Minerals Engineering 108 (2017) 71–82

Contents lists available at ScienceDirect

Minerals Engineering
journal homepage: www.elsevier.com/locate/mineng

Flocculation of oil sands tailings by hyperbranched functionalized

polyethylenes (HBfPE)
Linda Botha a, Stepheney Davey a, Benjamin Nguyen a, Anindya K. Swarnakar b, Eric Rivard b,
João B.P. Soares a,⇑
a
Department of Chemical and Materials Engineering, University of Alberta, Edmonton, Alberta, Canada
b
Department of Chemistry, University of Alberta, Edmonton, Alberta, Canada

a r t i c l e i n f o a b s t r a c t

Article history: We synthesized a series of hyperbranched functional polyethylene (HBfPE) flocculants with functional
Received 30 September 2016 group contents ranging from 30 and 50 mol%, and used them to flocculate oil sands mature fine tailings
Revised 6 January 2017 (MFT). MFTs (diluted to a concentration of 20 wt%) treated with the novel HBfPE had higher settling rates,
Accepted 11 February 2017
lower capillary suction time (CST), and lower turbidity than when treated with a reference polyacry-
Available online 21 February 2017
lamide (PAM). Within the HBfPE series, the settling performance improved when the functional group
content in the copolymer increased. When combined with centrifugation, the HBfPE flocculants gener-
Keywords:
ated sediments with higher solids contents than PAM within a shorter centrifugation time. This may
Oil sands tailings
Flocculation
be attributed to the tendency of HBfPE to form smaller flocs that compact more tightly than PAM. Floc
Dewatering development for HBfPE and PAM was compared with focused beam reflectance measurement (FBRM).
Functionalized polyethylene While floc formation with HBfPE and PAM was similar, it required significantly less mixing energy for
optimal performance, due to the significantly lower molecular weight (<10,000 g/mol) of HBfPE.
Ó 2017 Elsevier Ltd. All rights reserved.

1. Introduction
Alberta’s oil sands deposits are the third largest reserves of pet-
roleum in the world. They are unquestionably a vital source for
Canada’s energy needs. Unfortunately, the associated waste mate-
rial, in the form of fluid fine tailings (FFT) and mature fine tailings
(MFT) is a serious concern for the sustainability and environmental
responsibility of this industry (Kasperski and Mikula, 2011). Bitu-
men producers use different chemical and mechanical strategies
for the remediation of FFT, such as consolidated tailings (CT) or
non-segregating tailings (NST), thin lift dewatering, centrifugation,
rim ditching, and wet landscape reclamation (Kasperski and
Mikula, 2011; Masliyah et al., 2011). Neither of these individual
treatments or their combinations can currently dewater tailings
above the 75 wt% solids level required for land reclamation.
The slow consolidation and dewatering of oil sands tailings
originate from a combination of factors, such as the presence of
ultrafine clay particles (Mercier et al., 2012), interstratified clays
with swelling properties (Omotoso and Mikula, 2004; Omotoso
et al., 2006), medium pH, and process water composition
(Kotlyar et al., 1996; Fine Tailings Fundamentals Consortium
⇑ Corresponding author.
E-mail address: jsoares@ualberta.ca (J.B.P. Soares).
(FTFC), 1995; MacKinnon et al., 2001) often with a high concentra-
tion of Na+ ions. In addition, the clays present in MFT are partially
coated with residual bitumen and other organic materials that
limit the potential adsorption points for water soluble polymers
(Klein et al., 2013; Kotlyar et al., 1998; Majid and Sparks, 1996).
For example, Klein et al. showed that the progressive removal of
bitumen from tailings could improve the flocculation performance
of polyacrylamides (PAM), still the most commonly used polymer
flocculant in the oil sands industry (Klein et al., 2013).
Various types of PAM have been used as flocculants because
they have fast settling and initial water release rates. In thickener
applications, high molecular weight PAM is more effective as there
are statistically more units available in the polymer chains for
adsorption onto the clay surface. Longer chains can also span larger
distances between clay particles, flocculating them through a
bridging mechanism. Contrarily to the initial rapid water release
that occurs during the short term process of sedimentation, other
processes such as consolidation, creep, and desiccation of the floc-
culated sediments occur over a much longer time frame (Soleimani
et al., 2014). Therefore, good initial flocculation performance does
not necessarily translate into subsequent desirable long term
dewatering properties because the latter depends on the ability
of the flocculated sediment to release water trapped within the
sediment network. It therefore follows that polymers with some

