CHALLENGES IN NIOBIUM FLOTATION

*C.Gibson1,2, M. Aghamirian1, S. Kelebek2

1
SGS CANADA INC.
185 Concession Street
Lakefield, Ontario, Canada K0L 2H0
(*Corresponding author: 5ceg@[Link])
2
Queen’s University
Robert M. Buchan Department of Mining
Goodwin Hall, 25 Union Street,
Kingston, Ontario, Canada, K7L 3N6

ABSTRACT

Global niobium production is currently dominated by three operating mines, Araxá and Catalão in Brazil, and
Niobec in Canada. Although over 90% of the world’s niobium is mined and processed in Brazil, a number of high
grade niobium deposits exist worldwide. The advancement of these deposits depends largely on the development of
operable beneficiation flowsheets. Pyrochlore, the primary niobium bearing mineral, is typically upgraded by
flotation with amine collectors at acidic pH, and often includes a reverse gangue flotation stage prior to pyrochlore
flotation. Challenges related to the flotation of niobium ores include complicated flowsheets which require pulp pH
adjustment from alkaline to highly acidic conditions, difficulty recovering fine grained pyrochlore, producing
concentrates that meet specifications, and minimizing the losses to reverse pre-flotation gangue concentrates, slimes,
and tailings. The specific flotation challenges associated with different niobium ores are discussed. In some cases,
the use of alternate flotation reagents and flowsheets are required to achieve reasonable gangue rejection and
maximize recovery, particularly of very fine pyrochlore grains.

KEYWORDS

Flotation, niobium, pyrochlore, benzohydroxamic acid, columbite

244
 47th Annual Canadian Mineral Processors Operators Conference©, Ottawa, Ontario, January 20-22, 2015

INTRODUCTION

Niobium is a transition metal primarily found in the oxide minerals pyrochlore (Ca,Na)2Nb2O6(OH,F) and
columbite (Fe,Mn)Nb2O6. After physical upgrading, pyrochlore is converted into a ferroniobium product which is
largely used in the production of high strength low-alloy steel. Today’s niobium market is dominated by the
Companhia Brasileira de Metalurgia e Mineração, whose Araxá operation produces over 80% of the world’s
niobium. There are two smaller operations, Niobec in Canada, and Catalão in Brazil, which account for the
remaining production (U.S. Geological Survey, 2013).

A number of niobium deposits exist globally, their advancement dependent on the development of operable
beneficiation flowsheets. Pyrochlore, the predominant niobium bearing mineral, is commonly upgraded by flotation
with cationic amine collectors in acidic pulp conditions. However, many niobium ores do not respond to this typical
flowsheet because they are either highly refractory or niobium deports to other minerals such as columbite, ilmenite
and rutile, which are not as responsive to this reagent scheme (Bulatovic, 2010). Other reagents, such as alkyl
hydroxamic acid, have been studied extensively and while promising, are not without their own inherent issues
including lack of selectivity over iron oxide minerals, and high addition rates (Ni & Liu, 2012; Gorlovskii, Eropkiny,
Koval, Strelstin, Khobotova, & Shtchukina, 1968; Gorlovskii, Khobatova, & Shtchukina, 1966).

It is the aim of this report to highlight the challenges associated with the physical upgrading of niobium
minerals through examination of the flotation behavior of two different ore systems. In order to overcome these
challenges it is important to thoroughly understand the limitations of typical flowsheets/reagent schemes currently
employed in operation.

MINERALOGY

A number of bench scale batch flotation tests were performed on different carbonatite ores. QEMSCAN
anlaysis was performed on a representative head sample of both ores. Ore A is a sövite, containing over 60% calcite
and lesser amounts of dolomite, ankerite, apatite, silicates and iron oxides. Ore B is a ferro-carbonatite, containing
mainly dolomite, ankerite, siderite, barite, fe-oxides, and silicates with lesser quantities of ilmenite, pyrite, and
apatite. Head assays for each ore type can be found in Table 1. Full modal mineralogy of the gangue minerals in each
ore type can be found in Figure 1.

