Liquid coordination complexes: A new class of Lewis acids as safer alternatives to BF3 in synthesis of polyalphaolefins

Article abstract of DOI:10.1039/c4gc02080d

Liquid coordination complexes (LCCs) are a new class of liquid Lewis acids, prepared by combining an excess of a metal halide (e.g. GaCl₃) with a basic donor molecule (e.g. amides, amines or phosphines). LCCs were used to catalyse oligomerisation of 1-decene to polyalphaolefins (PAOs). Molecular weight distribution and physical properties of the produced oils were compliant with those required for low viscosity synthetic (Group IV) lubricant base oils. Kinematic viscosities at 100 °C of ca. 4 or 6 cSt were obtained, along with viscosity indexes above 120 and pour points below-57 °C. In industry, to achieve similar properties, BF₃ gas is used as a catalyst. LCCs are proposed as a safer and economically attractive alternative to BF₃ gas for the production of polyalphaolefins. This journal is

Full text of DOI:10.1039/c4gc02080d

Green Chemistry
View Article Online
View Journal
PAPER
Liquid coordination complexes: a new class
of Lewis acids as safer alternatives to BF3 in
Cite this: DOI: 10.1039/c4gc02080d
synthesis of polyalphaolefins
James M. Hogg, Fergal Coleman, Albert Ferrer-Ugalde, Martin P. Atkins and
Małgorzata Swadźba-Kwaśny*
Liquid coordination complexes (LCCs) are a new class of liquid Lewis acids, prepared by combining an
excess of a metal halide (e.g. GaCl3) with a basic donor molecule (e.g. amides, amines or phosphines).
LCCs were used to catalyse oligomerisation of 1-decene to polyalphaolefins (PAOs). Molecular weight dis-
tribution and physical properties of the produced oils were compliant with those required for low viscosity
synthetic (Group IV) lubricant base oils. Kinematic viscosities at 100 °C of ca. 4 or 6 cSt were obtained,
along with viscosity indexes above 120 and pour points below −57 °C. In industry, to achieve similar pro-
perties, BF3 gas is used as a catalyst. LCCs are proposed as a safer and economically attractive alternative
to BF3 gas for the production of polyalphaolefins.
Received 26th October 2014,
Accepted 14th January 2015
DOI: 10.1039/c4gc02080d
www.rsc.org/greenchem
thesis, followed by fractionated distillation, blending and for-
mulation with additives.
Introduction
Polyalphaolefins
The catalytic syntheses of PAOs have been expertly dis-
Polyalphaolefins (PAOs) are oligomers of higher α-olefins, pre- cussed by Ray et al.4 Boron trifluoride, promoted with water,
dominantly of 1-decene. They are used as base stock oils for alcohol or acids, is the only known catalyst used commercially
synthetic (Group IV) lubricants, destined mainly for auto- to produce low viscosity PAOs (Kv100 = 4 cSt or 6 cSt), character-
motive and industrial applications.1,2 PAOs are commonly con- ised by moderately high VI > 120 and low pour points (as low
sidered to have superior properties compared to mineral as −66 °C). In contrast, AlCl3 and EtAlCl2, typically promoted
(Group I–III) base stocks,3,4 allowing for formulation of high- with water or alcohols, give heavier products of very high Kv100
performance engine oils, e.g. low-viscosity automotive oils, (>40 cSt), with VI > 130 and PP > −42 °C. Similar results were
which can yield >2% fuel savings. PAOs are classified and mar- obtained when chloroaluminate ionic liquids, [cation][Al2Cl7],
keted depending on their kinematic viscosity at 100 °C, Kv100
,
were used as catalysts. Lighter products were obtained using
which has to fall within a specific bracket for each PAO grade.5 chlorogallate(III) ionic liquids,8 but no physical properties have
Furthermore, they are required to have excellent thermal pro- been reported. Among other catalysts, those based on low-
perties. They must have a viscosity index (VI) value of at least valent chromium, TiCl4 or Zr-based metallocenes promote the