http://dx.doi.org/10.1016/j.mineng.2017.02.004
0892-6875/Ó 2017 Elsevier Ltd. All rights reserved.
72 L. Botha et al. / Minerals Engineering 108 (2017) 71–82
hydrophobicity may be able to repel water trapped in the sedi-
ments, which would further aid in the dewatering of these
sediments.
Recent work on chemical modifications of PAM indicate that the
inclusion of hydrophobic groups in the PAM chains may help floc-
culate and dewater oil sands tailings, supposedly because the
hydrophobic groups repel water in the sediments and could also
interact with bitumen and organics present on the clay surfaces.
(Li et al., 2015; Reis et al., 2016). The structures of these modified
polymers are shown in Fig. 1. Poly-(n-isopropyl) acrylamide (PNI-
PAM) is temperature-sensitive and undergoes hydrophilic-to-
hydrophobic transition at solution temperatures above their lower
critical solution temperature (LCST). PNIPAM produces super-
natant water with lower turbidities, enhances the consolidation
of sediments, and improves the geotechnical properties of oil sands
tailings (Li et al., 2015). PNIPAM was used to treat a fresh
tailings stream between 40 and 50 °C to drive the hydrophilic-
hydrophobic thermal transition; however, application of PNIPAM
to MFT at ambient temperature is not considered to be economical,
as it would require additional energy input to heat up these large
volumes of tailings above the critical solution temperature of
PNIPAM (32 °C).
Other studies have focused on reducing the hydrophilicity of
PAM by grafting weakly polar polypropylene oxide chains onto
PAM backbones (Reis et al., 2016). This also improved the turbidity,
dewatering, and solids content (when combined with centrifuga-
tion) of the sediments, compared to a reference PAM flocculant.
Generally, PAM-PPO with shorter graft lengths performed better
than those with longer grafts. These lower molecular weight poly-
mers were also mush easier to dose and handle, compared to
PAM
PNIPAM
PAM-PPO
Fig. 1. Chemical structure of PAM and some modifications to reduce its
hydrophilicity.
higher molecular weight PAM flocculants; unfortunately, high
dosages (up to 10,000 ppm) were required for optimal perfor-
mance, which may make these flocculants less attractive from an
economical point of view.
These previous studies highlight the potential of hydrophobic
modification to improve flocculation and dewatering performance
of flocculants for oil sands tailings. These polymers, however, rely
on PAM backbones that inherently retain water through their
propensity to from hydrogen bonds. This pertains to the floccu-
lated sediments, where the concentration of polymer is expected
to be higher rather than for the low concentrations of polymer in
the actual process of flocculation.
Fig. 2A shows hydrogen bonding interactions between units of
different chains of an anionic hydrolyzed PAM, which are known
to swell in water due to these hydrogen-bond networks. As the
pH increases, some of the carboxylic acids of the anionic PAM are
deprotonated, reducing the number of hydrogen bond interactions;
however, the vast majority of the polymer (the acrylamide units),
still participates in hydrogen bonding (Garces et al., 1994). The
sum of these interactions increases with polymer molecular
weight: the higher molecular weight, the more likely is the poly-
mer to swell and retain water, which has a negative impact on
the long term dewatering and consolidation of the sediments.
We wanted to address this phenomenon by limiting the ability
of the polymer to form hydrogen bonds, while still working as an
efficient flocculant. In the present study, we used a purely
hydrophobic polymer backbone (polyethylene) functionalized
with charged functional groups to enable polymer dispersal in
water and interaction with the clay surfaces. This concept is illus-
trated in Fig. 2B.
Our group has previously synthesized partially-anionic slightly
hydrophobic hyperbranched functionalized polyethylenes (HBfPE)
with approximately 10–12 mol% functionality. Our proof of
A Anionic polyacrylamide
B Functionalized polyethylene
Fig. 2. Potential hydrogen bonding interactions (dashed lines) for an anionic
polyacrylamide (A) compared to a functionalized polyethylene (B).
 L. Botha et al. / Minerals Engineering 108 (2017) 71–82 73

