced by gold.
ing and elution behavior was observed. Effective elution of the loaded gold and copper can
It is perhaps worth noting that, with Eluant A be achieved using a number of composite eluant
(Table 4), the amount of gold finally remaining on solutions. The eluant containing tetrathionate is dis-
the resin was slightly higher than with the other two advantageous due to its instability and the need of
eluants. This is likely to result from Cycle 1 during resin reconditioning before returning the resin to
which gold was incompletely eluted for some reason. loading. Sodium sulfite is the best eluant in terms of
The remaining gold might have precipitated when the efficiency, while sodium chloride is likely the most
column was treated with NaOH solution. The precip- economical. The addition of thiosulfate reduces the
itated gold was then locked in the resin and could not elution efficiency of the sulfite eluant. The mechanism
be removed in the following column operations. This of thiosulfate inhibition of elution requires further
The loading/elution cycling experiments demon-
Results of loading/elution cycling experiments from Amberlite IRA-
strate that the resin can be used repeatedly with-
410 resin (2.00 g) using Eluant C out deterioration in the resin loading or elution
Cycle Metals loaded Metals eluted Recovery
Au/mg Cu/mg Au/mg Cu/mg Au (%) Cu (%)
1 18.56 17.00 19.22 22.80 103.6 134.1 Acknowledgements
2 19.08 21.56 19.31 20.80 101.2 96.5
3 19.58 21.80 19.17 21.00 98.0 96.3
4 19.58 29.16 19.39 20.20 99.1 69.3 The authors wish to acknowledge the financial
5 20.04 26.10 19.65 20.80 98.1 79.7 support of Anglogold, Barrick Gold, Kinross Gold,
Overall 96.82 115.6 96.75 105.6 99.9a 100.0a Newcrest Mining, Normandy, Placer Dome and Teck
Total metals remained on resin: Au 0.06 mg and Cu 0.02 mg. Cominco for this work.
234 H. Zhang, D.B. Dreisinger / Hydrometallurgy 72 (2004) 225–234
References fate solutions and pulps with ion-exchange resins. In: Young, C.,
Twidwell, L.G., Anderson, C.G. (Eds.), Cyanide: Social, Indus-
trial and Economic Aspects. TMS, Warrendale, Pennsylvania,
Abbruzzese, C., Fornari, P., Massidda, R., Veglio, F., Ubaldini, S.,
pp. 469 – 483.
1995. Thiosulfate leaching for gold hydrometallurgy. Hydrome-
Perez, A.E., Galaviz, H.D., 1987. Method for recovery of precious
tallurgy 39, 265 – 276.
metals from difficult ores with copper – ammonium thiosulfate.
Berezosky, R.M.G.S., Sefton, V.B., 1979. Recovery of gold and
United States Patent, US 4,654,078, Mar. 31.
silver from oxidation leach residues by ammoniacal thiosulfate
Ritchie, I.M., Nicol, M.J., Staunton, W.P., 2001. Are there realistic
leaching. 108th AIME Annual Meeting, New Orleans, Feb.
alternatives to cyanide as a lixiviant for gold at the present time?
18 – 22.
In: Young, C., Twidwell, L.G., Anderson, C.G. (Eds.), Cyanide:
Fleming, C., Wells, J., Thomas, K.G., 2002. Process for recovering
Social, Industrial and Economic Aspects. TMS, Warrendale,
gold from thiosulfate leach solutions and slurries with ion ex-
Pennsylvania, pp. 427 – 440.
change resin. United States Patent, US 6,344,068 B1, Feb. 5.
Tao, J., Jin, C., Shi, X., 1993. Electrochemistry and mechanism
Gallagher, N.P., Hendrix, J.L., Milosavljevic, E.B., Nelson, J.H.,
of leaching gold with ammoniacal thiosulphate. XVIII Inter-
Solujic, L., 1990. Affinity of activated carbon towards some
national Mineral Processing Congress, Sydney, 23 – 28 May.
gold(I) complexes. Hydrometallurgy 25, 305 – 316.
AusIMM, pp. 1141 – 1146.
Guerra, E., 1997. A study of the factors affecting copper cementa-
Zhang, H., Dreisinger, D.B., 2002a. The adsorption of gold and
tion of gold from ammoniacal thiosulfate solutions. Masters
copper onto ion exchange resins from ammoniacal thiosulfate
thesis, The University of British Columbia.
solutions. Hydrometallurgy 66, 67 – 76.
Li, J., Miller, J.D., Wan, R.Y., 1996. Important solution chemistry
Zhang, H., Dreisinger, D.B., 2002b. The kinetics for the decompo-
factors that influence the copper-catalyzed ammonium thiosul-
sition of tetrathionate in alkaline solutions. Hydrometallurgy 66,
fate leaching of gold. 125th SME Annual Meeting, Phoenix,
59 – 65.
AZ, March 11 – 14.
Zipperian, D., Raghavan, S., 1988. Gold and silver extraction by
Lyons, D., Nickless, G., 1968. In: Nickless, G. (Ed.), Inorganic
ammoniacal thiosulfate leaching from a rhyolite ore. Hydrome-
Sulfur Chemistry. Elsevier, Amsterdam, pp. 526 – 529.
tallurgy 19, 361 – 375.
Merritt, R.C., 1971. The Extractive Metallurgy of Uranium. Colo-
Zur, C., Ariel, M., 1982. Selective recovery of gold from spent gold
rado School of Mines Research Institute.
sulphite solutions by electrodeposition on graphite cloth electro-
Nugent, E.A., 1956. The chemistry of the poisons associated with
des. J. Appl. Electrochem. 12, 231 – 234.
the ion exchange process. S. Afr. Ind. Chem. 10, 282 – 290.
O’Malley, G.P., Nicol, M.J., 2001. Recovery of gold from thiosul-