Research Article

Cite This: ACS Sustainable Chem. Eng. 2018, 6, 1478−1488 pubs.acs.org/journal/ascecg

Selective Process Steps for the Recovery of Scandium from Jamaican

Bauxite Residue (Red Mud)
Remya P. Narayanan,‡ Nikolaos K. Kazantzis,† and Marion H. Emmert*,‡
‡
Department of Chemistry and Biochemistry and †Department of Chemical Engineering, Center for Resource Recovery and
Recycling, Worcester Polytechnic Institute, 100 Institute Road, Worcester, Massachusetts 01609, United States
*
S Supporting Information

ABSTRACT: We report the development of a process allowing

the selective, sustainable recovery of scandium (Sc) with 75%
efficiency from Jamaican bauxite residue (red mud), a waste
product from aluminum production. The process design is inspired
by green chemistry principles and focuses on establishing highly
selective process steps (sulfation, leaching, and precipitation) in
order to minimize costs and waste produced. In addition to
scandium oxide, the chosen approach produces mixed rare earth
oxides as a side product, thus isolating an average of 88% of all rare
earth elements contained in red mud. Furthermore, in light of
predicted supply shortages of the critical material Sc and the need to establish cost-effective bauxite residue remediation
techniques, a systematic Monte Carlo-based economic performance assessment framework is developed in order to evaluate the
economic prospects of the proposed process system in the presence of irreducible uncertainty.
KEYWORDS: Sustainable process design, Rare earth elements, Scandium, Jamaican bauxite, Red mud, Remediation techniques

■ INTRODUCTION
Since the Chinese government introduced export restrictions
for rare earths (REs; lanthanides + Sc, Y, and La) in 2011,
significant price fluctuations for these critical materials have
occurred.1−3 However, even though prices have significantly
relaxed since 2011, uncertainties in the supply of REs remain
that are likely to influence current and future decisions
regarding the use of RE containing materials in many
applications (e.g., wind and tidal turbines, hybrid, and electric
vehicles).4
Despite these uncertainties in the supply chain, the scandium
(Sc) market in particular is predicted to grow substantially over
the next decades due to the projected use of Sc in lightweight,
high-strength aluminum alloys for aerospace and fuel-efficient
vehicle applications.5 For example, a recent market analysis
predicts a significant growth of Sc consumption from a current
value of 10−15 t/y in 20156 to 98 t/y for aircrafts and 3000 t/y
for light vehicle production by 2023.7 As currently available Sc
is mainly produced as a byproduct of other processes,
increasing production rates is likely difficult;6 therefore, an
urgent need for new Sc sources exists in order to prevent future
shortages and price increases.
Bauxite residue, also called red mud (RM), is customarily
considered a waste product resulting from aluminum ore
refining but may also be viewed as a potential alternative source
of Sc.8 RM is being produced and stockpiled on scales as large
as 150 million t/y, of which only less than 4 million t/y are
found in follow-up processes.9,10 The Sc content of RM varies
between 15 and 170 ppm, depending on the source of bauxite;
furthermore, other REs are also present (∼850 to 1200 ppm of
mostly Y, La, Ce, Pr, and Nd).11 If all Sc from RM could be
recovered, 6600 to 20 400 t/y Sc could be made available just
based on the yearly production of RM. These amounts could
easily address projected supply shortages of Sc, while at the
same time contribute to remediating RM, which is dangerous to
aqueous ecosystems due to its strongly basic pH.12,13 As such,
using RM as a resource in a commercially viable process may
also serve to enhance the economic performance of alumina
refineries.14
Several previous attempts have been made to establish red
mud as a resource; however, only few processes that establish a
complete pathway from RM to Sc2O3 are known,12,15−21 but
many of these processes are unlikely to be realized on a large
scale due to reagent costs. Among the research in the relevant
literature that describes various optimization approaches of
single process steps, leaching studies to liberate REs from red
mud are the most common.22 Unfortunately, the majority of
the published studies generally only focus on recovery
efficiencies and selectivities without taking reagent costs into
account upon reaction design.23 Furthermore, several compa-
nies have claimed in recent years to have successfully realized
the recovery of Sc from RM,24,25 but very little is known about
the details of the employed process steps.
With this in mind, our work focused on optimizing
separation selectivity and minimizing reagent use in each step
of a newly design process. Both of these principles have
previously been formulated within the broader context of green
Received: October 30, 2017
Published: November 21, 2017