principle experiments showed that MFT sediments flocculated characterized by 1H and 13C NMR in CDCl3 using a u5 VNMRS
with HBfPE retained less water and had lower capillary suction 500 MHz Spectrometer with a C13/H1 cold dual probe at
times (CST) than when a reference PAM flocculant was used. HBfPE 25–27 °C. Delays of 0.1 s were used and acquisition times of 3.4 s
flocculants also captured more fine particles, leading to super- for the 13C and 5.0 s for the 1H NMR. The spectral width was
natants with lower turbidity (Botha and Soares, 2016). These approximately 33,000 Hz for carbon and 5971 Hz for proton
HBfPE, unfortunately, had slower settling rates and formed less NMR. The MA content was calculated from the proton NMR spec-
compact sediment beds than when a reference PAM was used tra, using the intensity of the (AOCH3) resonance at 3.6 ppm as
under the same conditions, even though we expected to improve described by Popeney et al. (2007). The peak assignments of
these properties by increasing the fraction of functional groups Guironnet et al. (2009) were used for interpretation of the 13C
incorporated into the HBfPE chains. NMR spectra.
In this article we describe the synthesis of a new series of HBfPE
flocculants with functional group content varying from 30 to 2.2. HBfPE base hydrolysis
50 mol%, and compare their flocculation performance to that of a
commercial anionic PAM (Mw > 10,000,000 g/mol, approximately Approximately 2 g of each HBfPE grade was dissolved in 60 mL
30% anionic charges in the form of carboxylic acids) for MFT sam- of a 9:1 dichloromethane/methanol solvent mixture. After com-
ples diluted to solid contents of 5 wt% and 20 wt%. The commercial plete dissolution, 6 mL of 2 M KOH in methanol was added to the
anionic PAM is typically used as a reference polymer in flocculation dissolved polymer and the mixture allowed to hydrolyze for
studies of oil sands tailings (Li et al., 2015; Reis et al., 2016; 3 days. An equal volume of deionized water was added to the
Alamgir et al., 2012; Wang et al., 2010). The results are quite organic phase (dichloromethane) and the polymer was extracted
promising and indicate that a new class of flocculants for MFT by mixing into the aqueous phase as an opaque suspension. The
and other mineral tailings may be derived from our innovative two phases were allowed to separate and the organic phase dried
concept. to determine the concentration of the polymer in the aqueous
phase by mass balance. In all cases, the whole HBfPE sample was
extracted into the aqueous phase as a suspension (Fig. 3).
2. Experimental
2.3. Settling tests and flocculation performance
2.1. Synthesis and characterization of HBfPE
The MFT sample was obtained from Albian oil sands and was
The palladium catalyst, [(ArN@C(CH3)AC(CH3)@NAr)Pd(MA)] characterized by Dean-Stark extraction to determine bitumen,
BArF4 (Ar = 2,6-iPr2C6H3, MA = methyl acrylate), (refereed from now solids, and moisture content. A Malvern Mastersizer 3000 was used
on as the Pd catalyst) was synthesized according to a procedure pub- to determine the particle size distribution of the MFT samples.
lished in the literature (Johnson et al., 1996). All reactions involved Major ions isolated from the MFT process water through
in the synthesis of the Pd catalyst were performed using standard centrifugation (900g for 30 min) were determined by atomic
Schlenk line techniques under an atmosphere of nitrogen or in an absorption (AA).
inert atmosphere glove box (Innovative Technology, Inc.). Solvents Settling tests were done on 100 mL MFT solutions diluted to
were dried using a Grubbs-type solvent purification system either 5 wt% solids or 20 wt% with deionized water. For consis-
(Pangborn et al., 1996) manufactured by Innovative Technology, tency, all samples were dosed with 1000 ppm Ca2+ (from a 1 M
Inc., degassed (freeze-pump thaw method), and stored under an CaCl2 stock solution) and different volumes of polymer to obtain
atmosphere of nitrogen prior to use. NaBArF4 (ArF = 3,5-(F3C)2C6H3) a range of dosages (typically 250–4000 ppm or mg polymer per
was purchased from Matrix Chemicals and dried under vacuum at kg solids, ratioed to the solids content of the MFT). The optimal cal-
110 °C for 48 h. cium dosage of 1000 ppm (mg per kg solids) was determined for
The Pd catalyst was used to make the HBfPE. A 300 mL Parr
reactor was heated up to 120 °C, rinsed with dry toluene and
nitrogen-vented/vacuum-purged five times before use. A range of
methyl acrylate (MA) comonomer concentrations (0–6 M) in a total
volume of 200 mL dichloromethane (DCM) was added to the
reactor to make HBfPE with different comonomer contents. For
the HBfPE series with higher MA contents, varying amounts of tri-
isobutyl aluminium (TIBAL) ranging from 0 g to 0.6 g was intro-
duced with a fixed concentration of MA (2.9 M) to protect the
polar functionality and for increased incorporation of the comono-
mer. The reactor temperature was kept constant at 35 °C and the
reactor was charged with ethylene at 2 atm pressure. The Pd
catalyst (0.2 mmol) was injected as soon as a stable ethylene
pressure was obtained and the polymerization was allowed to con-
tinue for 18 h before discharging the reactor contents in methanol.
The HBfPEs were insoluble in methanol, either in the form a vis-
cous oil (at 10 mol% MA content), which phase separated, or a
solid (at >30 mol% MA content). The polymer was washed with
methanol until the methanol solution was clear. The methanol
phase was either decanted or removed by filtration, depending
on the physical state of the polymer, and the polymer was dried
under vacuum at 60 °C. Polymer yields were typically between
10 and 15 g.
The HBfPE products were characterized by FTIR and HT-HPLC
using trichlorobenzene (TCB) as solvent. The HBfPE were also Fig. 3. HBfPE suspensions in water - 5 mg/mL (left) and 1 mg/mL (right)
74 L. Botha et al. / Minerals Engineering 108 (2017) 71–82