120, which indicates small changes in viscosity as a function formation of highly linear PAOs with high viscosities and VIs,4
of temperature. In addition, pour points (PP) must be below whilst solid Brønsted acids (e.g. zeolites) yield very light
−40 °C, ensuring that PAOs flow at extremely low temperatures. products.9
Finally, they must be characterised by low volatility and high
oxidative stability.4
Liquid coordination complexes (LCCs)
There is a strong relationship between molecular structure
LCCs are a new class of liquid Lewis acid, comprising of equili-
of PAOs and their physical properties.6,7 Linear oligomers have
brated cationic, anionic and neutral coordination complexes
high VI (desired), but also high pour point (undesired in many
(Scheme 1).10
applications); increase in chain branching decreases both the
LCCs are prepared by a one-step, solventless reaction of a
VI and PP. Achieving optimum physical properties requires a
simple donor molecule with an anhydrous metal halide, typi-
finely-tuned catalyst and reaction conditions during PAO syn-
cally with the metal halide in excess. Donor molecules may
include O-, N-, S- or P-donors, such as ureas,11 thioureas,
trialkylphosphines, trialkylphosphine oxides or some amides.
The metal is typically, but not exclusively, AlIII or GaIII. The
The QUILL Research Centre, The Queen’s University of Belfast, David Keir Building,
Stranmillis Road, BT9 5AG, Belfast, UK. E-mail: m.swadzba-kwasny@qub.ac.uk
molar fraction of metal halide to neutral ligand in an LCC,
This journal is © The Royal Society of Chemistry 2015
Green Chem.

View Article Online
Paper
Green Chemistry
phine, trihexyl(tetradecyl)phosphonium chloride and tetrabu-
tylphosphonium chloride were provided by Cytec Industries.
1-Ethyl-3-methylimidazolium chloride17 and 1-octyl-3-methyl-
imidazolium chloride18 were synthesised as described in the
literature.
Trioctylphosphine, supplied in an air-tight gas cylinder,
was used as received. Urea, thiourea, acetamide and trioctyl-
phosphine oxide were dried under reduced pressure (over-
night, 80 °C, 10−2 mbar) and stored in the glovebox (MBraun
labmaster dp, <0.3 ppm of H2O and O2). Octanenitrile was
dried over 3 Å molecular sieves and stored under argon.
Dimethylacetamide was stirred with calcium hydride (1 h), dis-
tilled under reduced pressure and stored under argon.
For small-scale experiments, anhydrous metal chlorides
were purchased: AlCl3, 99.99%; GaCl3, 99.999%, both packed
in ampoules under argon. The ampules were opened in a glove-
box and used as received. For scale-up experiments, alu-
minium(III) chloride, 98% anhydrous, was doubly sublimed,
transferred to glovebox and stored there.
Scheme 1 Main equilibria in LCCs, inferred from spectroscopic studies
for the key compositions (M = AlIII or GaIII, XMCl – mol fraction of metal
3
chloride in LCC).10
of metal halide in an LCC, χMCl , is of key importance: LCCs
3
with χMCl = 0.50 are considered neutral, as they do not contain
3
oligonuclear, Lewis acidic anions, [MxCl(3x+1)]−. Compositions
where χMCl > 0.50 have been shown to be Lewis acidic,10 as they
3
contain oligonuclear anions, which readily react with bases
following eqn (1).
In a typical sublimation procedure, aluminium(III) chloride,
dry sodium chloride and aluminium wire (each at 2.5 wt% per
mass of AlCl3) were, in the glovebox, transferred into a round-
bottomed flask. The flask was then fitted with a condenser,
removed from the glovebox and placed on a hotplate. The sub-
limation was carried out under a flow of nitrogen (190 °C, 2 h).
The setup was cooled, closed with a stopper and transferred to
the glove box. Sublimed AlCl3 (off-white solid) was crushed
using a mortar and pestle and the process was repeated to
obtain white doubly-sublimed aluminium chloride.
