Handling As in Cu Processing

Overview

The Industrial Separation of Copper

and Arsenic as Sulfides
P.J. Gabb and A.L. Davies

At the Kennecott Utah copper smelter in filter. When the solids

Magna, Utah, bleed streams from the refin- density is maximized, it
ery tankhouse and precious-metals plant are is possible to produce a
combined with smelter weak acid and elec- cake thickness of up to
trostatic precipitator dust to produce leach 50 mm. Despite the co-
solutions containing copper and impurities. herence of the cake, the
Copper and arsenic are precipitated from the moisture content is rela-
solutions as sulfides in a two-stage continu- tively high by normal
ous process that enables excess arsenic to be standards at levels be-
removed from the circuit and routed to haz- tween 50 percent and 60
ardous waste disposal as a highly concen- percent.
trated material. The arsenic stage is
equipped with a Pneuma-
INTRODUCTION
press™ pressure filter for
In the Kennecott smelter hydrometal- dewatering arsenic/cad-
lurgical impurities process, which has mium sulfides. It is criti-
been described in a previous publica- cal in this application that
tion,1 copper and arsenic are precipi- the cake does not drain
tated from acid leach solutions as sul- water on standing to meet
fides in a two-stage process that enables the criterion for hazard-
excess arsenic to be removed from the ous-material disposal. Figure 1. A process flow diagram of the copper/arsenic sepa-
circuit and routed to hazardous-waste From an economic stand- ration stages.
disposal as a highly concentrated mate- point, the cake moisture
rial. The process is conducted continu- should also be low to minimize disposal a theoretical viewpoint, the reactions can
ously under normal temperature and costs. The pressure filter meets these be expected to occur sequentially, and
pressure and avoids the arsenic stabili- criteria and provides unattended auto- complete copper precipitation can be
zation issues of large-volume chemical matic operation. achieved before the onset of arsenic
neutralization products. In practice, the precipitation.
THEORETICAL ANALYSIS
separation of copper from arsenic in MTDATA, a software package for
the first copper precipitation stage is in- Copper, dissolved in an impure acidic metallurgical thermochemistry,2 was
complete. aqueous solution, can be precipitated as used to model the precipitation process.
A schematic of the copper/arsenic an insoluble sulfide by reaction with a MTDATA is a rigorous, robust, and reli-
section of the Kennecott flow sheet is sulfidizing agent, such as NaHS or H2S, able implementation of a free-energy
shown on Figure 1. Each stage consists as shown by the following reactions: minimization algorithm that is capable
of two leach reactors with an associated of computing the amounts and compo-
CuSO4(aq) + H2 S
thickener and filter. The leach reactors sitions of co-existing phases in a system
= CuS + H2SO4(aq) (1)
are equipped with the Tetra Technolo- of given bulk composition and under
gies high-density solids system for in- given physical conditions. The thermo-
2CuSO4(aq) + 2NaHS
creasing sulfide-solids density. It is well dynamic data used for aqueous solu-
= 2CuS + H2 SO4(aq) + Na2 SO4(aq) (2)
known that metal sulfides have the ten- tions and the various solid phases have
dency to form slimes rather than coher- Unfortunately, the analogous reactions been critically assessed by the U.K. Na-
ent cakes when filtered. The benefit of tional Physics Laboratory for consistency
2HAsO2(aq) + 3H2 S = As2S3 + 4H2 O (3)
the high-density solids system for sul- with key thermochemical and phase-
fide-precipitation systems is the ability equilibria data in the literature.
4HAsO2(aq) + 6NaHS + 3H2SO4(aq)
to produce washable cakes at reasonable Figure 2a shows the fraction of copper
= 2As2S 3 + 3Na2SO4 (aq) + 8H2O (4)
cake thickness. and arsenic remaining in solution as a
The copper-precipitation stage is could lead to contamination of the pre- function of the quantity of H2S added.
equipped with an EIMCO vacuum belt cipitate with arsenic-bearing solids. From The stoichiometry of Reaction 1 implies
that one mole of H2S will precipitate one
Table I. Calculated Solution Composition as a Function of H2 S Addition mole of copper, and the abscissa scale is
H2S pH Eh / V Phases Present Cu As shown as a fraction of this stoichiometric
0.000 0.60 0.398 Aqueous 1.000 1.000 requirement. The significant features of
0.999 –0.35 0.385 Aqueous, CuS 0.001 1.000 the figure are that copper precipitation
1.001 –0.35 0.365 Aqueous, CuS, As2S3 0.000 0.997 proceeds according to Reaction 1, and
1.360 –0.34 0.356 Aqueous, CuS, As2S3 0.000 0.040 arsenic, from Reaction 3, does not occur
1.376 –0.34 0.354 Aqueous, CuS, As2S3, S 0.000 0.016 until copper precipitation is complete.
1.500 –0.33 0.353 Aqueous, CuS, As2S3, S 0.000 0.015
The pH of the solution falls throughout