Table 1 - Head assays for two types of niobium bearing carbonatite ores
Assays %
Nb2O5 CaO MgO Fe 2O3 P2O5 TiO2 SiO2 Al2O3 K2O Na 2O LOI

Ore A: Sövite 0.50 43.40 3.79 5.03 3.87 0.26 5.27 0.49 0.17 0.10 34.60

Ore B: Ferro-carbonatite 0.86 19.10 8.36 19.00 1.11 3.22 7.43 1.86 1.31 0.21 25.50

245
 47th Annual Canadian Mineral Processors Operators Conference©, Ottawa, Ontario, January 20-22, 2015

60 Modals: Gangue Minerals

Ore A: Sovite
50 Ore B: Ferro-Carbonatite

Mineral Mass (%)

40

30

20

10

fe-oxides

calcite

dolomite
pyrite

apatite

other
quartz
ilmenite

barite
other silicates

ankerite/siderite
Figure 1 - Gangue Modal mineralogy of two different niobium bearing carbonatite ores

Figure 2 presents a breakdown of the niobium minerals and niobium elemental deportment to different ore
minerals. Niobium in Ore A deports mainly to pyrochlore ~67%, and columbite ~27%, with a minor amount in rutile.
Similarly, in Ore B the majority of Nb reports to pyrochlore (nearly 80%). However, Ore B contains no columbite,
and approximately 20% of the niobium is found in an Nb-rich rutile.

3.0 80
Modals: Nb Minerals Nb Elemental Deportment
Ore A: Sovite 70 Ore A: Sovite
2.5
Ore B: Ferro-Carbonatite Ore B: Ferro-Carbonatite
60
Mineral Mass (%)

2.0
50
Mass (% Nb)

1.5 40

30
1.0
20

0.5
10

0.0 0
pyrochlore

columbite

Nb-rutile

ilmenite

ilmenorutile
pyrochlore

columbite

Nb-rutile

Figure 2 - Niobium mineral modals and elemental deportment for two different carbonatite ores

Niobium mineral liberation data for the two ores can be found in Tables 2 and 3. Liberation is categorized
based on mineral of interest area percent: free (>=95%), and liberated (<95% and >= 80%). Examining Ore A at a
size fraction of -150µm/ +75µm, which is generally an acceptable size range for flotation, niobium minerals are over
56.7% free and liberated, with only 10.5% of Nb minerals locked. By contrast, over a finer size range of -
106µm/+25µm, niobium minerals in Ore B are only 8.8% free and liberated. Only in the -25µm fraction does
niobium mineral liberation of Ore B become comparable to that of Ore A in the -150µm/ +75µm fraction.

246
 47th Annual Canadian Mineral Processors Operators Conference©, Ottawa, Ontario, January 20-22, 2015

Table 2 - Liberation of Nb Minerals in Ore A

ORE A: Sövite Combined +212µm -212/+150µm -150/+75µm -75/+20µm -20µm

Free Nb Minerals, % 34.50 0.08 39.09 26.67 46.78 51.84

Lib Nb Minerals, % 18.79 0.00 12.28 30.06 23.32 25.20
Midds Nb Minerals, % 14.16 32.51 3.73 13.16 9.48 11.41
Sub-Midds Nb Minerals, % 15.48 29.49 19.84 19.65 9.31 5.01
Locked Nb Minerals, % 17.07 37.92 25.05 10.47 11.10 6.54
Total 100.0 100.0 100.0 100.0 100.0 100.0

Table 3 - Liberation of Nb Minerals in Ore B

ORE B: Ferro-Carbonatite Combined +106µm -106/+25µm -25µm

Free Nb Minerals, % 25.91 1.01 4.95 55.27

Lib Nb Minerals, % 3.62 2.09 3.81 4.22
Midds Nb Minerals, % 17.60 8.20 16.96 22.81
Sub-Midds Nb Minerals, % 29.04 52.51 42.45 6.52
Locked Nb Minerals, % 23.82 36.20 31.83 11.18
Total 100.0 100.0 100.0 100.0