© 2017 American Chemical Society 1478 DOI: 10.1021/acssuschemeng.7b03968

ACS Sustainable Chem. Eng. 2018, 6, 1478−1488
ACS Sustainable Chemistry & Engineering
chemistry principles.26 Furthermore, the same design param-
eters have been applied for efficient and cost-effective process
design in RE recovery from magnets in motors.27 Employing
these principles, we identify and examine in this manuscript
salient economic performance profile characteristics of the
proposed process system for Sc recovery from RM (in
particular 75% Sc recovery as Sc2O3). Moreover, a commercial
viability analysis of the devised flowsheet shows highly favorable
prospects for process commercialization, even at currently low
Sc prices. Additionally, a mixed RE oxide is obtained as side
product of the process, which may generate further value-
enhancing opportunities. All of these exciting results are due to
rigorous optimization efforts, which establish highly selective
process steps and minimize undesired reagent and metal oxide
waste.
■
EXPERIMENTAL SECTION
Analysis of RM. The chemical analysis of a Jamaican red mud
sample was performed after dissolving the bauxite residue by an alkali
fusion/acid digestion sequence: The alkali fusion was carried out by
mixing 0.5 g of bauxite residue with 1.5 g of sodium carbonate and 1.5
g of sodium tetraborate decahydrate, followed by heating the mixture
in an alumina crucible at 1100 °C for 30 min. The residue (3.381 g)
was extracted using 100 mL 1:1 HCl:H2O, and the resulting
suspension was filtered. The ICP-OES analysis of the obtained filtrate
was performed six times to provide statistically meaningful average
values of elemental contents.
Sulfation and Roasting. The sulfation (acid mixing) of red mud
was performed by moistening 0.50 g of dried red mud with 40 wt %
water (0.20 mL) and 80 wt % concentrated H2SO4 (0.40 g, 0.22 mL)
in an alumina crucible. The mixture was then heated to 120 °C for 14
h in a vacuum oven. The XRD analysis (see the SI) of the obtained
material confirms the completion of sulfation. The sulfated RM was
then roasted at 700 °C for 1 h in a tube furnace to decompose low
thermal stability sulfates to oxides.
Leaching. The roasted RM (50 g) was leached by adding 50 wt %
water (25 mL) to the obtained residue, followed by ball milling (3000
rpm, glass balls; ULTRA-TURRAX Tube Drive) for 30 min. The
composition of the leach liquor was then analyzed using ICP-OES
(PerkinElmer Optima 8000), and the solid residue was analyzed by
TXRF (Bruker S2 Picofox). Both analyses confirm the leaching of 84%
Sc, 0% Fe, 8.2% Al, 0% Ti, 35% Ca, 100% Y, 100% La, 98% Ce, 100%
Pr, 99% Nd, 100% Gd, 100% Dy, 68% Er, and 100% Yb.
Mixed Rare Earth Oxide Precipitation. The pH of 25 mL of the
obtained leach liquor was adjusted to pH 8 using 2 mL of 2 M NaOH.
The mixed rare earth precipitate formed through pH adjustment was
removed by filtration and the filtrate composition was analyzed using
ICP-OES. The obtained precipitate was analyzed by TXRF. Both
analyses confirm that all rare earths but Sc (Y, La, Ce, Pr, Nd, Gd, Dy,
Er, and Yb) are separated as a precipitate and that all Sc remains in
solution in the filtrate.
Scandium Oxalate Precipitation. The pH of 500 mL filtrate was
adjusted to pH 1 using 15 mL of concentrated H2SO4. Then, scandium
oxalate was precipitated by adding solid oxalic acid (1.0 mg; 2.02
equiv. compared to the amount of Sc in the solution). The precipitate
was filtered to obtain 3.5 mg scandium oxalate from overall 10 g of RM
originally leached. The precipitate was analyzed by TXRF. The filtrate
was analyzed for remaining Sc by ICP-OES. Both TXRF and ICP-OES
analysis confirm the presence of only Sc in the precipitate and only
<1% Sc left in the filtrate. The overall Sc recovery is calculated to be
75%, based on the amount of Sc originally present in Jamaican red
mud.
■
RESULTS AND DISCUSSION
Process Design Overview. The chosen process design
(Figure 1) included initially a three-step approach, consisting of
selective leaching of REs, separation of REs from all bulk
1479
Research Article
Figure 1. Initial process design for Sc recovery from red mud.
elements, and separation of Sc from the obtained RE mixture.
In developing all of these general steps, separation efficiency
was of utmost concern in order to maximize the recovery of Sc
and to avoid producing waste oxide material.
Thus, we chose to explore a leaching step similar to a report
by Borra et al.,19 which combines sulfation, roasting, and
leaching and achieves a recovery of ∼60% Sc and >90% of
other REs after 7 days, while leaching only small amounts of Fe
and Ti (∼1% and ∼2%, respectively). We considered these
results to be a good starting point for further studies, if (i) the
recovery of Sc can be increased; (ii) leaching of bulk elements
can be further decreased (even 1% Fe and 2% Ti is considered a
significant amount in the leach liquor, as both elements occur
in high concentrations in RM); and (iii) the leaching times can
be significantly decreased.
If successful, the first process steps would be expected to
result in a large volume of leach liquor; as such, RE
precipitation as a second step could be expected to concentrate
the desirable RE components and also serve to separate REs
contained in the leach liquor from bulk elements. Finally,
separation of Sc from the RE concentrate may be possible by
either selective precipitation28 or solvent extraction.29
The success of this process design strategy depends strongly
on the selectivities that can be achieved in each step; therefore,
detailed optimization efforts for each step are described in the
following sections.
Determining RE Contents in Different Red Mud
Resources. Generally, the source of RM will determine its
Sc content, which in turn will likely influence the exact outcome
of process optimization. Therefore, we started the process
development by comparing the RE contents of different RM
sources. To this end, several RM samples of different origins
were collected (named in this manuscript Alcan, Alcoa, Korean,
and Jamaican RM in a reflection of their origins; however, no
further information regarding the source refineries or holding
ponds could be obtained due to trade secrets) and analyzed.
Comparison of the SEM pictures obtained for each sample
showed very fine materials that form agglomerates (Figure 2).
In a next step, bulk element concentrations were obtained for
all samples, using EDX and/or ICP-OES after alkali fusion (see
the SI for details). Table 1 summarizes the results of this
analysis. Interestingly, Al (0.05−0.23 kg/kg) and Ti (0.02−0.2
kg/kg) contents vary significantly between RM obtained from
Alcoa and the other RM samples, while Fe (0.18−0.3 kg/kg)
and Ca (0.02−0.03 kg/kg) contents are more comparable
between samples of different origins.
In order to determine the concentrations of trace elements in
the different RM samples obtained, RM samples were subjected
to alkali fusion and leaching: RM was first reacted with Na2CO3
and Na2B4O7·10H2O at 1100 °C, which was followed by acid
digestion in HCl solution, followed by filtration (see the SI for
DOI: 10.1021/acssuschemeng.7b03968
ACS Sustainable Chem. Eng. 2018, 6, 1478−1488
ACS Sustainable Chemistry & Engineering Research Article

Figure 3. Comparison of RE element contents of different red mud

samples. The graph shows the average values and standard deviations
obtained from alkali fusion and acid digestion processing of at least 7
independently prepared samples.

∼60% of Sc, while a large amount of other bulk elements (30%

Ca, 100% Na) are also leached out of RM. Thus, our work
focused on minimizing the amount of leached bulk elements,
while maximizing Sc leaching.
In an initial series of optimization steps, we focused on
optimizing the sulfation/roasting sequence, while keeping the
Figure 2. SEM images of RM samples of different origins. Sources: leaching conditions unchanged (H2O, 7 days, room temper-
Alcoa (top left), Jamaican RM (top right), Alcan RM (bottom left), ature).19 The sulfation/roasting sequence relies on the chemical
Korean RM (bottom right). reaction that converts oxides into hydrated sulfates at relatively
low temperature (120 °C), which is then followed by
details).30 The RE concentrations obtained from leaching are dehydration and SO3 extrusion at higher temperatures to
depicted in Figure 3. form oxides from thermally unstable sulfates. As the RE-based
Interestingly, the RE content varies dramatically between the sulfates exhibit higher decomposition temperatures than Fe
samples of different origins, with Jamaican RM being the richest sulfates (see Table 2), the roasting step would ideally transform
in both Sc (55 ppm) and other RE elements (e.g., 373 ppm Y; all Fe contained in RM into insoluble oxides,23 while all REs
287 ppm La; 366 ppm of Ce; 74 ppm Pr; 70 ppm of Nd; 37 remain soluble sulfates33−35 that can be dissolved.
ppm Gd). This suggests that Jamaican RM is the most suitable
RM source obtained for Sc recovery. According to other Table 2. Literature Values of Thermal Decomposition
reports, two types of RM (Australian red mud and Greek red Temperatures of Relevant Metal Sulfates
mud) with similar or higher Sc contents (54 and 120 ppm) may
also be suitable as a Sc resource;21,31,32 however, these types of metal sulfate onset temperature for sulfate to oxide conversion
RM samples could not be obtained within the time frame of Fe2(SO4)3 545 °C36
this study. TiOSO4 340 °C37
Selectivity in Sulfation and Roasting. Of particular CaSO4 >1000 °C37
importance in the recovery of Sc from RM is the selectivity in Na2SO4 870 °C37
the first step, as coleaching of REs and bulk elements can cause Al2(SO4)3 524 °C36
problems later in the recovery process, e.g. in potential solvent Sc2(SO4)3 700 °C38
extraction and precipitation steps. Due to the much higher Fe Y2(SO4)3 850 °C39
and Ti content in comparison to the RE content in RM, La2(SO4)3 840 °C39
entirely preventing those bulk elements from leaching is a Ce2(SO4)3 666 °C36
significant challenge; at the beginning of our experimental Nd2(SO4)3 800 °C39
studies, no reports that had met this challenge were available in
the literature. As mentioned previously in the process design We hypothesized that the removal of SO3 from the mixture
section, one report by Borra et al.19 showed results close to would have a significant effect on the conversion due to
completely selective leaching, with only ∼1% of Fe and ∼2% of chemical equilibrium considerations. Therefore, we performed
Ti leached from Greek RM. However, the process only recovers all roasting optimizations in a tube furnace open to air; in

