XXXI International Mineral Processing Congress 2024 Proceedings/Washington, DC/Sep 29–Oct 3 1227
For TREO+Y, WHIMS efficiency reached as high as
93% recovery at 75 µm, with grade peaking at 1.22%.
Recovery in the size range of 38 to 212 µm is above 90%,
indicating the optimal size range for high recovery. Recovery
fell away sharply below 20 µm. Above 212 µm recovery also
falls away to as low as 73.7% but averages 86%, indicating
finer allanite particles encapsulated within coarser silicate
particles, reducing the composite particle’s overall magnetic
susceptibility, but recovery is still considered quite good
relative to the 90 to 92% seen in batch tests. Losses in the
minus 20 µm fraction are generally responsible for lower-
ing the overall TREO+Y recovery (minus 20 µm TREO+Y
recovery is only 41.1%).
Iron recovery response generally followed the shape of
the rare earths response curve, likely due to the amphibole
mineral hastingsite having the same order of magnitude of
magnetic susceptibility and degree of liberation as allanite.
There is also iron in allanite which influences the response.
Silicon is present as silica, feldspars, amphiboles and
allanite in the ore, hence tracking silica rejection is not
overly meaningful. Recovery generally improves with
increasing fineness, though this likely indicates the tran-
sition away from silicate gangue minerals at coarser sizes,
which are efficiently rejected, to solid solution silica in alla-
nite and hastingsite which are recovered due to the miner-
als’ paramagnetic behaviour.
Al2O3 is usually a good indicator of feldspar minerals,
though it is also present in hastingsite and allanite. Low
recoveries across the size range can be seen, indicating the
success of WHIMS in rejecting aluminium bearing non-
magnetic minerals. Rejection of aluminium is important as
it can consume reagents in refinery leaching, depending on
the process adopted.
Products from the bulk primary WHIMS test were
subjected to quantitative XRD analysis to allow a mass
balance across the circuit to be undertaken by mineral
type. Hastingsite makes up a calculated 11.2% of feed but
is the dominant mineral in the four magnetics products,
particularly in the 3000 to 1000 gauss fractions. Allanite
was not identified individually in the analysis. A search of
Raman spectra for allanite and hastingsite shows the peaks
are very close together and allanite may have been masked
by hastingsite given the tenfold higher concentration (see
Figure 6).
Figure 7 provides a plot of cumulative magnetics mass
yield versus gangue mineral recovery. Although allanite was
not identified by XRD, TREO+Y recovery is predomi-
nantly represented by allanite so recovery by stage can be
inferred for the comparison.
Microcline (K-Al felspar) and Albite (Na feldspar)
demonstrated increasing grades with increasing field
strength, but overall recovery to magnetics was low at 15.6
and 12.3% respectively. It is likely that their recovery to
magnetics was due to incomplete liberation from the para-
magnetic minerals.
Quartz grades were generally low, suggesting that the
degree of liberation was high.
Hastingsite and inferred allanite recovery responses
have similar gradients, supporting the premise that they
have similar magnetic susceptibilities since the two miner-
als are large liberated from each other. Hastingsite’s ulti-
mate recovery was 85%, compared to allanite at 72.3%.
Due to the high concentration of hastingsite in ore feed, its
content in the collective concentrate was 40.4%, compared
with an estimated 1.6% allanite based on Ce content (Ce
makes up 27% of total allanite mass).
Secondary WHIMS
Primary WHIMS magnetics was subjected to cleaner
WHIMS testing to determine if regrinding and cleaning
would improve the rare earths content. A range of P80 sizes
from 106 to 38 µm was tested. A grind P80 size of 106 µm
was selected from these batch tests as optimal, with 76.9%
magnetics mass yield and 96.7% TREO+Y recovery. Finer
regrind sizes rejected only small amounts of incremental but
TREO+Y recovery suffered mass (69.3% mass yield at 38
µm for 90.4% TREO+Y recovery). Fe recovery remained
high across all grind sizes, hence, cleaner WHIMS served
to further enrich hastingsite. Further machine optimisation
ensued and the bulk cleaner run undertaken, yielding (rela-
tive to new feed) a combined magnetics of 16.1% mass at
1.51% TREO+Y and 24% Fe, for recoveries of 69.2% and
78.4% respectively.
QEMSCAN mineral abundance determination on the
secondary WHIMS magnetics is presented as Figure 8.
Allanite was the dominant REE host (mainly Ce, La
and Nd), accounting for 90% of each of these elements.
This deportment data has been calculated assigning 10%
Ce, 5% La and 4.5% Nd to allanite, with these values cal-
culated by off-line spectral analysis of the QEMSCAN data.
Other, far less significant REE hosts include:
• A Ca-Fe-Ti-REE-silicate, possibly chevkinite, which
typically accounts for 2% of the Ce, La and Nd
• Synchysite and/or parisite which account for a fur-
ther 2% of the Ce, La and Nd
• REE-bearing apatite which accounts for about 1% of
the Ce, La and Nd.