FLOTATION

Batch Flotation with Amines

Batch flotation test work revealed that Ore A could be successfully upgraded using a either a two-stage
flotation process, with reverse flotation of carbonate minerals using a fatty acid collector, followed by pyrochlore
flotation with an amine collector or by directly floating pyrochlore with an amine collector after grinding to 100%
passing 105 microns. Amine collectors are currently used to recover pyrochlore at all three niobium flotation
operations worldwide (Guimares & Weiss, 2003; Filho, Riffel, & de Faria Sousa, 2002; Oliveira, Saraiva, &
Oliveira, 2001; Gendron, Biss, & Rodrigue, 1981). Selective flotation of pyrochlore using amines requires pulp pH
to be quite low, reaching approximately pH 2 in the final cleaning stages. Carbonate minerals are considered to be
acid consumers and therefore can be floated out prior to niobium flotation in order obtain and maintain acidic pulp
conditions. Two stage batch flotation of Ore A produced a final concentrate grading 40.6% Nb2O5 with 53.0% Nb
recovery. The direct flotation flowsheet produced a final concentrate with an assay of 51.7% Nb2O5 and 49.0% Nb
recovery.

In both flowsheets, fluorosilicic acid was used to adjust pH. In two-stage flotation, after carbonate flotation
there were often still carbonate minerals remaining in the pulp which consumed acid in niobium rougher flotation.
Thus, one of the main requirements in the niobium rougher circuit was to perform flotation quite quickly before the
pH rose to around 7. In the first few cleaner flotation stages, mainly carbonate minerals were rejected. Once a
sufficient amount carbonate minerals were rejected, pH could be effectively lowered to below 4 and rejection of
silicates occurred. It should be emphasized that silicate rejection to the below the target level (usually around 3.5%
SiO2) is still a serious challenge. Rejection of silicates can be achieved mainly through further cleaning stages and
further pH reduction with fluorosilicic and oxalic acid, but this will be at the cost of higher reagent consumption and
higher Nb losses. Overall, there was not any apparent advantage to two stage flotation over direct flotation.

Ore B was initially ground to 100% passing 53 microns to improve liberation, necessitated by the highly
complex nature of the ore. Batch tests results indicated that the ferro-carbonatite material was not susceptible to
flotation using amine collectors following either the two stage or the direct flotation flowsheet. An upgrade ratio of 8
was obtained using the two stage flotation flowsheet, however niobium recovery at this grade did not exceed 5%.
Likewise, direct amine flotation resulted in minimal upgrading (upgrade ratio of 3) with niobium recovery limited to
7.3%.

247
 47th Annual Canadian Mineral Processors Operators Conference©, Ottawa, Ontario, January 20-22, 2015

Grade recovery curves in Figure 3 demonstrate that a high grade concentrate can be obtained using either
two stage or direct amine flotation on the sövite ore. The ferrocarbonatite ore sees minimal upgrade when treated by
the same flowsheet and reagent scheme.
80

70

Ore A: Sövite
60
Distribution (% Nb2O5)

50

40

30
Ore B: Ferro-carbonatite
20 Ore B: Two Stage Flotation
Ore B: Direct Amine Flotation
10 Ore A: Direct Amine Flotation
Ore A: Two Stage Flotation
0
0 10 20 30 40 50 60

Assay % Nb2O5

Figure 3 - Grade recovery relationships for two stage and direct amine flotation of sövite and ferro-carbonatite ores

Challenges: Flotation of Nb minerals with amine collectors

Complicated Flowsheets
While Ore A was successfully upgraded by flotation with amines, a number of challenges are associated
with this flowsheet. A summary of niobium recovery to various stages of the process, as well of major sources of
niobium losses can be found in Table 4.

Table 4 - Niobium distribution in various products of two stage and direct amine flotation of Ore A
Flowsheet Product Wt. (%) Nb Distribution (%)

Overall Feed 100.0 100.0

After initial de-slime 84.5 89.6

Two Stage Flotation Combined Slimes 20.3 14.4

Pyrite Concentrate 1.6 1.7
Carbonate Concentrate 44.1 10.0
Niobium Circuit Feed 33.6 73.6
Overall Feed 100.0 100.0
After initial de-slime 84.7 89.4
Direct Flotation Combined Slimes 29.2 22.9
Pyrite Concentrate 2.0 1.8
Niobium Circuit Feed 68.4 75.0