this specific MFT sample. Polymer stock solutions of 5 mg/mL were scooping an equivalent portion of the sediment with a spatula.
prepared by dissolving the polymer in deionized water, while mix- The supernatant turbidities were measured after 24 h of settling,
ing for at least two days to ensure complete dissolution and homo- using a Hach 2100 AN turbidimeter. The sediments were trans-
geneity. Blank tests were done where calcium and the maximum ferred to pre-weighed drying pans, dried in a low temperature
volume of base hydrolysis solution (without polymer) was added oven (60 °C) until constant weight and the solids content
to exclude any effects from the ions introduced during base hydrol- calculated.
ysis. Atomic absorption (AA) comparison of the calcium contents of
the process water and supernatants after flocculation, indicated 3. Results and discussion
that the polymers consume the calcium during flocculation.
A Heidolph Hei-Torque Precision 200 overhead mixer with a 3.1. Polymer synthesis
4-blade, pitched blade impeller (diameter 5 cm) was used to opti-
mize the mixing parameters for each polymer. With this system, Table 1 summarizes the polymerization conditions and proper-
the suspension torque can be followed in real-time. The mixing ties of the new HBfPE series (HBfPE-35, HBfPE-40, and HBfPE-50),
vessel dimensions were as follows: 6.5 cm diameter, 5 mm and the HBfPE produced without TIBAL (HBfPE-12). The addition
distance between the vessel floor and the impeller. The optimal of TIBAL significantly improved the incorporation of methyl acry-
mixing protocol for the HBfPE was a lower shear regime late (MA), as observed in the proton NMR spectra of HBfPE-35,
(300 rpm for 10 s). For the reference PAM however, this low shear HBfPE-40, and HBfPE-50 (Fig. 4). The frequency of adjacent MA
mixing protocol was not optimal and the original mixing condi- units (represented by the resonances at 1.5, 1.7 and 1.9 ppm)
tions of 650 rpm for 2 min, followed by slow mixing (230 rpm) increases in the spectra as the total MA content of the polymer
for 8 min was used as described by Reis et al. (2016). increases. In addition, the presence of 13C NMR resonances repre-
We tested the effect of using dilute (1 mg/mL) polymer solu- senting adjacent MA and ethylene units, e.g. >1 ethylene attached
tions, compared to the original 5 mg/mL solutions. The existing to a MA unit (ad+A at 27 ppm and bd+A at 32 ppm) or 1 ethylene
mixing protocol was however not optimal for the dilute solutions, inserted between 2 MA units (bbA at 25 ppm and acA at
specifically the reference PAM. In this case, the impeller position 33 ppm), confirm that the HBfPE series are true copolymers and
had to be changed to ensure proper mixing (due to the low volume not mixtures of the two homopolymers (Fig. 5).
of MFT, required to maintain the same overall solids content). The The new polymerization method makes copolymers not only
same trends for settling rate and CST were observed for optimal with increased functionality, but also with higher molecular
mixing of the dilute (1 mg/mL solutions) as with the 5 mg/mL solu- weights (Table 1). Since the molecular weight determination is
tions. More dilute polymer solutions were not tested, as this would based on hydrodynamic volume, the branched structure will result
lower the solids content of the whole suspension below 20 wt%.
During the optimization of the flocculant dosing process, we
also investigated the effect of the sequence of calcium and polymer
addition to the tailings. The HBfPE performed optimally (signifi-
cantly lower CST) when the calcium was added simultaneously
or a few seconds after the polymer was added. For the reference
PAM, calcium had to be added before the polymer for optimal per-
formance. The same protocols were used in preparation for the
centrifugation tests, but undiluted tailings was used and after mix-
ing, the flocculated MFT was transferred to 50 mL corning tubes
and centrifuged at 900g for a total of 140 min. Samples were
removed after 2, 8, 32 and 64 min to determine the supernatant
turbidity and sediment solids contents. The solids contents of each
sample after polymer and calcium addition was 27 wt%.
Capillary suction times (CST) were determined with a Triton
319 multi-purpose CST instrument, using the appropriate Triton
CST filter paper (7  9 cm). Three to six repeat CST measurements
were made per sample, to minimize the variation (coefficient of
variance <10%). The CST values were measured prior to settling
to ensure comparable water contents and on the sediments after
24 h. After 24 h of settling, the top supernatant and bottom sedi-
ments were separated and the interface discarded. A pipette with
a wide bore opening (3 mm) was used for this. In cases where
large flocs were obtained, the pipette tip was sawed off to provide Fig. 4. 1H NMR spectra in CDCl3 of the HBfPE series synthesized with TIBAL (HBfPE-
a wider opening, to prevent shearing of the flocs. Pipetting a fixed 35, HBfPE-40, and HBfPE-50), compared to HBfPE-12 synthesized without TIBAL.
volume of suspension gave more reproducible results than The asterix (*) indicates residual solvent signals for diethyl ether, toluene, and THF.

Table 1
HBfPE polymerization conditions and polymer properties.

Polymerb TIBAL (mmol) Ethylene pressure (atm) Methyl acrylate feed (M) mol% MA incorporated Mw (g/mol)a
HBfPE-12 0 2 2.9 12 3000
HBfPE-40 0.15 2 2.9 40 N/A
HBfPE-50 0.30 2 2.9 50 7000
HBfPE-35 0.60 2 2.9 35 N/A
a
Mw analysis done on HBfPE before hydrolysis.
b 13
Samples are labeled HBfPE-##, where ## is the mol% of MA incorporated in the copolymer, as measured by C NMR.
 L. Botha et al. / Minerals Engineering 108 (2017) 71–82 75

