AmericanMineralogist, Volume62,pages377-381, 1977
ImprovedheavyJiquidseparationat fine particle sizes
Knrru J. HnNr-ev
The Australian Mineral DeuelopmentLaboratories(AMDEL)
FlemingtonStreet, Frewuille,Adelaide,South Australia 5063
Abstract
A method is described for effective heavy-liquid separation of material in the particle size
range 50 to l0 pm. The method involves centrilugal separationof the material in a necked
glasstube, in which a densitygradient column spanningthe specificgravity range ofthe light
minerals is positioned above the constriction, the specific gravity of the heavy liquid from the
base of the density gradient column to the bottom of the tube being the desiredspecificgravity
of separation.When the sample is stirred into the gradient and centrifuged,the light mineral
particlesspreadout looselythrough the densitygradient and allow the heavy particlesto pass
through into the lower compartment of the tube. This method has been tested with various
heavy liquids and mineral mixtures,and has beenshown to be consistentlymore effectivethan
the normal method using a liquid of uniform specificgravity, in which heavy particlestend to
be entrained in the compact mat of light particleswhich risesto the top of the liquid during
centrifueation.
Introduction
of valuable heavy minerals for microscopic study,
Heavy-liquidseparation
is an importanttool of the and for determining the liberation characteristics of
mineralogist
in the concentration
of specificminerals the valuable minerals (Muller et al., 1969; Henley et
or groupsof minerals,and facilitiesfor heavy-liquid al.,1973).In addition, pure samplesof specificminerseparationare alrnostubiquitousin geologicaland als may need to be preparedfor analysisto determine
mineralogical
laboratories.
The mostcommonlyused how closely they meet commercial specifications.In
liquidsare bromoform(sp.gr.2.89),tetrabromoeth- contrast to much of the normal geologicaland miner(sp.gr.3.32)and alogical heavy-liquid separation work, ore-dressing
ane (sp.gr.2.96),di-iodomethane
Clerici solution (sp.gr.up to 4.3 at room temper- mineralogicalevaluation commonly requiresseparaature).The first three(organic)liquidscan be diluted tion of finely-groundmaterial, much of which may lie
with acetone,alcohol, or NN-dimethylformamide, within the size range 50 to l0 pm. Heavy-liquid sepaamongothers,whereas
Clericisolutionis dilutedwith ration within this size range requires the use of a
distilled water. Separationof grains coarserthan centrifuge, and it is the purpose of this note to deabout 50 pm is normallydone staticallyin a separa- scribean improved separationprocedurefor this type
tion funnelor similardevice,whereasfiner-sized
ma- of material.
terial is normallyseparated
centrifugally,usingone
The improved method
of a numberof separationtubesor sleeves.
With the advent of the electron-probemicroimproved
method was developedas a result of
The
analyzerthe need for separatingpure mineral sam- problemsencounteredin the centrifugalseparationof
plesfor chemicalanalysishasdecreased
considerably. particlesin the size range 50 to l0 pm using a heavy
However,suchsamplesarestill requiredfor geochro- liquid of uniform density. Our normal centrifugal
nology studies,and in beach-sand
explorationthe heavy-liquidseparationprocedureinvolvesthe use of
proportionof 'heavyminerals'is the first parameter a simple necked glass tube (Fig. l) of about 130 ml
determinedon drill-core material.In ore-dressing capacity in which the angle of necking is very shalmineralogicalstudies,heavy-liquidseparationstill low. After centrifugationa conical plunger is inserted
forms an integralpart of the assessment
procedure, gently through the 'floats' to seal off the neck, at
both with regardto concentrating
smallproportions which stage'floats'and 'sinks' can be recoveredsepa377
HENLEY' HEAVY-LIQUID SEPARATION
liquid of uniform density. The gradient is chosento
range in specificgravity from that at which the actual
separationis required (e.g., 3.3) to a value which is
slightly less than the majority of the light minerals
present (e.g., 2.4, if quartz, feldspar, and biotite are
the major minerals present).On centrifugation,each
of the light minerals positions itself at a level corresponding to its density in the gradient, and the mass
of light minerals is thus distributed through a significant proportion of the volume of the gradient (Fig.