Carbonate & Pyrite Concentrates

In two stage flotation, approximately 10% of the niobium is lost to the carbonate concentrate. QEMSCAN
analysis of the carbonate concentrate from a locked cycle testing on Ore A revealed that of the pyrochlore reporting
to the carbonate concentrate, nearly 68% was free and liberated suggesting that entrainment, possibly together with
inadvertent activation, was culpable for this loss. In a second locked cycle test, fatty acid collector addition was held
constant while niobium mineral depressant (causticized corn starch) was increased by nearly 1000g/t. Under these
modified conditions, free and liberated pyrochlore reporting to the carbonate concentrate was reduced to just below

248
 47th Annual Canadian Mineral Processors Operators Conference©, Ottawa, Ontario, January 20-22, 2015

44%, and overall niobium reporting to the carbonate concentrate was reduced from 24.7% to 7.6%. The mass pull to
the carbonate concentrate in both tests was 41%. It can be concluded that consistency in carbonate flotation with
minimal niobium losses is highly sensitive to operating conditions (such as pulp temperature and conditioning
density), and while losses can potentially be reduced through higher additions of niobium depressants, some loss to
this stream is inevitable.

Pyrite flotation was also performed prior to niobium flotation, as sulphur levels in the final concentrate must
be minimized for pyrometallurgical processing. It was found that pyrite left in the system floated with the amine
collector and reported to the niobium concentrate. Similar to the carbonate concentrate, this pyrite concentrate was
also a source of niobium losses, although the loss was less pronounced than in the carbonate concentrate due to much
lower mass pull. Cleaning of the pyrite concentrate may help in reducing niobium losses.

Slimes
To achieve successful flotation with amines, the fines fraction must be removed. This practice of desliming
the plant feed before flotation with amine collectors is practiced at Niobec, Catalao and Araxa, all operations
removing the -10 to -15 micron fraction in several stages using a series of hydrocyclones (Guimares & Weiss, 2003;
de Souza Paraiso Filho & de Fuccio, 1982; Gendron et al., 1981). When treating Ore A during batch testing, the -15
micron material was removed prior to carbonate flotation, and then a second time prior to niobium flotation. A
similar practice was employed prior to flotation of Ore B.

For Ore A, the use of direct flotation has proven to be equal in grade-recovery performance as the two stage
flotation process, as can be observed from the curves in Figure 3. The benefit of this flowsheet option is primarily
embodied by circuit simplification; less equipment, fewer reagents and fewer opportunities for niobium losses. The
elimination of carbonate flotation should inherently lead to higher overall recoveries. However, as is evident from
Table 4, instead of reporting to the final concentrate, the 10% of niobium that would have likely been lost to the
carbonate concentrate reports to the slimes fraction, further supporting the notion that pyrochlore is lost to the
carbonate concentrate due primarily to entrainment of fine particles.

Magnetic Concentrate
In some cases, low intensity magnetic separation may be run on the overall circuit feed, the niobium circuit
feed or any of the concentrate streams. While mass pull to this magnetic fraction is generally quite low, niobium loss
does occur.

Combined Niobium Losses

Overall, it should be noted that the both amine flotation flowsheets are quite complex and there are several
opportunities for niobium loss. While some of these losses are low, when compounded the overall loss can be
significant. In the case of the batch tests summarized on Table 4, approximately 25% of the niobium is lost before
even reaching the niobium flotation circuit.

Acidic Pulp and High Acid Addition Rates

As previously discussed, Nb flotation with amines must be carried out under increasingly acidic conditions,
pulp pH is usually around 5.5 in niobium rougher flotation and decreasing to as low as 1.8 in the final cleaners. This
necessitates anti-corrosive equipment, and high addition rates of acids which may be both dangerous and costly.

From Table 5, acid consumption, particularly of fluorosilicic acid, is significantly reduced in the two stage
flotation process. However, both flowsheet options require large quantities of acid in order for flotation to be
successful. Further, acid consumption in the two stage process is dependent on mass pull to the carbonate
concentrate.

Table 5 - Acid consumption in batch flotation tests on Ore A

Flowsheet H2 SiF6 (g/t) Oxalic (g/t)

Two Stage Flotation 4600 2400

Direct Flotation 7575 2100

249
 47th Annual Canadian Mineral Processors Operators Conference©, Ottawa, Ontario, January 20-22, 2015

While increased mass pull in the carbonate circuit may result in lower acid consumption, it can also lead to increased
niobium losses, possibly negating any additional financial benefit derived from lower acid requirement.