13
Fig. 5. C NMR spectra in CDCl3 of HBfPE and a polymethylacrylate (PMA) made by free-radical polymerization.

in an underestimation of the actual molecular weight of the HBfPE.
The molecular weight is calculated based on the Mark-Houwink
relationship with intrinsic viscosity for linear polyethylene. Conse-
quently, the molecular weight values reported in Table 1 are
expected to be lower than the absolute molecular weights. Since
all HBfPE were produced at the same ethylene pressure, they are
expected to have the same branching frequency. In addition, all
HBfPE have resonances associated with branch formation by chain
walking polymerization, such as sec-butyl branches, represented
by various smaller resonances between 20 and 34 ppm (Cotts
et al., 2000). The relative intensities of the methyl end group
resonances to that of the methylenes (dd+A) in the 13C NMR spectra
further confirms that HBfPE polymers produced with TIBAL had
higher molecular weights than HBfPE-12 made without TIBAL.
3.2. MFT sample composition
the mixture, being present in concentrations at least 10 times
higher than those of K+, Ca2+, Mg2+. The MFT sample is composed
of predominantly fine particles, 90% of which are smaller than
1 lm (measured by the light scattering number average). The
MFT sample was diluted with deionized water (as indicated in
Table 2) to obtain samples with solids contents of 5 wt% or 20 wt
%. We used these diluted samples in the flocculation tests dis-
cussed below.
3.3. Settling tests for MFT diluted to 5 wt% and 20 wt% solids
In a previous publication (Botha and Soares, 2016), we observed
that the settling rate increased fivefold when the functional
content of the HBfPE increased from 12 mol% to 50 mol%. For the
current series of tests, Fig. 6 and Table 3 shows that settling rates
of the 5 wt% MFT sample increase as the functional group content

Table 2 lists composition of the MFT sample used in all floccu-

lation tests. Typical for MFT, sodium is the most abundant ion in

Table 2
MFT sample composition.

Characterization Original Diluted sample Diluted sample

technique sample (5 wt% solids) (20 wt% solids)
Dean stark Bitumen 2.93 0.36 1.45
analysis (wt%) Water 54.10 94.65 78.58
Solids 40.27 4.99 19.97
Major ions (AA) Na+ 267.2 33.1 132.5
(ppm) K+ 20.8 2.6 10.3
Ca2+ 22.9 2.8 11.4
Mg2+ 12.8 1.6 6.3
Particle size distribution (mm)
Volume average d10 1.48 1.48 1.48
d50 10.20 10.20 10.20
d90 59.50 59.50 59.50
Number average d10 0.37 0.37 0.37
d50 0.53 0.53 0.53
d90 0.99 0.99 0.99
pH 7.8 7.0 7.6 Fig. 6. Settling rates obtained for 5 wt% solids MFT treated with 2000 ppm HBfPE
and PAM.
76 L. Botha et al. / Minerals Engineering 108 (2017) 71–82

Table 3 the improved performance of the HBfPE series relative to PAM at

Polymer performance in 5 wt% MFT. 20 wt% solids to the fact that the lower molecular weight HBfPE
Polymer Initial settling rate (mm/s) Final mudline (mL) chains face smaller distances between clay particles when the solid
HBfPE-35 0.054 30 content is higher, enabling them to flocculate the MFT particles
HBfPE-40 0.066 28 more efficiently than when the solid content is lower. Within the
HBfPE-50 0.083 28 HBfPE series, HBfPE-40 and HBfPE-50 had similar, but slightly fas-
PAM 2.370 17 ter, settling rates than HBfPE-35, which highlights the importance
of the MA functional groups in the flocculation efficiency of these
polymers. From Fig. 7, it is also evident that the HBfPE polymer
generated more compact sediments than PAM.
Fig. 8 compares the percentage water released after 24 h of set-
tling, based on the volume of the final mudlines. The higher the
dosage of PAM, the more water was released from the flocculated
MFT, but high PAM dosages are detrimental to other flocculation
measures, as we will explain in the next section. Contrarily, the
HBfPE flocculants released more water than PAM, and performed
significantly better at 250–1000 ppm dosages. At 4000 ppm,
HBfPE-50 released the most water of all tested flocculants; more
importantly, we saw no evidence of overdosing at this dosage, as
will be explained in the next section.

3.4. Sediment dewaterability

Capillary suction time (CST) measurements were used to quan-

Fig. 7. Settling rates obtained for 20 wt% solids MFT treated by 1000 ppm and
2000 ppm of the HBfPE series and PAM.
increases from 35 to 50 mol%. Fig. 6 compares settling rates only
for the 2000 ppm HBfPE dosage, but the other dosages followed
similar trends. Though most settling happened in less than
20 min, the initial settling rates were still significantly lower than
those for the reference PAM. The reference PAM also generated
more compact mudlines.
When we repeated settling tests for the 20 wt% MFT sample, the
HBfPE flocculants surprisingly showed higher settling rates than
PAM within the lower dosage range of 250–1000 ppm. Fig. 7 com-
pares settling rates for 1000 ppm and 2000 ppm dosages. The
HBfPE flocculants outperformed PAM for 1000 ppm dosages, but
all settling rates were similar for 2000 ppm dosages. We attribute
tify the influence of these flocculants on the dewaterability of the
MFT. CST was lower for HBfPE than for PAM when treating 5 wt%
MFT (Fig. 9A), although for some dosages, such as 1000 ppm, these
differences do not seem to be statistically significant. The CST gen-
erally increased for higher flocculant dosages, but this effect was
more pronounced for PAM, which increased from 20 s (250 ppm)
to 156 s (4000 ppm), more than doubling the CST value of the
blank test, when no flocculant was added to the MFT. At this high
dosage, PAM actually performed worse than pure MFT; the sedi-
ment appeared gel-like, with visible localized patches of polymer,
which indicates overdosing. On the other hand, we did not observe
any signs of overdosing for the HBfPE polymers from 250 to
2000 ppm, but CSTs increased slightly for HBfPE-35 and HBfPE-40
at the highest dosage of 4000 ppm. The CST continued to decrease
with dosage all the way to 4000 ppm for HBfPE-50, although the
differences were minor and likely not statistically significant. The
sediment CST measurements after 24 h presented similar trends
(Fig. 9B), but the interpretation of these measurements is more
complex, as differences in water content between the different
sediments may influence the CST measurements.
Higher CSTs were measured for 20 wt% MFT (Fig. 10). HBfPE-40
performed best at 1000 ppm, and HBfPE-50 at 2000 ppm, both out-
performing PAM slightly (Fig. 10A). The HBfPE flocculants also
obtained lower CST values within a wider range of dosages.
HBfPE-35 had the highest CST values in the HBfPE series, but they
were still better than that of the reference PAM for most dosages.
These results show that HBfPE with higher functional group
content dewater MFT more efficiently. Similarly, the CST of the
sediments after 24 h were consistently lower for HBfPEs than for
PAM (Fig. 10B).
3.5. Supernatant turbidity
Fig. 11 shows that the supernatant turbidities for 5 wt% MFT
treated with the HBfPE where lower than for PAM. HBfPE-50 con-
sistently gave supernatants with the lowest turbidity; interest-
ingly, turbidity did not depend on polymer dosage. Contrarily,
the turbidity increased when the dosage of HBfPE-35 increased,
suggesting that the functional content of this polymer was not high
enough to allow it to interact with, and subsequently precipitate all
fine clay particles in the MFT. The optimum performances for
 L. Botha et al. / Minerals Engineering 108 (2017) 71–82 77