For TREO+Y, WHIMS efficiency reached as high as
93% recovery at 75 µm, with grade peaking at 1.22%.
Recovery in the size range of 38 to 212 µm is above 90%,
indicating the optimal size range for high recovery. Recovery
fell away sharply below 20 µm. Above 212 µm recovery also
falls away to as low as 73.7% but averages 86%, indicating
finer allanite particles encapsulated within coarser silicate
particles, reducing the composite particle’s overall magnetic
susceptibility, but recovery is still considered quite good
relative to the 90 to 92% seen in batch tests. Losses in the
minus 20 µm fraction are generally responsible for lower-
ing the overall TREO+Y recovery (minus 20 µm TREO+Y
recovery is only 41.1%).
Iron recovery response generally followed the shape of
the rare earths response curve, likely due to the amphibole
mineral hastingsite having the same order of magnitude of
magnetic susceptibility and degree of liberation as allanite.
There is also iron in allanite which influences the response.
Silicon is present as silica, feldspars, amphiboles and
allanite in the ore, hence tracking silica rejection is not
overly meaningful. Recovery generally improves with
increasing fineness, though this likely indicates the tran-
sition away from silicate gangue minerals at coarser sizes,
which are efficiently rejected, to solid solution silica in alla-
nite and hastingsite which are recovered due to the miner-
als’ paramagnetic behaviour.
Al2O3 is usually a good indicator of feldspar minerals,
though it is also present in hastingsite and allanite. Low
recoveries across the size range can be seen, indicating the
success of WHIMS in rejecting aluminium bearing non-
magnetic minerals. Rejection of aluminium is important as
it can consume reagents in refinery leaching, depending on
the process adopted.
Products from the bulk primary WHIMS test were
subjected to quantitative XRD analysis to allow a mass
balance across the circuit to be undertaken by mineral
type. Hastingsite makes up a calculated 11.2% of feed but
is the dominant mineral in the four magnetics products,
particularly in the 3000 to 1000 gauss fractions. Allanite
was not identified individually in the analysis. A search of
Raman spectra for allanite and hastingsite shows the peaks
are very close together and allanite may have been masked
by hastingsite given the tenfold higher concentration (see
Figure 6).
Figure 7 provides a plot of cumulative magnetics mass
yield versus gangue mineral recovery. Although allanite was
not identified by XRD, TREO+Y recovery is predomi-
nantly represented by allanite so recovery by stage can be
inferred for the comparison.
Microcline (K-Al felspar) and Albite (Na feldspar)
demonstrated increasing grades with increasing field
strength, but overall recovery to magnetics was low at 15.6
and 12.3% respectively. It is likely that their recovery to
magnetics was due to incomplete liberation from the para-
magnetic minerals.
Quartz grades were generally low, suggesting that the
degree of liberation was high.
Hastingsite and inferred allanite recovery responses
have similar gradients, supporting the premise that they
have similar magnetic susceptibilities since the two miner-
als are large liberated from each other. Hastingsite’s ulti-
mate recovery was 85%, compared to allanite at 72.3%.
Due to the high concentration of hastingsite in ore feed, its
content in the collective concentrate was 40.4%, compared
with an estimated 1.6% allanite based on Ce content (Ce
makes up 27% of total allanite mass).
Secondary WHIMS
Primary WHIMS magnetics was subjected to cleaner
WHIMS testing to determine if regrinding and cleaning
would improve the rare earths content. A range of P80 sizes
from 106 to 38 µm was tested. A grind P80 size of 106 µm
was selected from these batch tests as optimal, with 76.9%
magnetics mass yield and 96.7% TREO+Y recovery. Finer
regrind sizes rejected only small amounts of incremental but
TREO+Y recovery suffered mass (69.3% mass yield at 38
µm for 90.4% TREO+Y recovery). Fe recovery remained
high across all grind sizes, hence, cleaner WHIMS served
to further enrich hastingsite. Further machine optimisation
ensued and the bulk cleaner run undertaken, yielding (rela-
tive to new feed) a combined magnetics of 16.1% mass at
1.51% TREO+Y and 24% Fe, for recoveries of 69.2% and
78.4% respectively.
QEMSCAN mineral abundance determination on the
secondary WHIMS magnetics is presented as Figure 8.
Allanite was the dominant REE host (mainly Ce, La
and Nd), accounting for 90% of each of these elements.
This deportment data has been calculated assigning 10%
Ce, 5% La and 4.5% Nd to allanite, with these values cal-
culated by off-line spectral analysis of the QEMSCAN data.
Other, far less significant REE hosts include:
• A Ca-Fe-Ti-REE-silicate, possibly chevkinite, which
typically accounts for 2% of the Ce, La and Nd
• Synchysite and/or parisite which account for a fur-
ther 2% of the Ce, La and Nd
• REE-bearing apatite which accounts for about 1% of
the Ce, La and Nd.