XXXI International Mineral Processing Congress 2024 Proceedings/Washington, DC/Sep 29–Oct 3 3159
phases such as the binder, which is fundamental for the
functioning of a LIB, covers and unselectively modifies the
surface properties of all phases, posing major challenges for
their separation (Vanderbruggen et al., 2021b). Thus, many
studies focus on pre-treatment options to minimize the
binder influence on the selective separation of CAMs and
graphite (Vanderbruggen et al., 2022b Yu et al., 2023).
Up to now, the energy intensive pyrolysis pre-treatment
remains as the most efficient pre-treatment for the selectiv-
ity of black mass flotation (Yu et al., 2023).
The field of automated mineralogy has contributed
to improving our understanding of mineral separation by
quantifying microstructural particle properties (e.g., size,
shape, composition, association, and liberation)—infor-
mation that can be correlated to process performance to
understand the natural limitations and/or opportunities for
selective mineral separation (Lotter, 2011). A recent parti-
cle-based separation modelling (PSM) methodology, which
combine automated mineralogy and statistical learning,
allow for quantifying the recovery of individual particles
in each separation process based on their microstructural
properties (Pereira et al., 2021b). This method has been
applied to diverse ores and separation processes and helped,
inter alia, to compute the flotation kinetics of individual
particles (Pereira et al., 2021a) and to identify the recovery
mechanisms of distinct phases in a feed material (Pereira et
al., 2023).
Lately, Vanderbruggen et al. (2021a) introduced a
routine to quantify the microstructural properties of black
mass particles—information that has already been used
to, for example, understand the influence of the overall
recycling chain on the selectivity of black mass flotation
(Vanderbruggen et al., 2022a). Moving forward, the meth-
odologies of Vanderbruggen et al. (2021a) and Pereira et al.
(2021b) are combined in this study to quantify the recov-
ery behaviour of individual graphite and CAM particles in
froth flotation under different pulp density conditions. A
model black mass composed of graphite, LiCoO2 (LCO),
LiNiMnCoO2 (NMC), aluminium and copper foils, and
quartz as entrainment tracer material is used for the inves-
tigation. Since all particles are liberated in this model black
mass, the relation between size and shape and the recovery
of distinct phases remains of interest for this study.
METHODOLOGY
In summary, a model black mass is compiled with
spheroidized gr`Taphite, LCO, NMC, aluminium and
copper foils, and quartz as entrainment tracer material.
Flotation experiments are performed at different percent-
age of solids in the pulp. Four concentrates are collected per
test. Samples from the concentrates and tailings fraction
are characterised with automated mineralogy, which data is
used to train particle-based separation models. Modelling
outcomes are used to evaluate the relation between the
recovery of individual particles and their composition, size,
and shape.
Samples
For this study, flotation tests of a model black mass consist-
ing of fully liberated and binder-free active particles. This
model black mass was prepared by mixing pristine lithium
metal oxides LiNi0.33Mn0.33Co0.33O2 (NMC-111,
MSE supplies, Product No. PO0126), LiCoO2 (LCO.
Targray, USA, Product No. SLC03007) and spheroidized
natural graphite (ProGraphite GmbH, product No. 1112-
1). In addition, some typical black mass impurities, such as
Cu foil and Al foil, were added. To understand the entrain-
ment degree of the particles as well, a SiO2 tracer was added.
Its D20, D50, and D80 are respectively 60 µm, 115 µm,
and 165 µm. Two solids percentage conditions were stud-
ied. The ratio between components in the model black
mass is the same for both cases, as displayed in Table 1.
Froth Flotation
The batch flotation tests were carried out in a laboratory-
scale mechanically stirred 2 L cell, specifically the GTK
Labcell from Outotec, which is equipped with automatic
scraping. This setup was chosen to ensure precise control
Table 1. Modal composition of model black mass used in the different tests
Phases Density, g/cm3 Ratio, %
Mass (g) at 12.7%
Solids Content
Mass (g) at 20.4%
Solids Content
Graphite 2.3 52.0 130.0 208.0
NMC 3.5 15.0 37.5 60.0
LCO 4.8 15.0 37.5 60.0
Quartz (SiO
2 )2.6 10.0 25.0 40.0
Cu foil 8.9 3.0 7.5 12.0
Al foil 2.7 5.0 12.5 20.0
Total 100.0 250.0 400.0
phases such as the binder, which is fundamental for the
functioning of a LIB, covers and unselectively modifies the
surface properties of all phases, posing major challenges for
their separation (Vanderbruggen et al., 2021b). Thus, many
studies focus on pre-treatment options to minimize the
binder influence on the selective separation of CAMs and
graphite (Vanderbruggen et al., 2022b Yu et al., 2023).