Table 1. Bulk Element Concentrations for Four Types of RM as Obtained by EDX and/or ICP-OES after Alkali Fusion

element (kg/kg) Alcoa Jamaican RM Alcan RM Korean RM

Fe 0.2 ± 0.01 0.36 ± 0.02 0.3 ± 0.002 0.18 ± 0.005
Ti 0.18 ± 0.02 0.02 ± 0.002 0.04 ± 0.001 0.02 ± 0.004
Ca 0.02 ± 0.005 0.03 ± 0.001 0.02 ± 0.001 0.02 ± 0.002
Na 0.059 ± 0.001 0.025 ± 0.001 0.14 ± 0.001 0.06 ± 0.001
Al 0.16 ± 0.03 0.05 ± 0.001 0.2 ± 0.001 0.23 ± 0.003
Zr 0.065 ± 0.005 0.065 ± 0.005 not detected 0.01 ± 0.001

1480 DOI: 10.1021/acssuschemeng.7b03968

ACS Sustainable Chem. Eng. 2018, 6, 1478−1488
ACS Sustainable Chemistry & Engineering Research Article

Figure 4. Effect of roasting temperature on metal ion recovery in leaching. Conditions: (sulfation) 100 wt % Jamaican RM, 40 wt % H2O, 80 wt %
conc H2SO4, aluminum crucible, 120 °C, 14 h, tube furnace;19 (roasting) temperature specified in graphs, 1 h; (leaching) H2O (solid/liquid ratio =
1/50), room temperature, 7 days.

Figure 5. Influence of H2SO4 loading on leaching selectivity and yield. Conditions: (sulfation) 100 wt % Jamaican RM, 40 wt % H2O, 60−100 wt %
conc H2SO4, aluminum crucible, 120 °C, 14 h, tube furnace; (roasting) 700 °C, 1 h; (leaching) H2O (solid/liquid ratio = 1/50), room temperature,
7 days.