½M2Cl7ꢀꢁ þ B ! ½MCl4ꢀꢁ þ ½MCl3Bꢀ
ð1Þ
LCCs are promising Lewis acidic catalysts in industrial
applications; they combine ease of preparation, low price, and
high Lewis acidity. Gutmann Acceptor Numbers (AN, a quanti-
tative measure of Lewis acidity) reported for some LCCs (AN =
ca. 102)10 were slightly higher than values in chloroaluminate(III)
ionic liquids (AN = ca. 96).12 Moreover, certain donor-AlCl3
mixtures have already been demonstrated to act as Lewis
acidic catalysts: dimethylformamide–AlCl3 and tetrahydro-
furan–AlCl3 were reported to catalyse Friedel–Crafts alkyl-
ation13 and Diels–Alder cycloaddition,14 respectively.
Synthesis of liquid coordination complexes (LCCs)
LCCs were prepared in the glovebox. Donor was added slowly,
in small portions, with stirring, to metal(III) chloride. Heat was
evolved. Afterwards, the mixture was stirred (0.5–24 h,
30–80 °C) to complete the reaction. All LCCs were stored in the
glovebox until used.
Rationale for this work
The physical properties of PAOs depend strongly on the cata-
lyst used in their synthesis; to date, boron trifluoride has been
the only industrially viable catalyst yielding low viscosity PAOs.
BF3 is a colourless, toxic gas (classified as poisonous gas) and
pulmonary irritant.15 In contact with moisture it hydrolyses to
release HF, which in turn dissolves in water to form hydrofluo-
ric acid. Hydrofluoric acid is an acute poison, able to permeate
human/animal tissue and react with Ca2+/Mg2+ ions to form
fluorides, potentially causing cardiac arrest and tissue poison-
ing leading to death.16
In contrast, LCCs are non-volatile liquids, which do not
contain fluorine. Furthermore, they are easier to handle, less
expensive, and versatile in terms of structure. In this work,
LCCs were tested as a safer and economically viable alternative
to BF3 in the oligomerisation of 1-decene to polyalphaolefins.
Synthesis of chloroaluminate(III) ionic liquids
Chloroaluminate(III) ionic liquids were prepared in the glove-
box. Aluminium(III) chloride (2 mol eq.) was added slowly, in
small portions, with stirring, to organic chloride salt (1 mol
eq.). Heat was evolved. Afterwards, the mixture was stirred
(0.5–24 h, 60 °C) to complete the reaction. All ionic liquids
were stored in the glovebox until used.
Oligomerisation – small scale
Experiments were carried out in a battery of eight glass compu-
ter-controlled H.E.L. reactors (120 cm3) designed for high cor-
rosion resistance.
The reactor vessels and stirrer propellers were dried over-
night in an oven, and then cooled in a desiccator, over phos-
phorus(V) oxide. The remaining parts were dried with a heat
gun, and the reactors were assembled. The feedstock
Experimental
Purification of starting materials
Unless otherwise mentioned, all chemicals were purchased (1-decene, dry by Karl-Fisher analysis) was poured into the
from Sigma-Aldrich. 1-Decene, +96%, was sourced from both reactors and stirred vigorously (600 rpm) at the reaction tem-
AlfaAesar and Sigma-Aldrich and used as received. Trioctylphos- perature, whilst the reactors were purged with dry argon.
Green Chem.
This journal is © The Royal Society of Chemistry 2015

View Article Online
Green Chemistry
Paper
The LCC catalyst was loaded, in the glovebox, into a dry
All fractions and the residue were analysed using SimDist
gas-tight syringe. The tip of the needle was sealed, the loaded GC. After blending, pour point, Kv40 and Kv100 were measured,
syringe was removed from the glovebox and immediately and VI was calculated.
plunged through a septum into the reactor. The catalyst
Physical properties tests
addition to the vigorously stirred feedstock, equilibrated at the
reaction temperature, was carried out drop-wise, as fast as
possible without thermal runaway.