2). Where a mineral of uniform density (e.g., qtartz)
is present,it will form a discrete,compact layer which
will tend to entrain heavy particles, but for many
minerals slight variations in density will tend to
spread out the layers vertically and allow the heavy
'sinks.'
m i n e r a l g r a i n st o p a s st h r o u g h i n t o t h e
Incorporation of a density gradient column into
the separationtube is a relatively simple matter. The
tube is half-filled with denseliquid and the light liq-
2.4Fig.I Neckedglasstube
"::":H,r;."t
heavyliquid
centrifugal
rately. This procedure has been found to be simpler
to use than any of the multiple-tube centrifuge systems, and can be used easily with all heavy liquids,
although care has to be taken with high-densityClerici solution becauseof the danger of shattering the
glass if the centrifugeis run at high speed.However,
despite its simplicity, repeated separations are required to give good recovery of heavy minerals into
'sinks,'
becauseof entrainment of a significant proportion of the heavy minerals in the predominant
light fraction. The entrainment problem increasesas
the particle size decreases(and the number of particles per unit weight increases),and is particularly
acute in the -20*10 pm size range (10 pm is the
approximate practical lower sizelimit of heavy liquid
separation;below this size entrainment and flocculation prevent effectiveseparation).
The new method involves incorporating a suitable
density-gradientcolumn in the centrifugetube above
the neck, to spreadout the light minerals and prevent
the formation of a compact mat of light mineral
particles such as is obtained when using a heavy
_fvrligr-ecline
2.6s3.3-
-Atbite
- Quqrtz
3.3-
t
Sp.gr
t
Minerol
Fig. 2. Density gradient separation of C3 fraction of granitelgalena mixture. Note the spreading of the light minerals
through the gradient, thereby reducingentrainmentof the dense
galena.
HENLEY HEAVY-LIQUID SEPARATION
379
uid is carefully poured in on top. Rotation of the tube
Table2. Percentageofgalena in size fraction reporting into
'floats'*
with a swinging motion of the hand will causeswirling in the tube which will tend to homogenizethe two
liquids at their junction and thereby createa density
(a)
SeDaration in licuid
of mifom
densitv
gradient. This procedure is continued until the deNunber of seDaratlm
st
Size
fraction
sired gradient is obtained,as indicated by the vertical
i2
o.2
0.6
0.2
c1
separationin the tube of two glass marker beadsof
1.3
I.0
u.o
c2
5.0
known intermediatedensities.The preciseprofile of
I
.
9
t,7
5
.
2
c3
10.r
42.6
3.5
c4
the gradient is generally not critical, the essential
23.0
18.4
18.4
28.3
c5
requirement being the spreadingapart ofthe various
(b)
gradient
colum
seDaratlon
ln density
light mineralsto prevententrainment of heavyminerNu[ber
of
separation
st'ages
Size
als. The useof the gradient also facilitatesinsertion of
fractionL234
the conical plunger to seal offthe neck of the tube, as
o.2
0.2
0.1
cl
the mineral grains are loosely in suspensionand not
r.3
1.0
0.7
0.3
c2
no
0.9
2.8
c3
present as a compacted mat. The density gradient is
4.7
0.8
c4
30.2
l?
remarkably stable, as once formed and left undisr2.5
44. I
3.3
c5
turbed, molecular diffusion is the only mechanism
*39 init:iaL
reLrtuzre.
uei,ght of granite-galena
tending to causehomogenization.As pointed out by
M u l l e r a n d B u r t o n ( 1 9 6 5 ) ,d e n s i t yg r a d i e n t sm a y r e of granite with about 2 percent galena and 2 percent
main stable from severaldays to severalweeks.
and sizedin a
chalcopyriterespectivelywere-prepared
Comparison of methods
Warman Cyclosizerto give sizefractions in the range
-l0lrm
A comparativestudy was made of the two separa- 5 3 t o l 0 p m ( e q u i v a l e n tq u a r t z s p h e r e ) ,t h e
tion methods (i.e., uniform-density liquid and den- fraction not being retained. A number of tests were
-5 3 + l0 pm fractions
sity-gradientcolumn), using a mixture of pulverized carried out using the combined
granite (composed of quartz, albite, microcline, and and also individual size fractions within this range.