contrast, the original report19 employs a muffle furnace. With
this setup, the effects of different H2SO4 loadings and roasting
temperatures were explored, using Jamaican RM as a starting
material.
Initial sulfation conditions (120 °C, 14 h) were chosen
similar to the original report by Borra et al.19 Figure 4 shows
the effect of roasting temperature on percent element recovery.
Interestingly, the average RE recovery declines steeply at
temperatures higher than 715 °C, while Fe recovery is very low
when roasting is performed at temperatures higher than 700
°C.
This suggests that either 700 or 715 °C may be the most
suitable temperature for selective RE recovery. Importantly, Sc
recovery remains quantitative up to 715 °C (Figure 4, right
side) but falls to less than 20% when roasting is performed at
750 °C. This is likely due to the decomposition of Sc2(SO4)3
under these roasting conditions; however, the demonstrated
quantitative recovery of Sc up to a roasting temperature of 715
°C is a significant advancement in comparison to the reported
conditions. We speculate that this is possible due to the
improved furnace setup used for the herein described studies,
which removes gaseous products from the chemical equili-
brium, thus allowing a higher-yielding formation of insoluble Fe
oxides and higher Sc selectivity.
In a next step, the effect of the used H2SO4 loading on Sc
recovery and selectivity was studied. We chose to perform this
study at a roasting temperature of 700 °C to maximize the
overall RE recovery (88%) at a relatively low Fe recovery level
(5%; see Figure 4 above, obtained with 80 wt % of conc H2SO4
1481
used for sulfation and roasting). Interestingly, employing higher
H2SO4 loadings (100%) provided the same 88% overall RE
recovery, but Fe recovery was further enhanced (34%), thus
diminishing the selectivity for RE recovery (Figure 5). This may
be due to the incomplete decomposition of Fe2(SO4)3 or to a
more complete conversion of Fe oxides to Fe2(SO4)3 during
the sulfation reaction. Due to the lower obtained selectivity, the
use of higher H2SO4 loadings than 80 wt % was considered
unfavorable for further development.
Figure 5 further shows that lower H 2 SO 4 loadings
significantly decreased RE recovery (only 36% average) and
Sc recovery in particular (0%). Despite the low Fe leaching
observed under these conditions (0%), they were thus deemed
unfavorable. In summary, sulfation with 80 wt % concentrated
H2SO4 followed by roasting at 700 °C for 1 h were the
optimized sulfation/roasting conditions, providing the max-
imum average RE recovery (88%, with 100% Sc recovery) along
with recovery of 4.9% Fe, 3.0% Ti, 22.4% Ca, 21.7% Al, and
70% Na.
Selectivity and Efficiency in Leaching. The above
detailed optimization steps had achieved one important goal:
Leaching Sc and other REs nearly quantitatively from Jamaican
RM. To further improve the selectivity of RE leaching vs bulk
element leaching, the next series of studies focused on
optimizing the leaching step itself. To this end, initial
experiments were performed in order to gain insight into the
leaching kinetics to determine if selectivity can be achieved by
adjusting leaching times. As is illustrated in Figure 6, leaching of
the solid residue that is obtained from roasting at room
DOI: 10.1021/acssuschemeng.7b03968
ACS Sustainable Chem. Eng. 2018, 6, 1478−1488
ACS Sustainable Chemistry & Engineering
Figure 6. Kinetics of metal ion leaching. Conditions: (sulfation) 100
wt % Jamaican RM, 40 wt % H2O, 80 wt % conc H2SO4, aluminum
crucible, 120 °C, 14 h, tube furnace; (roasting) 700 °C, 1 h; (leaching)
H2O (solid/liquid ratio = 1/50), room temperature, 1−7 days.
temperature is a slow process: The maximum RE recovery is
only achieved after 6 days. Furthermore, the selectivity of RE
leaching as compared to bulk element leaching does not
considerably change during the course of 7 days. However,
processing times of this magnitude are difficult to realize in
pilot or production plants; as such, we turned our efforts to
approaches that may possibly enable faster leaching.
We rationalized that higher leaching temperatures may
accelerate leaching rates by providing more thermal energy to
overcome activation barriers required for hydrating RE ions
contained in a solid matrix. However, leaching experiments at
50 or 80 °C led to even slower leaching of REs from the solid
residue, compared to the leaching behavior at room temper-
ature (Figure 7). This may be due to the lower solubility of RE
sulfates at higher temperatures.20 As such, leaching at higher
temperatures was considered unsuitable for improving the
process.
Figure 7. Kinetics of RE leaching at different leaching temperatures.
Conditions: (sulfation) 100 wt % Jamaican RM, 40 wt % H2O, 80 wt
% conc H2SO4, aluminum crucible, 120 °C, 14 h, tube furnace;
(roasting) 700 °C, 1 h; (leaching) H2O (solid/liquid ratio = 1/50),
room temperature/50 °C/80 °C, 1−6 days.
Figure 8. Leaching upon sonication of leaching mixture. Conditions: (sulfation) 100 wt % Jamaican RM, 40 wt % H2O, 80 wt % conc H2SO4,
aluminum crucible, 120 °C, 14 h, tube furnace; (roasting) 700 °C, 1 h; (leaching) H2O (solid/liquid ratio = 1/50), room temperature, sonication in
water bath/no sonication.
1482
Research Article
The formation of small, cocrystalline morphologies in RM
after acid digestion has previously been documented;40
therefore, we speculated that microcrystallinity of the residue
obtained by roasting may hinder liberation of REs from the
crystal matrix, thus leading to long leaching times. We further
reasoned that leaching from cocrystalline materials may be
accelerated by applying force to the material, which could be
performed by sonication or milling. Therefore, our next
optimization study measured leaching performance upon
sonicating (Figure 8, left).
Excitingly, the average RE leaching observed was 80% af ter only
1 h of sonication, with almost quantitative RE leaching after 5 h.
Moreover, the leaching of bulk elements was signif icantly
decreased upon sonication leaching (to 0% Fe, 6% Ti, 22%
Ca, and 6% Al). This is an exciting discovery, as Fe and Ti are
present in large quantities in RM and may interfere in RE
isolation from the leach liquor. Sonication has previously been
documented to lead to a reduction in particle size of crystalline
material.41 We speculate that a reduction of particle size of the
roasted RM residue through sonication may accordingly result
in larger surface areas at which ion exchange with the
surrounding leach liquor can occur, thus accelerating the
leaching process. The right part of Figure 8 contrasts RE
leaching kinetics of the sonication method with the kinetics of
traditional, stirred RE leachinga clear improvement can be
observed.
Similar to sonication, ball milling also resulted in a significant
increase in leaching rate and selectivity (Figure 9): after only 30
min, 88% average RE recovery (89% Sc) was obtained in the
absence of any Fe and Ti leaching and at relatively low Al and Na
leaching (10% and 10%, respectively). Extending the ball
milling time to 60 min increased the recovery of REs to 95%;
however, the recovery for Sc remained at 89%, while Fe and Ti
leaching (0.01 and 1.06%, respectively) started to occur. The
exact mechanisms for achieving such highly selective Sc
leaching are currently not understood in detail. In-depth
understanding will likely require further detailed studies, which
were considered outside the scope of this manuscript.
Finally, the influence of the solid/liquid ratio on the
selectivity and efficiency of leaching was investigated; we
reasoned that a lower volume of leach liquor would be desirable
for subsequent processing. However, lowering the amount of
H2O used for leaching (Figure 10) resulted in a significant
decrease of overall RE recovery from 95% (with 50 wt % H2O)
to 69%, 65%, and 61% (with 30, 20, or 10 wt % H2O). We
speculate that this decrease in leaching efficiency may be caused
by solubility limitations of the leached salts in the leach liquor.
This decrease in RE recovery was undesirable; therefore,
DOI: 10.1021/acssuschemeng.7b03968
ACS Sustainable Chem. Eng. 2018, 6, 1478−1488
ACS Sustainable Chemistry & Engineering Research Article

Figure 9. Metal ion leaching after 30 (left) and 60 min (right) upon ball milling. Conditions: (sulfation) 100 wt % Jamaican RM, 40 wt % H2O, 80
wt % conc H2SO4, aluminum crucible, 120 °C, 14 h, tube furnace; (roasting) 700 °C, 1 h; (leaching) H2O (solid/liquid ratio = 1/50), room
temperature, ball milling (IKA ULTRA-TURRAX Tube Drive, glass ball diameter 6 mm, 3000 rpm), 30 or 60 min.

Thus, we explored conditions for precipitation by adjusting

the pH of the leach liquor with NaOH. The original pH of the
obtained solution was measured to be 3.57; the addition of
different amounts of NaOH resulted in pH values of 8, 10, 12,
and 14. In each case, a precipitate was formed. Determining the
remaining RE concentrations in solution by ICP-OES revealed
that the majority of REs had precipitated quantitatively even at
pH 8 (Figure 12), with some Ca also coprecipitating. However,

Figure 10. Ball-mill supported leaching of RM residue at different

solid/liquid ratios. Conditions: (sulfation) 100 wt % Jamaican RM, 40
wt % H2O, 80 wt % conc H2SO4, aluminum crucible, 120 °C, 14 h,
tube furnace; (roasting) 700 °C, 1 h; (leaching) H2O (1.0 wt % solid
residue, 10−50 wt % H2O), room temperature, ball milling (IKA
ULTRA-TURRAX Tube Drive, glass balls, 3000 rpm), 60 min.