Recovery of fine grained pyrochlore and other Nb-minerals

QEMSCAN analysis was performed on the 2nd Cleaner tailings and the final concentrate from two-stage
locked cycle testing on Ore A. QEMSCAN was also performed on the rougher tailings and final concentrates of two-
stage batch flotation tests on Ore B. A particle size distribution of these flotation products can be found in Figure 4.

100 100
90 90
80 80
Cumulative % Passing

70

Cumulative % Passing
70
60 60
50 50
40 Ore A: Sövite 40 Ore B: Ferro-carbonatite

30 2nd Cl Tails: Nb-Minerals Ro tails: pyrochlore

30
2nd Cl Tails: Overall size distribution Ro Tails: Overall size distribution
20 20
Concentrate: Nb Minerals
Concentrate: pyrochlore
10 Concentrate: Overall size distribution 10
Concentrate: Overall size distribution
0 0
0 15 30 45 60 75 90 105 120 0 15 30 45 60
Particle Size (µm) Particle Size (µm)
Figure 4 - Particle size distribution of entire product and Nb-minerals in concentrates and tailings after two stage
flotation with amines

While the initial grind size of Ores A and B were different, the trend in product size distribution is very
similar. In both cases, the Nb minerals reporting to the concentrate are notably coarser than those reporting to the
tailings and the Nb minerals in the tailings are considerably finer than the overall tailings particle size distribution. In
Ore A, Nb minerals in the concentrate product are approximately 80% passing 60 microns as opposed to 80%
passing 25 microns in the 2nd cleaning tailings. In Ore B, pyrochlore in the concentrate is nearly 80% passing 28
microns compared to 80% passing 10 microns in the tailings. Since the hardness of pyrochlore group (~5 on the Moh
scale) is greater than those of the major gangue minerals, calcite (~3) and dolomite (~3.7), the differences in size
distributions are expected.

From these results it is possible to deduce that amine collectors have difficulty in recovering extremely fine
grained pyrochlore particles. In the case of ore A, inability to recover these fine particles may not be entirely
detrimental to the project. However, in a highly refractory deposit such as Ore B, where liberation only becomes
significant below 25 microns, the inability to recover extremely fine pyrochlore grains indicates that it will be
extremely difficult to achieve a high grade concentrate with reasonable niobium recovery. This may also explain
some of the losses during Nb cleaner flotation. Since the main strategy is to minimize flotation time to prevent
gangue mineral recovery, some of the finer pyrochlore particles with inherently slow flotation kinetics may not have
enough time to be recovered.

QEMSCAN analysis also provided a breakdown of different niobium minerals each of the flotation
products. These values have been compared to the distribution of niobium minerals in the feed in Table 6.

Table 6 – Distribution of niobium bearing minerals in feed, concentrate and tailings products after two stage flotation
with amines
Ore A: Sovite Ore B: Ferro-Carbonatite
Mineral Mass (%)
Feed 2nd Cl Tails Conc Feed Ro Tails Conc
pyrochlore, % 0.91 6.30 65.63 1.01 0.68 2.25
columbite, % 0.34 2.57 18.02 0.00 0.00 0.00
Nb-rutile, % 0.62 1.71 0.79 2.67 1.54 2.97
pyrochlore: columbite 2.68 2.45 3.64 - - -
pyrochlore: Nb-rutile 1.47 3.69 82.59 0.38 0.44 0.76

250
 47th Annual Canadian Mineral Processors Operators Conference©, Ottawa, Ontario, January 20-22, 2015

Most notable is that in the flotation of Ore B, the ratio of pyrochlore in the concentrate to pyrochlore in the
feed is significantly higher than the upgrade ratio of the Nb rutile. The ratio of pyrochlore to nb-rutile is 0.38 (by
weight) in Ore B feed material. This ratio is increased to 0.76 times in the flotation concentrate. A similar trend can
be observed when examining the upgrade ratio of pyrochlore compared to columbite in Ore A. There is 2.6 times
more pyrochlore than columbite in Ore A feed, and 3.64 times more pyrochlore than columbite in Ore A flotation
concentrate. Also in Ore A, the ratio of pyrochlore to Nb-rutile changes from 1.47 in the feed to 82.59 in the
concentrate, showing that pyrochlore is certainly being preferentially floated over Nb-rutile.