After a given reaction time, the reaction was quenched by
stirring with deionised water (600 rpm, 10 min, ambient tem-
perature) and centrifuged to enhance phase separation. All
products were analysed using SimDist GC.
Pour point was measured according to ASTM D97-11. The pour
point cryostat (Stanhope-Seta, Model 93531-7) was validated
against a standard gas oil sample (99851-0) according to ASTM
method (pour point −15 °C, cloud point −4 °C).
Kinematic viscosity (Kv) was measured according to ASTM
D445 – 11a. The measurements were performed at 40 and
100 °C using the appropriate Cannon-Fenske kinematic vis-
cosity glassware and a dedicated, precisely-controlled heating
bath. Kv was found by timing the gravitational flow of the
sample through a capillary; temperature was maintained using
a high-accuracy heating bath.
Oligomerisation – scale-up
Scale-up reactions were carried out in a computer-controlled
glass H.E.L. reactor (5 L). The reactor was dried (60 °C, 30 min,
10−2 mbar) and then 1-decene was added under positive
pressure of argon; the remaining air was purged by the flow of
argon. Subsequently, vigorously stirred (500 rpm) 1-decene was
equilibrated at the reaction temperature.
Viscosity index (VI) was calculated according to eqn (2),
where L and H are constants based on Kv100, tabularised in
ASTM 2270.
The LCC catalyst (0.25–3.00 wt% per mass of the feedstock)
was loaded into a dry syringe as described for the small-scale
oligomerisation. The catalyst was added using a syringe pump
(>0.2 cm3 min−1), slow enough to prevent thermal runaway.
After a given reaction time, the stirring was stopped and the
system was allowed to cool to ambient temperature. The reac-
tion mixture was washed with deionised water (3 × 1 L). Each
batch was analysed using SimDist GC.
VI ¼ 100ðL ꢁ Kv40ÞðL ꢁ HÞꢁ1
ð2Þ
Results and discussion
Synthesis and properties of LCCs
LCCs used in this work were based on aluminium(III) chloride
or gallium(III) chloride, and on a range of ligands, listed with
abbreviations in Table 1.
Simulated distillation GC (SimDist GC)
Samples for SimDist GC analysis were dissolved in toluene
Ligands were selected from off-the-shelf chemicals, and rep-
(100 mg cm−3), dried over magnesium sulfate, filtered and resented a selection of donor atoms (O, S, P and N-donors)
analysed according to ASTM D6352. The analysis were carried and different affinity to olefins (from urea to trioctylphos-
out using Agilent 6890N GC, equipped with an FID detector phine). It is noteworthy that in all cases miscibility of LCCs
and a high-performance HT PTV (high-temperature program- and olefins was limited, and biphasic mixtures were formed.
mable temperature vaporising) inlet with optimised design for
In selecting the metal halide to ligand ratio, χMCl , care was
3
SIMDIS applications. Column: Agilent J&W DB-HT Sim Dis taken to form mobile liquids. LCCs based on AlCl3 formed
Column.
homogenous liquids around χAlCl = 0.60.10 Some LCCs based
3
on GaCl3 were liquid within a wide compositional range
Fractionated distillation
(χGaCl = 0.50 to 0.75), but to ensure all GaCl3-based LCCs
3
Fractionated distillation was carried out in a computer-con- being liquid, the composition of χGaCl = 0.67 was required.
3
trolled Distrilab D5236 Crude Oil Distillation System with 6 L The C7CN–AlCl3 combination did not form a homogenous
boiling flask, according to ASTM D5236.
liquid and hence wasn’t further studied.