further experiments were performed with a leach liquor

obtained after 30 min of ball milling with a solid/liquid ratio
of 1:50 (w/w). As a result of these studies, the leach liquor
contains the relevant metal concentrations as shown in Figure
10 for 50 wt % water.
Selectivity for RE Precipitation. The obtained RE
solution (leach liquor) contained all REs, including Sc and Y,
as well as La, Ce, Pr, Nd, Gd, Dy, Er, and Yb. Typical literature
approaches to separate and concentrate the REs have used
solvent extraction methods,42 which typically use relatively
expensive organic extractants and solvents. Therefore, a direct
use of solvent extraction with the obtained, dilute leach liquor
as aqueous phase was deemed undesirable from a cost
perspective. We postulated that precipitation of the REs
contained in the solution may allow concentrating the valuable
components while decreasing the volume of the solutions for
further processing (Figure 11).
Figure 12. RE precipitation from leach liquor. The leach liquor was
obtained after filtering off bulk materials after sulfation, roasting, and
leaching (30 min, ball milling). The pH of the leach liquor was
adjusted with NaOH (2 M). Precipitation percents are obtained by
analyzing the filtrate by ICP-OES.
Sc does not precipitate under these conditions, allowing for
separation of Sc f rom the other REs at this stage of the process.
This separation can be rationalized with the amphoteric
behavior of Sc(OH)3, which shows significant solubility at
both low and high pH values;43 similar amphoteric behavior is
known for Al(OH)3,44 which also reports to the filtrate.
Furthermore, analysis of the precipitate by TXRF confirmed
that only traces of Sc were contained in the formed solid. These

Figure 11. Weight balance and further separation options after roasting and leaching.

1483 DOI: 10.1021/acssuschemeng.7b03968

ACS Sustainable Chem. Eng. 2018, 6, 1478−1488
ACS Sustainable Chemistry & Engineering Research Article

data suggest that precipitation does not only concentrate REs Table 4. Element Concentrations of Filtrate after
but that this step also allows for an efficient separation of Sc Precipitationa
from the other REs originally contained in RM. Interestingly,
element concentration (g/L)
no precipitation of Al occurs under these conditions, suggesting
that Al is separated together with Sc from the REs through Sc 0.051 ± 0.001
precipitation. Ca 2.045 ± 0.002
Considering the mass balance (Figure 13, Table 4), it is Al 2.606 ± 0.001
a
inferred that the NaOH precipitation step provides access to a Conditions: (sulfation) 100 wt % Jamaican RM, 40 wt % H2O, 80 wt
% conc H2SO4, aluminum crucible, 120 °C, 14 h, tube furnace;
(roasting) 700 °C, 1 h; (leaching) H2O (1.0 wt % solid residue, 10−50
wt % H2O), room temperature, ball milling (IKA ULTRA-TURRAX
Tube Drive, glass balls, 3000 rpm), 60 min.