18 JOM • September 1999

copper precipitation, but is not affected
by arsenic precipitation, as suggested by
the production of H2SO4 in Reaction 1
and not Reaction 3. The Eh of the solu-
tion shows a marked step change at the
completion of copper precipitation. A
small amount of arsenic remains in solu-
tion. The data from which the figure was
drawn are given in Table I.
The theoretical results can be com-
pared with the laboratory test work per-
formed during the process-engineering
design stage (Figure 2b). The test work
did not show a particularly clean separa-
tion of copper and arsenic. It has been
suggested that the difficulty of distribut-
ing H2S and avoiding a localized reac-
tion is especially acute in the benchscale
apparatus. The concurrent development
work on a mini-pilot plant indicated a
more favorable separation, implying
better mixing is achieved with increas-
ing scale and continuity of operation.
Further calculations have been per-
formed to investigate the effect of the
presence of other elements in solution,
such as iron and cadmium. These calcu-
lations show that iron is reduced from
ferric to ferrous forms, with a conse-
quent increase in reagent demand be-
fore the complete precipitation of cop-
per, but with no effect on the purity of
the precipitated solids. Cadmium pre-
cipitates with arsenic.
Temperature variation does not have
a significant effect within the operating
range 40–60°C. Initial acid concentra-
tion does not have a great effect on the
separation, and the generation of H2SO4
by copper precipitation rapidly lowers
the pH below zero in any case. The use of
NaHS, rather than H2S, as the sulfidizing
agent does not affect the process, al-
though the pH change is less marked.
PROCESS OPERATION
The two-stage operation provides the
capability of selectively separating cop-
1.0
0.8 Arsenic
Fraction in Solution
Copper
0.6
0.4 Eh/V
0.2
0.0
–0.2
–0.4
0.00 0.25 0.50 0.75 1.00
a
Fraction H2S Added
2,500
Concentration (mg/l)
2,000
1,500
1,000
Copper
500 Arsenic
0
5 6 7 8 9 10
Sample Number
b
Figure 2. The (a) calculated solution compo-
sition and (b) experimental solution composi-
tion as a function of H2S addition.
1999 September • JOM
Table II. Commercial Plant Solution Compositions
Flow
(l/min.) (g/l)
Influent 1 409 10.7
Influent 2 49 10.1
Influent 3 261 4.5
Total In 719 — 362
Effluent 832 0.01
Precipitation — —
per from arsenic and cadmium. The ra-
tio of the copper:arsenic split is deter-
mined by the desired level of arsenic in
anode copper to effect the impurities
control strategy at the refinery. Kennecott
has established that controlled levels of
arsenic and lead are required in anode
copper to ensure bismuth and antimony
are deported to anode slimes almost in
their entirety.
Thus, when arsenic is low in concen-
trate to the smelter, the copper precipita-
tion stage is operated at an Eh around
+40 mV, referred to the calomel elec-
trode, to essentially recycle all of the
copper and arsenic to the flash smelting
furnace. A small portion of the arsenic
passes to the second precipitation stage
to be precipitated with most of the cad-
mium. The arsenic bleed from the pro-
cess in this condition is almost wholly
through flash-furnace slag in a benign
form that complies with environmental
standards.
When the concentrate to the smelter
contains high levels of arsenic, the cop-
per precipitation stage is operated at an