In a typical procedure, the distillation rig was dried prior to
As a consequence of their high Lewis acidity, LCCs are
use (1 mmHg, 30 min). A sample (2 to 4 L) was dried over hydrolytically unstable, which makes them active catalysts in
MgSO4, filtered and loaded into the boiling flask. The con- carbocationic processes. In oligomerisation of olefins, LCCs
ditions required to remove each fraction were programed into
the software; the expected mass of each fraction was predicted
from SimDist GC. Firstly, the decene fraction was removed
(10 mmHg, vigorous stirring) at a carefully controlled heating
Table 1 Abbreviations and systematic names of donor molecules used
in this work
L
Name
rate. After the decene removal and cooling of the apparatus,
the dimer (C20) fraction was distilled (0.1 mmHg, vigorous
stirring). Then, the distillation rig was cleaned with toluene to
P888
P888
Trioctylphosphine
Trioctylphosphine oxide
Urea
Thiourea
Acetamide
Dimethylacetamide
Octanenitryl
O
prevent contamination of subsequent fractions and higher oligo- Ur
SUr
AcA
mers (mainly the trimer, C30) were fractionally distilled
(0.1 mmHg) to obtain PAO6. A minimum of 0.4 L (the heaviest
fraction) was left in the boiling flask as residue.
DMA
C7CN
This journal is © The Royal Society of Chemistry 2015
Green Chem.

View Article Online
Paper
Green Chemistry
Fig. 1 SimDist GC analysis of commercial PAO samples: (a) chromatograms, (b) simulated distillation curves, and (c) distribution of oligomer frac-
tions as a function of PAO grade. Dashed lines in graph (c) are only guidelines for the eye.
react with adventitious water in the system (eqn (3)), and
generates a Brønsted superacidic proton, which reacts with an
olefin to form a carbocation.
Table 2 Conversions and oligomer distributions, obtained using two
LCCs (120 °C, 1 h, 1 wt% of LCC). Values measured for each reaction
replicate are given, along with calculated averages and standard
deviations
½M2Cl7ꢀꢁ þ H2O ! ½MCl4ꢀꢁ þ ½MCl3OHꢀ þ Hþ
ð3Þ
Product distribution (%)
No.
Conversion (%)
C20
C30
C40
C50
C60
C70+
Distribution of oligomers in commercial PAO samples
Fig. 1 depicts SimDist GC analysis of commercial PAO4 and
PAO6 samples. Oligomers of 1-decene, Cn (n – number of
carbons in the alkyl chain) are easily identifiable in the chro-
matograms (Fig. 1a). Starting material (C10) or dimers (C20)
are absent in either chromatogram. PAO4 contains mainly the
trimer (C30), with small amounts of tetramer (C40), whereas
PAO6 contains similar quantities of C30 and C40, with signifi-
cant amounts of pentamer (C50) and some hexamers and
heavier oligomers (C60 and C70+).
The amount of each fraction in SimDist chromatogram can
be precisely quantified. The chromatograms can be processed
into simulated distillation curves (Fig. 1b), where the mass per-
centage of hydrocarbon fractions in the sample is plotted
against their boiling points. This allows for the construction of
Fig. 1c, where distribution of oligomers can be plotted as a
function of a selected factor (here, PAO grade).
Ur–GaCl3, χGaCl = 0.67
3
1
2
3
4
74
72
76
72
74
2
40
33
32
32
34
4
35.9
35.5
34.4
35.0
35.2
0.6
13.8
17.0
17.2
16.8
16.2
1.6
6.9
8.5
10.6
9.8
9.0
1.6
2.8
4.3
4.0
4.2
3.8
0.7
0.7
2.1
2.0
2.1
1.7
0.7
Aver
St dev
Ur–AlCl3, χAlCl = 0.60
1
2
3
4
78.53
80.5
77.5
78.5
78.5
78.7
1.1
35
30
38
37
36
35
3
37
37
34
30
31
34
3
15.5
17.6
15.7
14.2
16.8
16.0
1.3
5.2
8.8
7.8
9.0
9.0
8.0
1.7
5.2
3.8
2.6
5.2
3.9
4.1
1.0
2.0
3.1
2.0
4.5
3.2
3.0
1.0
5
Aver
St dev
Error assumed for all experiments (%)
2
4
3
2
2
1
1
From analysis of the commercial PAO samples, it transpires
that, to maximise PAO yield, high yields of C30 and C40 frac-
tions are desired. Some heavier oligomers are needed for
PAO6, but excessive amount would detrimentally affect pour
point, hence must be avoided. Monomer and C20 can be
recycled, but recycling of large volumes may not be economi-
cally viable.