Thus, it appears that not only does this the reagent system suffer difficulty in recovering the fine grained
pyrochlore, amine collectors also seem to have a stronger affinity for pyrochlore over columbite and Nb rich titanium
minerals. While columbite did float with amines to a certain extent, recovery was much lower than pyrochlore. Work
by Chelgani, Hart, Marois, & Ourriban (2012) demonstrated that amine collector adsorption on the surfaces of low
Fe-pyrochlore grains was greater than on the surfaces of high Fe-pyrochlore grains. Since columbite is an iron
containing mineral, it is reasonable to assume that collector adsorption on columbite surfaces is weaker than on
pyrochlore, much in the same way as on the high Fe-pyrochlore surfaces. The same may also be true for the niobium
bearing rutile mineral surfaces. However, further surface characterization of these different niobium bearing minerals
is required to fully understand the differences in their floatability with amines.

Batch flotation with hydroxamates

A large fraction of pyrochlore in Ore B was considered to be slimes material (i.e. sub 15µm), a novel
approach was taken in an attempt to reject sufficient carbonate material and mass while maintaining niobium
recovery for further downstream processing. One known method of recovering extremely fine particles is through the
use of chelating reagents such as hydroxamates, which have a strong affinity for iron, titanium, and niobium.

Ore B was ground to 100% passing 37µm and then floated directly using benzohydroxamic acid in five
rougher stages, followed by one cleaner and three cleaner scavenger stages. No desliming was carried out prior to
flotation. The result was a combined cleaner and cleaner scavenger product with an upgrade ratio of 6.5 with over
60% of niobium recovered in less than 10% of the mass. Titanium recovery to the concentrate using benzohydroamic
acid followed niobium recovery and was just above 50%, compared to less than 5% in flotation with amines. This
indicates that niobium rich titanium minerals (Nb-rutile and ilmenrutile) are also recovered using this reagent
scheme. These results, along with the results from amine flotation for comparison purposes, can be found in Figure
5.

100

90 Ore B: Two Stage Flotation

Ore B: Direct Amine Flotation
80 Ore B: Hydroxamate Flotation

70
Distribution (% Nb2O5)

60

50

40

30

20

10

0
0 1 2 3 4 5 6 7 8

Assay % Nb2O5

Figure 5 - Comparison of amine and hydroxamate flotation on Ore B

QEMSCAN analysis of a rougher concentrate obtained from Ore B using benzohydroxamic acid identified
that in this product; the pyrochlore was 87% passing 12 microns, which is significantly finer than the pyrochlore
recovered to the concentrate using amines.

While the final flotation concentrate is still far from meeting specifications for a pyrochlore end product
which can be used to produce ferroniobium, the upgrade ratio and recovery are enough for further
251
 47th Annual Canadian Mineral Processors Operators Conference©, Ottawa, Ontario, January 20-22, 2015

hydrometallyurgical processing with greatly reduced acid consumption compared to hydrometallurgical treatment of
the whole ore.

A similar reagent scheme/flowsheet was adopted and tested on Ore A. The results are presented in Figure 6.
It can be observed that until a concentrate grade of above 20% Nb2O5, the hydroxamic reagent system out
performed both two stage and direct amine flotation in terms of recovery, due to the fact that no desliming was
performed prior to flotation. However, in the benzohydroxamic acid case, it was impossible to obtain a final
concentrate grade which meets industry specifications for ferroniobium production (generally above 50% Nb2O5).
This is largely due to the collector’s affinity for iron which was recovered into the final concentrate, diluting the
grade.

100

90 Ore A: Direct Amine Flotation

Ore A: Two Stage Flotation
80 Ore A: Hydroxamate Flotation

70
Distribution (% Nb2O5)

60

50

40

30

20

10

0
0 10 20 30 40 50 60

Assay % Nb2 O5

Figure 6 - Comparison of amine and hydroxamate flotation on Ore A

In the flotation of Ore A with benzohydroxamic acid, a fourth cleaner scavenger concentrate with a grade of
28.74% Nb2O5 (an upgrade ratio of nearly 57) at 53% Nb recovery was obtained. Examining the curves in Figure 6,
it is evident that in rougher flotation and initial cleaning stages, the benzohydroxamic acid outperforms amine
collectors by more than 15% recovery at any given grade until the concentrate reaches an assay around 20% Nb2O5.