Na, Ca, and Al. In agreement with the literature,45,46

precipitating Sc from this solution by NaOH addition to
directly obtain Sc(OH)3 was unsuccessful, suggesting that other
precipitation reagents need to be explored. Previous studies
suggest that Sc can be precipitated from aqueous solutions as
the oxalate salt upon addition of oxalic acid.24,47 We considered
oxalic acid a desirable reagent, because oxalate salts can be
converted into oxides simply by roasting.48 Furthermore, oxalic
acid is a solid reagent, and therefore, less hazardous than other
acidic reagents.
When oxalic acid precipitation was attempted with the
Figure 13. Mass balance and separation obtained by precipitation. solution obtained after filtering off the mixed RE/Ca precipitate
(pH 8), several interesting observations were made (Figure 14,
Table 3. Element Concentrations of Leach Liquora top left chart): (i) Ca coprecipitates with Sc under these
element concentration (g/L) conditions; (ii) the Sc/Ca selectivity increases with the amount
of oxalic acid equivalents added; (iii) the recovery of Sc also
Na 3.71 ± 0.0005
increases when more oxalic acid is added. The latter two
Al 2.60 ± 0.005
observations both suggest that Sc precipitation selectivity may
Ca 6.06 ± 0.005
improve under more acidic conditions.
Sc 0.055 ± 0.008
Indeed, when performing analogous precipitation trials after
Y 0.363 ± 0.001
adjusting the pH of the solution by addition of H2SO4, both
La 0.317 ± 0.001
increased Sc selectivity and Sc recovery were observed. Ideal
Ce 0.406 ± 0.002
conditions were identified at pH 0 with 1 equiv (compared to
Pr 0.082 ± 0.001
the original Sc content in RM) of oxalic acid added (Figure 14,
Nd 0.078 ± 0.001
bottom right). The elemental analysis of the obtained
Sm ND
precipitate (by TXRF; see the SI) confirms that the only
Eu ND
metal present is Sc, while Ca, Al, Na, or even sulfate are not
Gd 0.032 ± 0.001
detected.
Tb ND
Roasting of the precipitate to obtain Sc2O3 was then
Dy 0.034 ± 0.0001
performed in a tube furnace in analogy to a literature
Ho 0.009 ± 0.0007
Er 0.011 ± 0.0006
procedure.48 This provided 1.0 mg Sc2O3, which had been
Tm 0.01 ± 0.0001
obtained from a combined weight of 10 g of RM (Figure 15).
Yb 0.012 ± 0.0005
Summarizing all optimized process steps provides the final
a process flow sheet (Figure 16), in which each step relies on
Conditions: (sulfation) 100 wt % Jamaican RM, 40 wt % H2O, 80 wt selective separations. This results in a highly efficient, 4-step
% conc H2SO4, aluminum crucible, 120 °C, 14 h, tube furnace;
(roasting) 700 °C, 1 h; (leaching) H2O (1.0 wt % solid residue, 10−50
separation of Sc as Sc2O3 from RM. The key to developing this
wt % H2O), room temperature, ball milling (IKA ULTRA-TURRAX rapid separation sequence is the optimization of each step and
Tube Drive, glass balls, 3000 rpm), 60 min. the separation of bulk elements by ball-mill accelerated leaching
of REs early on in the process. The overall recovery efficiency
of Sc is 75%, as calculated based on the original Sc content of
valuable, mixed RE oxide side product (mixed with Ca oxide). the used Jamaican RM and the obtained amount of Sc2O3.
Furthermore, Sc is completely separated in the precipitation Economic Analysis: Results of Probabilistic Assess-
step (see concentrations in leach liquor, Table 3). This ment. Once the flow sheet was finalized, we performed an
simplifies the separation of Sc in the next step, as only the bulk economic assessment of the developed process based on
elements Na, Ca, and Al have to be considered. Similarly to the statistical characterizations of economic performance outcomes
first highly selective leaching step, high selectivity in using Monte Carlo simulation techniques in the presence of
precipitation thus improves the overall efficiency of the process. multiple uncertainty sources (for all pertinent technical details
Selectivity in Sc Precipitation. Due to the Sc separation the reader is referred to the accompanying SI) . It should be
obtained in the precipitation step at pH 8, the remaining filtrate pointed out that one needs to be cognizant of the lack of
contains the vast majority of Sc leached from RM in addition to pertinent accumulated operating experience as well as the
1484 DOI: 10.1021/acssuschemeng.7b03968
ACS Sustainable Chem. Eng. 2018, 6, 1478−1488
ACS Sustainable Chemistry & Engineering
Figure 14. Optimization of oxalic acid precipitation of Sc. Conditions: pH < 8 adjusted by addition of H2SO4; addition of solid oxalic acid;
equivalents oxalic acid are calculated based on the original Sc content of RM. Average precipitation and standard deviations of three independent
experiments are shown.
Figure 15. Comparison of RM amount and Sc2O3 recovered from RM.
Figure 16. Updated process flowsheet for Sc recovery from RM.
inevitable paucity of reliable data that create significant process
economic performance evaluation challenges. However, the
proposed probabilistic approach is particularly suitable for
explicitly capturing inherent uncertainties (rather than con-
veniently ignoring them as in conventional single-point
economic performance assessment methods) associated with
the newly designed process49−52 and is also quite important in
1485
Research Article
the production of critical materials such as Sc53 due to the
underlying price volatility issues.
First, the development of baseline total capital investment
(TCI) and total product cost (TPC) models was pursued
following standard practices in process system economic
analysis.54 The above two quantities represent the two primary
components of a comprehensive cost analysis also known as
capital expenditure and operating expenses, respectively. In
particular, TCI was estimated as the sum of working capital
(WC) and fixed capital investment (FCI) encompassing both
direct and indirect costs. Furthermore, TPC is calculated by
forming the sum of production cost (PC) (such as operating
and maintenance costs) and general expenses (GE) (please see
pertinent tables in the SI).
Then, Monte Carlo simulation methods were integrated into
the aforementioned baseline cost models to derive full cost
distribution profiles in the presence of uncertainty associated
with commodity markets and the process system operating
environment. In particular, uncertain cost model input variables
were identified, following probability distribution profiles that
conform to ranges of reasonable empirical estimates50,54
reported in pertinent studies/reports (cited in the list of
bibliographical references). Furthermore, in the presence of
available historical data, the aforementioned distribution
profiles were derived using standard resampling methods as
commonly used in cost estimation studies.
Finally, it should be pointed out that it is the above
simultaneous inclusion of various sources of uncertainty as
multiple model input (random) variables that becomes feasible
through Monte Carlo techniques and differentiates the
proposed approach from the conventional sensitivity analysis
while effectively overcoming the “flaw of averages” (please see
pertinent tables in the SI for a detailed discussion).
After identifying all uncertain input variables, random
sampling from the distributions via a Monte Carlo simulator
was performed, allowing the model input uncertainties to
propagate through the cost models and generate probabilistic
DOI: 10.1021/acssuschemeng.7b03968
ACS Sustainable Chem. Eng. 2018, 6, 1478−1488
ACS Sustainable Chemistry & Engineering
Figure 17. Total product cost (TPC, left) and net present value (NPV, right) profiles for established Sc recovery process from RM. Profiles were
obtained through statistical characterizations of economic performance outcomes. Assumed production: 50 t Sc2O3/y.
cost distribution profiles that can be statistically characterized.
The derived profiles represent “risk−reward” profiles associated
with a range of economic performance outcomes where zones
of risks and opportunities can be identified and statistically/
probabilistically characterized. As a result, key economic
performance metrics such as the expected value of the specific
cost or the net present value, standard deviation, value at risk,
value at opportunity, and confidence intervals can be quantified
and insightfully characterized.
In the specific context of the established process, the
calculations assume the production of 50 t/y of Sc2O3, as this
value is one-third of the Sc projected to be consumed in 2023.7
Choosing this relatively low amount of Sc production is
expected to not significantly upset the Sc price movement
profile. Other key assumptions for the calculations are (i) a
product purity of 99.0% Sc2O3; (ii) a range of historical prices
(1991−2015) for Sc2O3 ($700−2000/kg); (iii) historical price
ranges for all employed chemicals and solvents (1991−2015;
see the SI); and (iv) an assumed plant life of 20 y.
A measure of the increase or decrease in wealth induced by
investment into a process system is the net present value
(NPV), as introduced in standard finance and investment
valuation theory. The NPV is a risk neutral valuation metric
used to assess economic performance over a project’s lifetime.
Within the proposed framework, a cumulative distribution NPV
profile was calculated as shown in Figure 17.
Notably, the NPV profile associated with the proposed
process remains positive, suggesting that the herein described
process development focusing on both selectivity and efficiency
yields a potentially economically viable process. The range of
NPV values between $0.18 billion and $0.27 billion further
suggest that it may be possible to drive remediation of Jamaican
RM simply by recovering Sc; if other REs are also recoverable
from the mixed RE oxide byproduct, the NPV profile may
become even more appealing. Overall, the economic analysis
performed validates the described process development
approach and its techno-economic performance prospects.
■
SUMMARY AND CONCLUSIONS
In summary, this manuscript describes the detailed develop-
ment of a cost-efficient, commercially viable process that allows