For the testswith uniform-densityliquid, a known
biotite and containing20 ppm Pb and 5 ppm Cu) and
pure galena;subsequentlya briefer seriesof testswas weight of sample was stirred into the liquid and any
carried out using a mixture of the pulverizedgranite entrained air removed by vacuum treatment; the
'floats' reand pure chalcopyrite.The heavy liquid used for the sample was then centrifuged, and the
main tests was di-iodomethane (sp.gr. 3.3); dilution rnouid, washed,dried, and weighed.One-eighthof the
'floats' was riffied out for analysisfor Pb or Cu, the
for the density gradients was with NN-dimethylf o r m a m i d e ( M e y r o w i t z e t a l . , 1 9 6 0 ) . I n a d d i t i o n , remainder was returned to a separatetube, and the
some tests were carried out using tetrabromoethane whole procedurerepeated.In this way representative
'floats' material at each stagewas obtained for analyas the heavy liquid. Tests were also,carried out on
different amounts of sampleto determinethe effectof sis, and by calculating the distribution of Pb or Cu
between'floats' and'sinks' at each separationstage,
this variable on separationefficiency.
The galena and chalcopyrite were hand-picked it was possibleto monitor the progressivechange in
from massiveore, crushed,separatedat specificgrav- separationefficiency.
A similar procedure was adopted for the densityity 3.3 to iemove any possible intergrown silicates,
-53
gradient
tests,exceptthat a measuredweight of matepm. The granite was
and then pulverized to
-53
was
stirred into a density gradient of known
rial
pm. Separatemixtures
separatelypulverized to
characteristics (determined using glass beads of
known specificgravity) rather than a liquid of uniTable l Percentageof galena in total -53 * l 0 pm size fraction of
g r a n i t e - g a l e n am i x t u r e r e p o r t i n g i n t o ' f l o a t s ' f o r v a r i o u s i n i t i a l
form density. The results of the initial comparison
sample weights (density-gradientseparation method)
test using l0g of -53+10 pm sample showed that,
whereasthe uniform-densityliquid gave'floats' prodNurnber of separatlon
Sanp le
stages
ucts containing 27.1 percent of the galena after one
(e)
L2
weieht
stageof separationand 6.8 percent after four stages,
'floats'products from the density8.0
5.0
3.7
the corresponding
t<
4.9
2 . r gradient separationcontained 7.7 and 0.6 percent of
2.8
r.25
r.2
the galena respectively.
L .
)
380
HENLEY: HEAVY-LIQUID SEPARATION
separationtestson the -53+10
Density-gradient
pm sample using progressively
decreasingsample
weight gave the resultsshown in Table l, clearly
the
indicating,as would be expected,that decreasing
sampleweight increasedseparationefficiency.
Comparativeseparationtests on individual Cyclosizer size fractions (Cl to C5') of the granite-galenamixturegavethe result$shownin Table2,
which indicatethat for all sizefractionsthe densitygradientseparationmethod is superiorto the uniform-densityseparationmethod and that there is a
parprogressive
in efficiencywith decreasing
decrease
ticle sizefor both separationmethods.The resultsfor
the C5 fraction are particularlyinteresting;with the
separationmethod,18.4percentof
uniform-density
the galenastill remainsin 'floats' after four stagesof
whereasthe corresponding
figurefor the
separation,
density-gradient
separationmethodis only 1.3 percent.The data for the Cl. C3. and C5 sizefractions
in Figure3.
are illustrateddiagrammatically
Table 3 summarizes
the data for the testson the
granite-chalcopyrite
mixture.It is interesting
to note
that thereis generallylessdifferencebetweenthe two
separation
methodsandthat at mostparticlesizesthe
separationhas been more efficientthan that for the
granite-galena
mixture.I suggest
that
corresponding
theseeffectsare due to the differencein averagepar-
Uniformdensity
D c n s i t yg r a d i c n l
tnfu
o
o
o
ol
,.
L
9r0
o,
c
o-
(,
o
o
N
U'"
;
c
o
ol
o
O n.q
Ol ""
o
c
o
(,
!