Fig. 8. Water released after 24 h of settling for 20 wt% solids MFT.

Fig. 9. Capillary suction times for 5 wt% MFT treated with different dosages of HBfPE and PAM obtained immediately after dosing (A) and after 24 h (B).

HBfPE-40 and PAM were observed at the intermediate dosages of
500 and 1000 ppm, respectively. For the 20 wt% MFT sample, the
supernatants after all the treatments had low turbidities (<20
NTU) at a dosage of 2000 ppm. For lower dosages (250, 500 and
1000 ppm), the volume of supernatant was not enough to allow
for turbidity measurements.
3.6. Sediment solids content
The solids contents of the sediments were determined after
24 h of settling. Fig. 12 shows that for the 5 wt% MFT sample,
higher solids contents were achieved for PAM than for the HBfPE
flocculants. This is to be expected, considering the more compact
78 L. Botha et al. / Minerals Engineering 108 (2017) 71–82

Fig. 10. Capillary suction times for 20 wt% MFT treated with different dosages of HBfPE and PAM obtained immediately after dosing (A) and after 24 h (B).

Fig. 11. Turbidities for MFT diluted to 5 wt% solids treated with different dosages of HBfPE and PAM.
 L. Botha et al. / Minerals Engineering 108 (2017) 71–82 79

Fig. 12. Sediment solids content for 5 wt% solids MFT treated with different dosages of HBfPE and PAM.

Fig. 13. Sediment solids content for 20 wt% solids MFT treated with different dosages of HBfPE and PAM.

Fig. 14. Mudlines after 24 h of settling for HBfPE-50 (A) and PAM (B). Dosages increase from left to right.
80 L. Botha et al. / Minerals Engineering 108 (2017) 71–82

final mudlines achieved with PAM, as shown in Table 3. In addition,

Fig. 15. Cake solids content obtained after centrifugation of 27 wt% MFT treated
with different dosages of HBfPE and PAM.
lower PAM dosages resulted in slightly higher solids content.
Within the HBfPE series, HBfPE-50 obtained the highest solids con-
tents. The solids content, however, decreased with dosage for
HBfPE-35, and no clear influence of dosage on the resulting solids
content was observed for HBfPE-40.
Considering the differences in the mudlines and water released
between the HBfPE and PAM for 20 wt% solids MFT (Fig. 7), we
expected higher solids contents for the HBfPE flocculants. Although
Fig. 13 shows that this was the case for dosages between 250 and
1000 ppm, the solids contents obtained with HBfPE flocculants
were similar to PAM’s, or slightly lower, for higher dosages.
The fact that the differences in the solids content were smaller
than anticipated based on the mudlines may be attributed to dif-
ferences in floc structure and packing between HBfPE and PAM.
Fig. 14 compares floc size and packing for HBfPE-50 and PAM,
showing that the flocs formed with HBfPE are smaller and more
tightly packed than those formed with PAM. Because PAM forms
a larger volume of sediments, one may underestimate the solids
contents based on mudline level, as water is still present among
the flocs.

Fig. 16. Mean square weight chord lengths of flocs formed after addition of PAM (A) or HBfPE-40 (B).
 L. Botha et al. / Minerals Engineering 108 (2017) 71–82 81

The solids content measurements, however, assume no con-

straints in the drying process, as these are done on small laboratory
scale samples using controlled conditions. Therefore, in addition to
the solids content, the CST provides complementary information
on the ease of removal of the water trapped in the sediments. As
discussed in the previous section, the HBfPE are easier to dewater
and may, therefore, attain higher practical solids contents than
PAM within a short drying period.