Error estimation. Ideally, all experiments should be carried
out in several replicates; however, considering the large
number of experiments in this work, this would not be practi-
cal. Instead of repeating all experiments, two model oligomeri-
sation reactions: with Ur–AlCl3, χAlCl = 0.60, and with Ur–
3
GaCl3, χGaCl = 0.67, were carried out in several replicates. Two
3
different batches of each catalyst were used. Conversions and
products distribution for each run, with calculated standard
deviations, are shown in Table 2. The higher of the two stan-
Small-scale screening
The carbocationic mechanism relies on traces of adventitious dard deviations calculated for each fraction, rounded to the
water for carbocation generation; consequently, the variation nearest integer, was assumed as error for this fraction in sub-
in the amount of adventitious water is the main source of vari- sequent experiments.
ation in the results. It is therefore essential to rigorously main-
Influence of the catalyst was investigated by varying the
tain consistent experimental procedure, and to assess the error metal halide, the ligand, and their molar ratio (χMCl ), in catalysts
3
before drawing any conclusions based on gathered data. More- used for oligomerisation of 1-decene under identical conditions.
over, experimental error is inversely proportional to the scale
The influence of the χMCl value on conversion and products
3
of reaction; hence, error estimation was of particular impor- distribution could be studied only for selected GaIII-based
tance in small-scale studies.
systems, which were liquid for a wide range of χGaCl values.
3
Green Chem.
This journal is © The Royal Society of Chemistry 2015

View Article Online
Green Chemistry
Paper
Table 3 Conversions and products distributions, obtained using LCCs
Table 4 Conversions and distributions of 1-decene oligomers,
obtained using LCCs based on GaCl3 and AlCl3, and a range of ligands
(120 °C, 1 h, 1.707 mmol of LCC)
based on Ur–GaCl3, where χGaCl = 0.60, 0.67 or 0.75 (120 °C, 1 h,
3
1.707 mmol of GaCl3 equivalent in each catalyst loading)
Product distribution (%)
Product distribution (%)
Conversion
χGaCl
Conversion (%)
C20
C30
C40
C50
C60
C70+
L
(%)
C20
C30
C40
C50
C60
C70+
3
0.50
0.60
0.67
0.75
—
—
—
—
—
—
—
—
1.7
—
L-GaCl3, χGaCl = 0.67
3
71.5
73.5
78.5
36.9
34.1
38.7
36.9
35.2
33.5
15.6
16.2
15.5
7.1
8.9
7.7
3.5
3.8
4.5
P888
P888
Ur
76.5
81.5
75.5
82.5
80.5
86.5
43.7
46.0
32.2
32.0
39.0
42.1
35.8
33.5
33.6
34.4
40.3
33.9
14.6
14.9
17.5
17.2
15.1
15.2
5.3
5.0
9.4
9.8
5.0
5.9
0.7
0.6
5.4
4.9
0.6
2.3
—
—
2.0
1.8
—
O
AcA
SUr
C7CN
0.6
L-AlCl3, χAlCl = 0.60
3
P888
P888
Ur
81.5
82.5
78.5
79.5
87.5
33.5
39.3
37.4
28.0
30.1
37.3
34.4
29.7
33.1
35.8
18.6
18.4
14.2
17.8
19.7
7.5
6.1
9.0
11.5
9.3
2.5
1.8
5.2
5.1
3.5
0.6
—
4.5
4.5
1.7
O
AcA
SUr
Conversions (Table 4) were consistently high, between
75.5 and 87.5%. Although this discrepancy lies outside of
the error bars assumed for conversion, no simple correla-
tion between the LCC structure and conversion could be
proposed.