the recovery of many valuable REs, and Sc in particular, from
Jamaican RM. As the methodology of process development
focuses on selectivity for each separation step, viable process
conditions are established, which allow for Sc recovery in 75%
efficiency in only four steps. The outcome of a probabilistic,
Monte Carlo-based economic analysis that takes into account
uncertainties such as potential price fluctuations further
validates the prospects of the chosen approach and technology
1486
Research Article
option. As such, we report a viable route for Sc recovery from
RM, which may contribute to remediation of RM on large
scales and provide options for the treatment of RM produced
continuously by the aluminum industry. Furthermore, the
process system under consideration could potentially alleviate
challenges associated with likely future supply shortages for Sc,
in particular related to the projected increased use of high
strength aluminum alloys in aerospace and vehicle manufactur-
ing.
■
ASSOCIATED CONTENT
*
S Supporting Information
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acssusche-
meng.7b03968.
Analyses of red mud samples, process intermediates, and
final products; process optimization studies; detailed
assumptions used for Monte Carlo simulations; detailed
information on Monte Carlo analysis approach (PDF)
■
AUTHOR INFORMATION
Corresponding Author
*E-mail: mhemmert@wpi.edu. Phone: 508-831-4115. Fax 508-
831-4116.
ORCID
Marion H. Emmert: 0000-0003-4375-8295
Funding
This work was supported by WPI, NSF (IIP-0968839 and IIP-
1464560) and the industrial members of the NSF I/UCRC
Center for Resource Recovery and Recycling (CR3).
Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS
We thank all members of the Center for Resource Recovery
and Recycling (CR3) for valuable discussions. We thank Dr.
Shajesh Palantavida (Tufts University) for assistance with SEM
and EDX measurements.
■
ABBREVIATIONS
Al, aluminum; Ca, calcium; Ce, cerium; Fe, iron; La,
lanthanum; Na, sodium; Nd, neodymium; Pr, praseodymium;
RE, rare earth; RM, red mud; Sc, scandium; SI, Supporting
Information; t/y, tons per year; Ti, titanium; TXRF, total
reflection X-ray fluorescence; Y, yttrium
DOI: 10.1021/acssuschemeng.7b03968
ACS Sustainable Chem. Eng. 2018, 6, 1478−1488
ACS Sustainable Chemistry & Engineering
■ REFERENCES
(1) Han, A.; Ge, J.; Lei, Y. An adjustment in regulation policies and
its effects on market supply: Game analysis for China’s rare earths.
Resour. Policy 2015, 46, 30−42.
(2) Bradsher, K. China is said to halt exports to U.S. of some key
minerals, 2010. http://www.nytimes.com/2010/10/20/business/
global/20rare.html (accessed December 15, 2016).
(3) U.S. Geological Survey. Rare Earth ElementsCritical Resources
for High Technology. https://pubs.usgs.gov/fs/2002/fs087-02/ (ac-
cessed March 15, 2017).
(4) Hurst, C. China’s Rare Earth Elements Industry: What Can the
West Learn? http://www.ensec.org/index.php?option=com_
content&view=article&id=240:hurstrareearth&catid=
103:energysecurityissuecontent&Itemid=358 (accessed December 15,
2016).
(5) Duyvesteyn, W. P. C.; Putnam, G. F. SCANDIUM: A review of
the element, its characteristics, and current and emerging commercial
applications. http://www.emcmetals.com/i/pdf/Scandium-White-
PaperEMC-Website-June-2014-.pdf (accessed December 15, 2016).
(6) U.S. Geological Survey. Mineral commodity summaries. https://
minerals.usgs.gov/minerals/pubs/commodity/scandium/mcs-2016-
scand.pdf (accessed April 18, 2017).
(7) IHS Automotive Global Light Vehicle Production Summary.
https://www.ihs.com/ja/pdf/IHS-Automotive-Global-Summary-
Production-LVP-03-16_222719111115583632.pdf (accessed April 18,
2017).
(8) He, A. P.; Hu, Z. L.; Cao, D. G.; Zeng, J. M.; Wu, B. L.; Wang, L.
J. Extraction of Valuable Metals from Red Mud. Adv. Mater. Res. 2014,
881−883, 667−670.
(9) Power, G.; Gräfe, M.; Klauber, C. Bauxite residue issues: I.
Current management, disposal and storage practices. Hydrometallurgy
2011, 108 (1−2), 33−45.
(10) Evans, K. The History, Challenges, and New Developments in
the Management and Use of Bauxite Residue. J. Sustain. Metall. 2016,
2 (4), 316−331.
(11) Wagh, A. S.; Pinnock, W. R. Occurrence of scandium and rare
earth elements in Jamaican bauxite waste. Econ. Geol. Bull. Soc. Econ.
Geol. 1987, 82 (3), 757−761.
(12) Ochsenkü h n-Petropoulou, M. T.; Hatzilyberis, K. S.;
Mendrinos, L. N.; Salmas, C. E. Pilot-Plant Investigation of the
Leaching Process for the Recovery of Scandium from Red Mud. Ind.
Eng. Chem. Res. 2002, 41 (23), 5794−5801.
(13) Day, M. Hungary threatened by ‘ecological catastrophe’ as toxic
sludge escapes factory. http://www.telegraph.co.uk/news/worldnews/
europe/hungary/8043969/Hungary-threatened-by-ecological-
catastrophe-as-toxic-sludge-escapes-factory.html (accessed December
15, 2016).
(14) Jones, B. E. H.; Haynes, R. J. Bauxite Processing Residue: A
Critical Review of Its Formation, Properties, Storage, and
Revegetation. Crit. Rev. Environ. Sci. Technol. 2011, 41 (3), 271−315.
(15) Wang, W.; Pranolo, Y.; Cheng, C. Y. Recovery of scandium from
synthetic red mud leach solutions by solvent extraction with D2EHPA.
Sep. Purif. Technol. 2013, 108, 96−102.
(16) Ochsenkühn-Petropulu, M.; Lyberopulu, T.; Parissakis, G.
Selective separation and determination of scandium from yttrium and
lanthanides in red mud by a combined ion exchange/solvent extraction
method. Anal. Chim. Acta 1995, 315 (1), 231−237.
(17) Borra, C. R.; Blanpain, B.; Pontikes, Y.; Binnemans, K.; Van
Gerven, T. Comparative Analysis of Processes for Recovery of Rare
Earths from Bauxite Residue. JOM 2016, 68 (11), 2958−2962.
(18) Onghena, B.; Borra, C. R.; Van Gerven, T.; Binnemans, K.
Recovery of scandium from sulfation-roasted leachates of bauxite
residue by solvent extraction with the ionic liquid betainium
bis(trifluoromethylsulfonyl)imide. Sep. Purif. Technol. 2017, 176,
208−219.
(19) Borra, C. R.; Blanpain, B.; Pontikes, Y.; Binnemans, K.; Van
Gerven, T. Recovery of Rare Earths and Major Metals from Bauxite
Residue (Red Mud) by Alkali Roasting, Smelting, and Leaching. J.
Sustain. Metall. 2017, 3 (2), 393−404.
1487
Research Article
(20) Borra, C. R.; Mermans, J.; Blanpain, B.; Pontikes, Y.;
Binnemans, K.; Van Gerven, T. Selective recovery of rare earths
from bauxite residue by combination of sulfation, roasting and
leaching. Miner. Eng. 2016, 92, 151−159.
(21) Borra, C. R.; Blanpain, B.; Pontikes, Y.; Binnemans, K.; Van
Gerven, T. Recovery of Rare Earths and Other Valuable Metals From
Bauxite Residue (Red Mud): A Review. J. Sustain. Metall. 2016, 2 (4),
365−386.
(22) Schmitz, C. Handbook of aluminium recycling; Vulkan-Verlag
GmbH, 2006.
(23) Fulford, G. D.; Lever, G.; Sato, T. Alcan International Limited.
Recovery of rare earth elements from sulphurous acid solution by
solvent extraction. U.S. Patent 5,015,447, 1991.
(24) Government Successfully Concludes Rare Earth Pilot. http://
www.mstem.gov.jm/government-successfully-concludes-rare-earth-
pilot (accessed March 14, 2017).
(25) UC RUSAL to produce high-margin products with the use of
red mud cycle wastes. http://www.rusal.ru/en/press-center/press-
releases/1033/ (accessed March 14, 2017).
(26) Anastas, P. T.; Warner, J. C. Green Chemistry: Theory and
Practice; Oxford University Press: New York, 1998; p 30.
(27) Bandara, H. M. D.; Field, K. D.; Emmert, M. H. Rare earth
recovery from end-of-life motors employing green chemistry design
principles. Green Chem. 2016, 18 (3), 753−759.
(28) Krishnamurthy, N.; Gupta, C. K. Extractive Metallurgy of Rare
Earths, second ed.; CRC Press, 2015.
(29) Wang, W.; Pranolo, Y.; Cheng, C. Y. Recovery of scandium from
synthetic red mud leach solutions by solvent extraction with D2EHPA.
Sep. Purif. Technol. 2013, 108, 96−102.
(30) Borra, C. R.; et al. Smelting of Bauxite Residue (Red Mud) in
View of Iron and Selective Rare Earths Recovery. Journal of Sustainable
Metallurgy 2016, 2 (1), 28−37.
(31) Wang, W.; Pranolo, Y.; Cheng, C. Y. Recovery of scandium from
synthetic red mud leach solutions by solvent extraction with D2EHPA.
Sep. Purif. Technol. 2013, 108, 96−102.
(32) Ochsenkühn-Petropulu, M.; Lyberopulu, T.; Parissakis, G.
Direct determination of landthanides, yttrium and scandium in
bauxites and red mud from alumina production. Anal. Chim. Acta
1994, 296 (3), 305−313.
(33) Poston, J. A., Jr; Siriwardane, R. V.; Fisher, E. P.; Miltz, A. L.
Thermal decomposition of the rare earth sulfates of cerium(III),
cerium(IV), lanthanum(III) and samarium(III). Appl. Surf. Sci. 2003,
214 (1−4), 83−102.
(34) Nathans, M. W.; Wendlandt, W. W. The thermal decomposition
of the rare-earth sulphates. J. Inorg. Nucl. Chem. 1962, 24 (7), 869−
879.
(35) Paama, L.; Pitkänen, I.; Valkonen, J.; Pärnoja, E.; Kola, H.;
Perämäki, P. Thermal and spectroscopic investigation of europium and
samarium sulphates hydrates by TG-FTIR and ICP-MS techniques.
Talanta 2005, 67 (5), 897−902.
(36) Tagawa, H. Thermal decomposition temperatures of metal
sulfates. Thermochim. Acta 1984, 80 (1), 23−33.
(37) Stern, K. H. High temperature properties and thermal
decomposition of inorganic salts with oxyanions; CRC press, 2000.
(38) Li, J.-G.; Ikegami, T.; Mori, T. Fabrication of Transparent,
Sintered Sc2O3 Ceramics. J. Am. Ceram. Soc. 2005, 88 (4), 817−821.
(39) Bergmann, H. Sc, Y, La and Lanthanide. Gmelin Handbuch der
Anorganischen Chemie Seltenerdelemente; 1981; Teil C8, pp 62−182
(40) Kurtoğlu, S. F.; Soyer-Uzun, S.; Uzun, A. Tuning structural
characteristics of red mud by simple treatments. Ceram. Int. 2016, 42
(15), 17581−17593.
(41) Párez-Maqueda, L. A.; Franco, F.; Avilés, M. A.; Poyato, J.;
Pérez-Rodríguez, J. L. Effect of Sonication on Particle-size Distribution
in Natural Muscovite and Biotite. Clays Clay Miner. 2003, 51 (6),
701−708.
(42) Yuan, C.; Ma, H. Extraction of rare earth elements using alkyl
phosphinic acid or salt/tetraalkylammonium salt as extractant.
US5622679A, 1997.
DOI: 10.1021/acssuschemeng.7b03968
ACS Sustainable Chem. Eng. 2018, 6, 1478−1488
ACS Sustainable Chemistry & Engineering Research Article