3.7. Centrifugation

Industrial centrifuges can improve the dewatering process by

means of shear in addition to the centrifugal forces. Since HBfPE
flocculants make smaller flocs than PAM, centrifugation may force
tighter packing of the sediments, speed up settling rates, and
release more water. Fig. 15 compares solids contents of the cen-
trifuge cake determined after various settling intervals, using two
different polymer dosages. HBfPE-40 generated the highest solids
content (approximately 60 wt%) after as little as 2 min of centrifu-
gation. The other flocculants led to slightly lower solids contents
under the same centrifugation time, with PAM’s being the lowest
of all tested polymers for a centrifugation time of 2 min. In fact,
PAM needed 140 min of centrifugation time to reach solids con-
tents comparable to those of HBfPE. Considering that centrifuga-
tion is an energy-intensive method, short times are highly
desirable. The turbidities of all the supernatant samples were low
(<20 NTU) for all centrifugation times.

3.8. Floc development

The flocculation performance of the HBfPE is surprising, consid-

ering that these polymers have lower molecular volumes than that
of the reference PAM. As seen above, HBfPE performed better for
20 wt% MFT samples, likely because the chains had to bridge smal-
ler distances between the suspended clay particles, but Fig. 14
shows that the flocs produced by HBfPE were visibly smaller than
those by PAM. The HBfPE flocculants also required short mixing
times (10 s at 300 rpm) because their solutions had lower viscosi-
ties than PAM solutions, which had to be initially stirred at
600 rpm, followed by an additional low speed (230 rpm) mixing
step to release more water between the flocs.
To better understand floc formation with these two very differ-
ent classes of flocculants, we used a focused beam reflectance mea-
surement (FBRM) system to track the development of flocs after
HBfPE or PAM addition. We used the mean square weighted chord
length to quantify the floc formation process, as it is more sensitive
towards larger flocculated particles compared to unflocculated
fines. Since the FBRM profiles obtained for the different HBfPE were
similar, we will only discuss the experiments with HBfPE-40. Fig. 17. Mean square weighted counts for MFT treated with HBfPE and PAM.
Fig. 16 compares dynamic changes in the mean square
weighted chord lengths with time for PAM and HBfPE-40. Both
flocculants have similar baseline chord lengths of approximately Fig. 16, a second stage of mixing at the same speed re-suspended
25 lm. A sharp increase in the chord length is observed after poly- the flocs, which retained their original sizes, as can also be seen
mer addition, with PAM forming flocs with larger chord lengths in the second plateau region in Fig. 16.
(250 lm) than HBfPE-40 (170 lm). The lag time between the addi- Fig. 17 compares the distributions of the mean square weighted
tion of PAM and floc formation is higher than for HBfPE-40, which chord length for MFT before treatment (t0), the first plateau after
reflects the need for greater mixing requirements for PAM. Con- polymer addition (t1), and the second plateau after one cycle of set-
versely, the lower mixing energy requirement of HBfPE-40 is also tling and re-suspension (t2). As expected, the t0 baseline distribu-
observed through a slight decrease in floc size to 150 lm during tions are similar for all polymers, but PAM forms larger flocs
longer mixing times, whereas the floc size of PAM increases with the broadest distribution ranging from 100 to 1000 lm. The
slightly, but continuously throughout the whole mixing period. chord size distribution remains the practically the same for all
When mixing was stopped, the size of flocs formed with HBfPE polymers after one cycle of settling and re-suspension.
increased sharply to 230 lm before settling. This observation indi- HBfPE-35 has a significant portion of small particles, or flocs,
cates a negative overmixing effect on HBfPE performance; luckily, between 10 and 100 lm, likely because it has the lower MA con-
this effect appears to be reversible, as larger flocs are formed as tent of all HBfPE flocculants. This measurement agrees with the
soon as mixing is interrupted. After the settling period shown in previous observations for turbidity and settling rates for this
82 L. Botha et al. / Minerals Engineering 108 (2017) 71–82
polymer. The population of these smaller particles, however, is
reduced after the second mixing step, shifting the distribution to
larger floc sizes.
Among the HBfPE series, narrower floc distributions are
obtained as the MA fraction increases, but HBfPE-40, with interme-
diate functional group content, has the largest overall floc sizes in
the series and an intermediate floc size distribution. This polymer
also produced the highest solids contents after centrifugation.
Therefore, this specific combination of average floc size and distri-
butions seems to favour compaction under centrifugal forces. On
the other hand, HBfPE-50 had the most uniform distribution of floc
sizes, and achieved best performance in the HBfPE series for dewa-