Product distribution (Table 4, Fig. 3) changed moderately
with changes in LCC In general, slightly lower MW oligomers
were produced using L-GaCl3 systems, than using analogous
L-AlCl3 systems. This was most evident when comparing C20
content in both graphs in Fig. 3. Moreover, distribution
shifted towards lighter products (C20) for LCCs with hydro-
phobic ligands: P888 and P888O (three long alkyl chains per
ligand) and C7CN (one long chain per ligand). In contrast,
small and hydrophilic ligands gave less C20 and slightly more
heavy oligomers; which was particularly visible for DMA–MCl3.
Fig. 2 Distributions of 1-decene oligomers, obtained using LCCs based
on Ur–GaCl3, where χGaCl = 0.60, 0.67 or 0.75 (120 °C, 1 h, 1.707 mmol
3
of GaCl3 in each catalyst loading).
The DMA–GaCl3 system, at four compositions: χGaCl = 0.50, This is rather surprising; increased miscibility with olefin was
3
0.60, 0.67 and 0.75, was investigated. In each reaction, equal expected to result in longer contact of propagating carbocation
amounts of GaCl3 equivalent (1.707 mmol) were used. The with the catalyst phase, whereas lower miscibility should result
results are compared in Table 3 and, to facilitate the analysis, in phase-separation from LCC of relatively short oligomers,
product distributions for successful experiments are plotted such as C20 or C30. Here, however, a reverse trend was
against χGaCl in Fig. 2.
observed.
For comparison, oligomerisation of 1-decene has been
3
In a control experiment with the neutral composition, χGaCl
3
= 0.50, no conversion of 1-decene was observed, as anticipated. carried out using a range of chloroaluminate(III) ionic liquids
For Lewis acidic compositions, conversion obtained for com- and aluminium(III) chloride powder, with equal mol equiva-
positions, χGaCl = 0.60 and 0.67, was the same within experi- lents of AlCl3 in each reaction. The selected cations ranged
3
mental error, but for the composition of highest acidity (χGaCl
from
very
hydrophilic
(1-ethyl-3-methylimidazolium,
3
= 0.75) it increased slightly. Product distribution remained [C2mim]+), through moderately hydrophilic (1-octyl-3-methyl-
unaffected by changes in χGaCl , taking into consideration 4% imidazolium, [C8mim]+ and tetrabutylphosphonium, [P4444]+),
3
and 3% error bars on C20 and C30 content (Fig. 2).
to hydrophobic (trihexyl(tetradecyl)phosphonium, [P66614]+).
In conclusion, the catalyst composition (χMCl ) had negli- The main anion in chloroaluminate(III) ionic liquids at χAlCl
=
3
3
gible impact on the reaction outcome, provided, that the cata- 0.67 is [Al2Cl7]−, the same as in LCCs at χAlCl = 0.60 (see
3
lyst was Lewis acidic (e.g. χMCl > 0.50).
Scheme 1). However, in ionic liquids there are no ligands
3
The influence of changes in LCC metal and ligand on the other than chloride coordinating to AlIII, whereas in LCCs the
oligomerisation outcome was investigated by testing L-AlCl3 [Al2Cl7]− anion remains in a dynamic equilibrium with other
systems at χAlCl = 0.60 and L-GaCl3 systems at χGaCl = 0.67, coordination complexes (Scheme 1).
3
3
under identical reaction conditions. Equal amounts of MCl3
equivalents (1.707 mmol) were used.
Miscibility of ionic liquids and olefin phase was limited;
biphasic mixtures were formed in all cases. Unlike in
This journal is © The Royal Society of Chemistry 2015
Green Chem.

View Article Online
Paper
Green Chemistry
Fig. 3 Distributions of 1-decene oligomers, obtained using LCCs based on GaCl3 or AlCl3, and a range of ligands (120 °C, 1 h, 1.707 mmol of LCC).
Dashed lines are only guidelines for the eye.