*
0.2
1231
Numberof sepqrqtionstqges
Fig. 3. Diagram illustrating the relationship between particle
size, separation efficiency,and number of separation stagesfor the
uniform-density and density-gradient centrifugal separation methods. Note the improved separation efficiency of the densitygradient method as compared to the uniform-density method, and
the increasing efficiency of both methods with increasing particle
size and number of stages of separation. Increased efficiency is
indicated by a decreasein the proportion of galena present in (r'e
e n t r a i n e di n ) ' f l o a t s . ' l
1The particlesizescorrespondingto theseCyclosizerfractions
are approximatelyas follows:
Cyclosizer
Fraction
cl
C2
C3
C4
C5
Granite
(pm)
Galena
-5 3+43
-43+32
-32+23
-23+t5
-15+ll
-53+22
-22+16
-t6+12
-12+8
- 8+6
0rm)
Table3. Percentage
ofchalcopyritein sizefractionreportinginto'floats'*
Size
fractlon
c1
c2
c3
c4
c5
*39 initiaL
Uni f o=rm-dens 1 tv 1 lq uid
Nunber of separation
stages
4
L
1.0
r.6
2.3
5.3
20.9
0.1
0.1
0.1
o.7
L.6
ueight of gz,anite-chalcopyrtte
I.3
1.8
1.3
3.2
5.7
mLrture
<0. 1
0.1
0.1
0.5
0.9
Chalcopyrite
(pm)
- 5 3 + 3I
-31+23
-23+17
-t7+lI
-l l+8
381
HENLEY HEAVY.LIQUID SEPARATION
ticle size of the heavy mineral in the two sets of
experiments;in any Cyclosizerfraction the grains of
galena are significantlysmaller than chalcopyritebecause of the difference in density between the two
minerals, and the efficiencyof separation is greater
the less the difference in size between heavy and light
minerals.
Tests using tetrabromoethane (sp.gr. 2.96) gave
similar results to those obtained using di-iodomethane.
Tests on the - l0*5 pm (equivalentquartz sphere)
fraction of the granite-chalcopyritemixture showed
that both methods of separation were equally ineffective---cven after five stages of separation the
'floats'
still contained about 25 percent of the chalc o p y r i t e i n t h e s i z ef r a c t i o n .
Conclusions
Overall, the resultsof the testsindicatethat: (a) the
density-gradientseparationmethod is more efficient
than the uniform-density separation method at all
particle sizes down to l0 pm; (b) the differencein
efficiency between the two methods increases with
increasingdensity of the heavy fraction; (c) the efficiency of both methods decreaseswith decreasing
particle size and with increasingsample weight; and
(d) both methods are relatively ineffectiveat particle
sizesbelow l0 pm.
Acknowledgments
I thank Manfred Imiela for carrying out most of theexperiment a l w o r k a n d t h e D i r e c { o r o f A m d e l f o r p e r m i s s i o nt o p u b l i s h t h i s
note.
References
Henley, K. J., R. S Cooper, G. G. Lowder, F. Radke and I G.
W a t m u f f ( 1 9 7 3 ) T h e q u a n t i t a t i v e m i n e r a l o g i c a le v a l u a t i o n o f
nickel ores and their processingproducts. ln, WesternAustralian
C o n l e r e n c e1 9 7 J ,A u s t r a l a s . I n s t . M i n . M e t a l l . , 4 0 1 - 4 1 5 .
Meyrowitz, R, F. Cuttitta and B. Levin (1960) N,N-dimethylformamide, a new diluent for methyleneiodide heavy liquid
Am. Mineral . 45. 1278-1280
Muller, L. D. and C. J Burton (1965) The heavy liquid density
gradient and its applications in ore-dressing mineralogy. Proceedings oJ the 8th Commonwealth Mining and Metallurgical
C o n g r e . s .6s,, I l 5 l - l 1 6 3
M u l l e r , L D . , K J . H e n l e y a n d R . E . K B e n j a m i n ( 1 9 6 9 )A p p l i e d
mineralogy in tin ore beneficiation. ln, A Second Techincal Conference on Tin W. Fox, Ed. International Tin Council, 2,
559-600.
Manuscript receiued,May 28, 1976: accepted
for publit'ation, September 17, 1976