(43) Shkol’nikov, E. V. Thermodynamic characterization of the

amphoterism of hydroxides and oxides of scandium subgroup elements
in aqueous media. Russ. J. Appl. Chem. 2009, 82 (12), 2098−2104.
(44) Gayer, K. H.; Thompson, L. C.; Zajicek, O. T. The solubility of
aluminum hydroxide in acidic and basic media at 25 °C. Can. J. Chem.
1958, 36 (9), 1268−1271.
(45) Yagmurlu, B.; Dittrich, C.; Friedrich, B. Precipitation Trends of
Scandium in Synthetic Red Mud Solutions with Different Precipitation
Agents. J. Sustain. Metall. 2017, 3 (1), 90−98.
(46) Zhang, W.; Koivula, R.; Wiikinkoski, E.; Xu, J.; Hietala, S.;
Lehto, J.; Harjula, R. Efficient and Selective Recovery of Trace
Scandium by Inorganic Titanium Phosphate Ion-Exchangers from
Leachates of Waste Bauxite Residue. ACS Sustainable Chem. Eng. 2017,
5 (4), 3103−3114.
(47) Yi, X. W.; Huang, C. H.; Wang, W.; Liu, Y .J.; Wu, J. G.
Scandium and Rare Earth Elements In Inorganic Chemistry Series;
Science Press: Beijing, 1992; Vol. 7, in Chinese.
(48) Kimura, E.; Nakanishi, H. Hexagonal-bipyramid crystalline
scandium oxide powder and a process for preparing the same.
US4980141 A, 1990.
(49) Glasserman, P. Monte Carlo Methods in Financial Engineering;
Springer: New York, 2003.
(50) de Neufville, R.; Scholtes, S. Flexibility in Engineering Design;
MIT Press: 2011.
(51) Ma, L.-C.; Castro-Dominguez, B.; Kazantzis, N. K.; Ma, Y. H. A
cost assessment study for a large-scale water gas shift catalytic
membrane reactor module in the presence of uncertainty. Sep. Purif.
Technol. 2016, 166, 205−212.
(52) Castro-Dominguez, B.; Ma, L.-C.; Mardilovich, I. P.; Kazantzis,
N. K.; Ma, Y. H. Integrated Experimental−Technoeconomic Analysis
of the Lifetime of Pd−Au Membranes. Ind. Eng. Chem. Res. 2016, 55
(38), 10160−10171.
(53) Bandara, H. M. D.; Mantell, M. A.; Darcy, J. W.; Emmert, M. H.
Rare Earth Recycling: Forecast of Recoverable Nd from Shredder
Scrap and Influence of Recycling Rates on Price Volatility. Journal of
Sustainable Metallurgy 2015, 1 (3), 179−188.
(54) Peters, M.; Timmerhaus, K.; West, R. Plant Design and
Economics for Chemical Engineers; McGraw-Hill Education, 2003.

1488 DOI: 10.1021/acssuschemeng.7b03968

ACS Sustainable Chem. Eng. 2018, 6, 1478−1488