ORP around +200 mV to perform a selec-
tive precipitation. From the above theo-
retical discussion, it is indicated that a
clean split is possible; actual operation at
Kennecott has confirmed the ability to
recover a high proportion of the copper
with about 30 percent of the arsenic dis-
tribution. Typical results from one pe-
riod of low-throughput operation are
shown in Table II.
The reason for the relatively poor sepa-
ration factor between practice and theory
is probably due to the difficulty in add-
ing NaHS in a uniform manner through-
out the liquid hold-up in the reactors.
Any arsenic that is precipitated in lo-
cally overreduced, or low copper tenor,
pH conditions will report with the final fil-
ter cake.
1.25 1.50 It is not critical under normal operat-
ing conditions at the Kennecott smelter
that a high separation factor is achieved
between copper and arsenic, as the an-
ode composition normally dictates a high
portion of arsenic recycle. However,
when new arsenic to the smelter is high,
larger quantities of smelter dust need to
be processed to reject additional arsenic;
11 12 13 this carries a penalty of precipitating
copper units with expensive reagents.
Therefore, it becomes important to maxi-
mize the first pass separation of arsenic
from copper to minimize reagent con-
sumption.
Copper Arsenic
(kg/h) (g/l) (kg/h)
262 5.88 144
30 6.80 20
70 0.13 2
— 166
0.5 2.23 111
99.9% — 33.0%
CONCLUSIONS
The degree of copper:arsenic separa-
tion at the Kennecott smelter will de-
pend on future levels of impurities in the
Bingham Canyon mine. As the mine
progresses into the surrounding halo,
more impurities are likely, making it
more important that precipitation stages
are tuned to their ultimate capability
and the cost of reagents is minimized.
Kennecott has developed the ability to
achieve almost complete selective sul-
fide precipitation of copper from arsenic,
and this methodology will be introduced
as the situation demands.
One cost-reduction approach is to re-
place the use of expensive NaHS with
site-generated hydrogen sulfide gas. This
can be accomplished by biological or
chemical means that are not dependent
on the high and variable cost of caustic
soda used in the manufacture of NaHS.
A low-inertia system is important to
enable rapid interruption in the supply
of this hazardous chemical. The cost of
sulfide reagent would be less than half
the current cost by this means.
References
1. P.J. Gabb et al., “The Kennecott Smelter Hydrometallurgi-
cal Impurities Process,” Proceedings of Copper ’95–Cobre ‘95
International Conference, vol. 3, ed. W.C. Cooper et al.
(Montreal, Canada: The Metallurgical Society of CIM, 1995),
pp. 591–606.
2. R.H. Davies et al., “MTDATA—The NPL Databank for
Metallurgical Thermochemistry,” User Aspects of Phase Dia-
grams: Proceedings of the International Conference, ed F.H. Hayes
(London: Institute of Metals, 1990), pp. 140–152.
P.J. Gabb and A.L. Davies are with Rio Tinto Technical
Services.
For more information, contact P.J. Gabb, Rio
Tinto Technical Services, P.O. Box 50, Cas-
tlemead, Lower Castle Street, Bristol BS99 7YR,
United Kingdom; telephone 44-117-927-6407; fax
44-117-927-3317; e-mail phil.gabb@ riotinto.com.
This month’s JOM-e
features the second
part of the series
Modeling and
Simulation.
The first part
appeared in August.
The articles can be
accessed at www.tms.org/
pubs/journals/JOM/
articles.html
19