Reagent addition rates and costs

While the benzohydroxamic acid system allows for major simplification of the flotation circuit and
eliminates the need for desliming, addition rates of this reagent are generally in the order of kilograms per ton of ore.
Depending on the flowsheet configuration, these high addition rates alone can bring project economics into question.
Furthermore, there is currently no flotation reagent grade commercial manufacturer of benzohydroxamic acid in
North America and reagents must be imported from China.

Rejection of impurities from the final concentrate

Regardless of the flowsheet or reagent scheme, selective rejection of impurities from the final concentrate is
imperative when producing a product that meets the minimum specification for ferroniobium production. In the
amine flotation reagent scheme, whether two stage or direct flotation, the presence of silica in the final concentrate is
problematic. While other impurities such as carbonate and phosphate minerals can be removed through atmospheric
leaching of the concentrate, silica cannot be removed in this process. Moreover, silica floats readily with amines and
has very similar physicochemical properties to pyrochlore, making selective flotation of pyrochlore over silica
extremely difficult.

In many cases, a SiO2 assay below 3.5% is required to meet pyrometallurgical specifications for
ferroniobium production. The rejection of silica can be achieved through reduction in pulp pH using fluorosilicic and
oxalic acid. From Figure 7, it is clear that silicate rejection only begins to occur only when pH drop below 4. This
corresponds to a drop in CaO assay to below ~15%. It is likely that only once a sufficient amount of acid consuming
carbonate minerals have been rejected, is it possible to achieve a pH of 4 or below, and thus reject silicates. The
rejection of silicates after reaching a pulp pH of 4 is consistent in both two stage and direct flotation. It should be
noted that even once low pulp pH conditions are reached, silica rejection may not be satisfactory. Silica rejection
through pH control is still not well understood and some operations (Catalão) employ a reverse silicate flotation
circuit ahead of niobium flotation to remove this impurity (Guimares & Weiss, 2003).

252
 47th Annual Canadian Mineral Processors Operators Conference©, Ottawa, Ontario, January 20-22, 2015

60 60
Direct Flotation (Amine) Two Stage Flotation (Amine)

50 Nb2O5 50 Nb2O5
SiO2 SiO2
CaO CaO

Distribution (%)
Distribution (%)

40 40

30 30

20 20

10 10

0 0
0 1 2 3 4 5 6 0 1 2 3 4 5 6
pH pH
Figure 7 - Assay of Nb2O5 and impurities with respect to pulp pH in the flotation of Ore A

Unlike amines, benzohydroxamic acid is highly selective over silicate and carbonate minerals. In this
reagent system the major impurities causing issue are titanium and iron oxides. Figure 8 shows recovery of both iron
and titanium in relation to niobium recovery. In flotation of both Ore A and Ore B, there was greater degree of
selectivity over iron oxides than titanium oxides. Even so, selectivity over iron oxides was not ideal and some iron
recovery cannot be avoided. A challenge exists in finding selective iron depressants, as many known depressants for
iron oxides (such as causticized corn starch) also depress pyrochlore.

Flotation of niobium minerals over titanium oxides appears to be even less selective. In the case of Ore B,
recovery of TiO2 in relation to recovery of Nb2O5 was almost perfectly linear. This indicates there was almost no
selectivity (the collector has equal affinity for both Nb and Ti), or that much of the niobium and titanium were
coming from the same mineral (i.e. Nb-rutile), or possibly middlings particles of pyrochlore associated with titanium
oxide minerals. If the latter is the case, selective titanium depressants cannot increase selectivity without hindering
niobium recovery unless finer grinding is undertaken (i.e. to 100% passing 20 micron), which may not be
economically practical. It is important to note that the recovery of titanium minerals to the final concentrate may be
detrimental to downstream processing, even if these minerals are niobium rich and contribute to final Nb recovery.

100

90

80
Distribution Nb2O5 (%)

70

60

50
Hydroxamate Flotation
40

30 Ore A: TiO2
Ore A: Fe2O3
20
Ore B: TiO2
10
Ore B: Fe2O3
0
0 20 40 60 80
Impurity Distribution (%)
Figure 8 - Distribution of Nb2O5 and impurities in flotation with benzohydroxamic acid

Selectivity over titanium oxides in the flotation of Ore A is much higher, and can be attributed to better
liberation in this sample, as well as lower niobium deportment to titanium oxide minerals. Even in this case the
collector’s strong affinity for titanium is apparent, and titanium recovery to the concentrate is nearly 20%.