tering (as measured by CST), and turbidity.
4. Conclusions
We synthesized a series of novel hyperbranched functional
polyethylenes (HBfPE) flocculants consisting of a hydrophobic
backbone containing anionic (carboxylic acid) functional groups
ranging from 35 to 50 mol%. These polymers have significantly
lower molecular volumes and viscosities than typical PAM floccu-
lants used in the industry, and require less mixing energy to be
adequately dispersed in mature fine tailings (MFT). The HBfPE per-
formed better for 20 wt% than 5 wt% MFT, obtaining lower CSTs
and turbidities, faster settling rates, and more compact sediments
than a reference PAM. Based on the CST measurements, the sedi-
ments generated by HBfPE are expected to dewater faster than
those formed with PAM. Tailings flocculated with HBfPE also had
higher solids contents after centrifugation than those treated with
PAM. The HBfPE could also be dosed over a wider optimum range,
and were not prone to overdosing, which improves the practical
application of these polymers, since dosing mistakes will not lead
necessarily to deleterious effects. Within the HBfPE series, poly-
mers with higher functional contents had better CSTs and turbidi-
ties, which could be related to the formation of more uniform floc
sizes.
The fact that optimal HBfPE performance was observed when
calcium was added in sequence after the polymer, instead of
before, suggests the occurrence of a hydrophobic transition caused
by the interaction between the functional groups in HBfPE and
calcium. Since HBfPE with high functional contents have many
adjacent methyl acrylate (MA) units, Ca2+ ions may chelate the car-
boxylic acids of adjacent groups, forcing the polymer to precipitate.
This hydrophilic-hydrophobic transition does not depend on
temperature or pH change, and uses a chemical that is already
conventionally combined with polymer treatments for MFT.
These are promising initial results from a new class of floccu-
lants. Future work from our group will include scaling up the
polymer synthesis to treat larger MFT volumes and evaluate their
saturated hydraulic conductivities and long-term dewatering
performance.
Acknowledgements
The authors would like to thank Mark Mizolskie and Nupur
Dabral (Department of Chemistry, University of Alberta, Edmonton,
Alberta, Canada) for the NMR runs. Eric Rivard thanks NSERC of
Canada for a Discovery Grant and the Canada Foundation for
Innovation for support. João Soares thank the Campus Alberta
Innovates Program (CAIP) and the Canada Foundation for Innova-
tion for their kind support.
References
Alamgir, A., Harbottle, D., Masliyah, J., Xu, Z., 2012. Chem. Eng. Sci. 80, 91–99.
Botha, L., Soares, J.B.P., 2016. Novel hyperbranched functionalized polyethylenes for
the treatment of oil sands tailings. In: 19th International Seminar on Paste and
Thickened Tailings, 5–8 July, Santiago, Chile.
Cotts, P.M., Guan, Z., McCord, E., McLain, S., 2000. Macromolecules 33, 6945–6952.
Fine Tailings Fundamentals Consortium (FTFC), 1995. Advances in Oil Sands Tailings
Research. Alberta Department of Energy, Edmonton, AB.
Garces, F.O., Sivadasan, K., Somasundaran, P., Turro, N.J., 1994. Macromolecules 27,
272–278.
Guironnet, D., Roesle, P., Rünzi, T., Göttker-Schnetmann, I., Mecking, S., 2009. J. Am.
Chem. Soc. 131, 422–423.
Johnson, L.K., Mecking, S., Brookhart, M., 1996. J. Am. Chem. Soc. 118, 267–268.
Kasperski, K.L., Mikula, R.J., 2011. Elements 7, 387–392.
Klein, C., Harbottle, D., Alagha, L., Xu, Z., 2013. Can. J. Chem. Eng. 91, 1427–1432.
Kotlyar, L.S., Sparks, B.D., Schutte, R., 1996. Clays Clay Miner. 44 (1), 121–131.
Kotlyar, L.S., Sparks, B.D., Woods, J.R., Raymond, S., Le Page, Y., Shelfantook, W.,
1998. Pet. Sci. Technol. 16, 1–19.
Li, H., Zhou, J., Chow, R., Adegoroye, A., Najafi, A.S., 2015. Can. J. Chem. Eng. 93,
1780–1786.
MacKinnon, M.D., Matthews, J.G., Shaw, W.H., Cuddy, R.G., 2001. Int. J. Surf. Min.
Reclamat. Environ. 15 (4), 235–256.
Majid, A., Sparks, B.D., 1996. Fuel 75 (7), 879–884.
Masliyah, J.H., Czarnecki, J.A., Xu, Z., 2011. Handbook on Theory and Practice of
Bitumen Recovery from Athabasca Oil Sands: Theoretical Basis, vol. 1. Kingsley
Knowledge, Cochrane, AB.
Mercier, P.H.J., Ng, S., Moran, K., Sparks, B.D., Kingston, D., Kotlyar, L.S., Kung, J.,
Woods, J., Patarachao, B., McCracken, T., 2012. Pet. Sci. Technol. 30 (9), 915–923.
Omotoso, O.E., Mikula, R.J., 2004. Appl. Clay Sci. 25, 37–47.
Omotoso, O., Mikula, R., Urquhart, S., Sulimma, H., Stephens, P., 2006. Clay Sci. 12
(2), 88–93.
Pangborn, A.B., Giardello, M.A., Grubbs, R.H., Rosen, R.K., Timmers, F.J., 1996. J. Am.
Chem. Soc. 15, 1518–1520.
Popeney, C.S., Camacho, D.H., Guan, Z., 2007. J. Am. Chem. Soc. 129, 10062–10063.
Reis, L.G., Oliveira, R.S., Palhares, T.N., Spinelli, L.S., Lucas, E.F., Vedoy, D.R.L., Asare,
E., Soares, J.B.P., 2016. Miner. Eng. 95, 29–39.
Soleimani, S., Simms, P., Dunmola, A., Freeman, G., Wilson, G.W., 2014. Dessication
and consolidation in thin-lift deposition of polymer-amended mature fine
tailings. In: 17th International Seminar on Paste and Thickened Tailings, 8–12
June, Vancouver, Canada.
Wang, X., Feng, X., Xu, Z., Masliyah, J., 2010. Can. J. Chem. Eng. 88, 403–410.