Up to now, the energy intensive pyrolysis pre-treatment
remains as the most efficient pre-treatment for the selectiv-
ity of black mass flotation (Yu et al., 2023).
The field of automated mineralogy has contributed
to improving our understanding of mineral separation by
quantifying microstructural particle properties (e.g., size,
shape, composition, association, and liberation)—infor-
mation that can be correlated to process performance to
understand the natural limitations and/or opportunities for
selective mineral separation (Lotter, 2011). A recent parti-
cle-based separation modelling (PSM) methodology, which
combine automated mineralogy and statistical learning,
allow for quantifying the recovery of individual particles
in each separation process based on their microstructural
properties (Pereira et al., 2021b). This method has been
applied to diverse ores and separation processes and helped,
inter alia, to compute the flotation kinetics of individual
particles (Pereira et al., 2021a) and to identify the recovery
mechanisms of distinct phases in a feed material (Pereira et
al., 2023).
Lately, Vanderbruggen et al. (2021a) introduced a
routine to quantify the microstructural properties of black
mass particles—information that has already been used
to, for example, understand the influence of the overall
recycling chain on the selectivity of black mass flotation
(Vanderbruggen et al., 2022a). Moving forward, the meth-
odologies of Vanderbruggen et al. (2021a) and Pereira et al.
(2021b) are combined in this study to quantify the recov-
ery behaviour of individual graphite and CAM particles in
froth flotation under different pulp density conditions. A
model black mass composed of graphite, LiCoO2 (LCO),
LiNiMnCoO2 (NMC), aluminium and copper foils, and
quartz as entrainment tracer material is used for the inves-
tigation. Since all particles are liberated in this model black
mass, the relation between size and shape and the recovery
of distinct phases remains of interest for this study.
METHODOLOGY
In summary, a model black mass is compiled with
spheroidized gr`Taphite, LCO, NMC, aluminium and
copper foils, and quartz as entrainment tracer material.
Flotation experiments are performed at different percent-
age of solids in the pulp. Four concentrates are collected per
test. Samples from the concentrates and tailings fraction
are characterised with automated mineralogy, which data is
used to train particle-based separation models. Modelling
outcomes are used to evaluate the relation between the
recovery of individual particles and their composition, size,
and shape.
Samples
For this study, flotation tests of a model black mass consist-
ing of fully liberated and binder-free active particles. This
model black mass was prepared by mixing pristine lithium
metal oxides LiNi0.33Mn0.33Co0.33O2 (NMC-111,
MSE supplies, Product No. PO0126), LiCoO2 (LCO.
Targray, USA, Product No. SLC03007) and spheroidized
natural graphite (ProGraphite GmbH, product No. 1112-
1). In addition, some typical black mass impurities, such as
Cu foil and Al foil, were added. To understand the entrain-
ment degree of the particles as well, a SiO2 tracer was added.
Its D20, D50, and D80 are respectively 60 µm, 115 µm,
and 165 µm. Two solids percentage conditions were stud-
ied. The ratio between components in the model black
mass is the same for both cases, as displayed in Table 1.
Froth Flotation
The batch flotation tests were carried out in a laboratory-
scale mechanically stirred 2 L cell, specifically the GTK
Labcell from Outotec, which is equipped with automatic
scraping. This setup was chosen to ensure precise control
Table 1. Modal composition of model black mass used in the different tests
Phases Density, g/cm3 Ratio, %
Mass (g) at 12.7%
Solids Content
Mass (g) at 20.4%
Solids Content
Graphite 2.3 52.0 130.0 208.0
NMC 3.5 15.0 37.5 60.0
LCO 4.8 15.0 37.5 60.0
Quartz (SiO
2 )2.6 10.0 25.0 40.0
Cu foil 8.9 3.0 7.5 12.0
Al foil 2.7 5.0 12.5 20.0
Total 100.0 250.0 400.0