253
 47th Annual Canadian Mineral Processors Operators Conference©, Ottawa, Ontario, January 20-22, 2015

CONCLUSIONS

In conclusion, many of the difficulties associated with niobium flotation with amine collectors such as complicated
circuits, acidic pulp conditions, difficulty recovering fine grained pyrochlore, columbite and Nb-rutile, can be
addressed through flotation with benzohydroxamic acid. However, this reagent system hosts its own set of issues,
many of which presently have no known solutions.

Overall, the reagent scheme and flowsheet selection should be dictated by the inherent mineralogical limitations of
the ore. Well liberated, coarser grained pyrochlore bearing ores may be better upgraded using a typical amine
flotation circuit, either two stage or direct. These ores can also be upgraded using benzohydroxamic acid, but not
without limitations in terms of concentrate grade. Ores with finely disseminated Nb minerals, which are not
amenable to flotation with amines can be upgraded into a low grade bulk concentrate using benzohydroxamic acid
for further downstream processing.

Impurity rejection from the flotation concentrates remains of chief concern, independent of the reagent scheme.
Better understanding of the optimal conditions and identification of selective depressants for the rejection of silicates
in amine flotation, and iron and titanium oxides in hydroxamate flotation is required for the realization of major
process improvements.

REFERENCES

Bulatovic, S. M. (2010). Flotation of Niobium/Tantalum Ores. In S. M. Bulatovic, Handbook of Flotation Reagents:

Chemistry, Theory and Practice - Volume 2: Flotation of Gold, PGM and Oxide Minerals (pp. 111-125).
Elsevier B.V.

Chelgani, S. C., Hart, B., Marois, J., & Ourriban, M. (2012). Study of pyrochlore surface chemistry effects on froth
flotation by SEM-EDX. Mineral Engineering (30) , 62-66.

de Souza Paraiso Filho, O., & de Fuccio, R. J. (1982). Mining, ore preparation and ferroniobium production at
CBMM. Niobium Proceedings of the International Symposium (pp. 113-133). San Francisco, California:
The Metallurgical Society of AIME.

Filho, A. I., Riffel, B., & de Faria Sousa, C. A. (2002). Some aspects of the mineralogy of CBMM niobium deposits
and pyrochlore ore processing. Niobium Proceedings of the International Symposium, (p. 13). Orlando, Fl,
USA.

Gendron, L., Biss, R., & Rodrigue, M. (1981). Underground mining and pyrochlore ore processing at Niobec mine,
Quebec, Canada. Niobium Proceedings of the International Symposium (pp. 79-113). San Francisco,
California: The Metallurgical Society of AIME.

Gorlovskii, S. I., Eropkiny, I., Koval, E. M., Strelstin, V. G., Khobotova, N. P., & Shtchukina, E. E. (1968).
Improvement in concentration technology of some rare metals ores, based on taking advantage of
complexing alkyl hydroxamic acids pecularities of action. VIII International Mineral Processing Congress,
(pp. D-3-9). Leningrad.

Gorlovskii, S. I., Khobatova, N. P., & Shtchukina, N. E. (1966). Concentration of niobium-containing ores with use
of a complexing collector. Tsvetnye Metally (7)2 , 7-12.

Guimares, H. N., & Weiss, R. A. (2003). The complexity of the niobium deposits in the alkaline ultramafic intrusions
Catalao I and II - Brazil. Sao Paulo, SP, Brasil: Mineracao Catalao.

Ni, X., & Liu, Q. (2012). Developing flotation reagents for niobium oxide recovery from carbonatite Nb ores.
Minerals Engineering (36-38) , 111-118.

Oliveira, J. F., Saraiva, S., & Oliveira, A. (2001). Technical note kinetics of pyrochlore flotation from Araxa mineral
deposits. Minerals Engineering (14) , 99-105.

U.S. Geological Survey. (2013). Minerals Commodity Summaries. Reston, Virginia: U.S. Geological Survey.

254