Table 5 Conversions and distributions of 1-decene oligomers, obtained
using chloroaluminate(III) ionic liquids, χAlCl
= 0.67 (120 °C, 1 h,
3
1.707 mmol of AlCl3 equivalent)
Product distribution (%)
C20 C30 C40 C50 C60 C70+
Conversion
(%)
Catalyst
AlCl3
55.5
18.4 29.4 23.8 20.2
31.9 24.2 15.3 14.0
20.1 27.2 20.1 16.6
8.3
8.9
9.5
—
5.7
6.5
[C2mim][Al2Cl7] 79.5
[P4444][Al2Cl7] 85.5
[C8mim][Al2Cl7] 86.5
[P66614][Al2Cl7] 95.5
17.5 19.9 16.4 15.2 11.7 19.3
5.3 12.7 14.8 19.0 19.1 29.1
LCCs, both conversions and products distributions were
closely related to expected miscibility of ionic liquids with
olefin.
As shown in Table 5, AlCl3 gave the lowest conversion. Con-
versions achieved with ionic liquids were higher (on par with
those obtained with LCCs), and increased with lipophilicity of
the cation. The distribution of products (Fig. 4) also corres-
ponded well to the cation lipophilicity: [P66614][Al2Cl7] gave
the most heavies and the least C20, whereas [C2mim][A2Cl7]
produced more C20 and less C60 and C70+. It is noteworthy
Fig. 4 Distributions of oligomers obtained using chloroaluminate(III)
ionic liquids, χ = 0.67 (120 °C, 1 h, 1.707 mmol of AlCl3 equivalent).
AlCl3
Considering the carbocationic mechanism,19 distribution
that products obtained with ionic liquids were heavier than of oligomers depends on the rate of propagation (chain
those synthesised using AlIII-based LCCs (see Fig. 4 vs. 3).
growth) vs. rate of termination. Under identical physical con-
It is apparent that structural factors have different effect on ditions (reaction temperature etc.), propagation depends on
the products distribution in ionic liquids and in LCCs. In complex interactions between all charged species present in
chloroaluminate(III) ionic liquids, lipophilic cations produced the reaction mixture. In both ionic liquids and LCCs, this
heavier oligomers, by enhancing longer contact of propagating interaction may be limited by mass transport, governed by
carbocation with the ionic liquid. In contrast, for LCCs, lipo- factors such as solubility of olefin in the droplet of catalyst, or
philic ligands gave lighter products and small hydrophilic viscosity of the catalyst. Provided good mass transport, the
ligands gave more heavies, for both GaIII and AlIII-based LCCs. chemical structure of interacting ions may be the crucial
Furthermore, ionic liquids based on [Al2Cl7]− gave heavier pro- factor: ‘hardness’ of the anion, as well as competitive inter-
ducts than LCCs containing the same anion. This indicates actions of the anion with carbocation and with cations in ionic
that factors other than solubility govern products distribution liquid/LCC. Understanding what governs products distribution
in LCC-catalysed oligomerisation.
when using LCCs, which would exceed beyond phenomenolo-
Green Chem.
This journal is © The Royal Society of Chemistry 2015

View Article Online
Green Chemistry
Paper
Table 6 Conversions and distributions of 1-decene oligomers, obtained using Ur–AlCl3, χAlCl = 0.60, at a range of temperatures (1 h, 1 wt% of LCC)
3
Product distribution (%)
Reaction
temperature (°C)
Conversion
(%)
C20
C30
C40
C50
C60
C70+
160
140
130
120
110
100
90
80
60
40
83.5
78.5
77.5
78.5
78.5
79.5
77.5
76.5
83.5
70.5
49.7
47.7
41.8
37.4
31.0
24.2
23.5
18.5
4.8
31.5
31.0
30.1
29.7
31.0
31.8
34.0
29.1
18.2
8.6
13.3
11.6
13.1
14.2
16.8
17.8
18.3
18.5
13.3
8.6
3.6
6.5
7.8
1.8
1.3
3.9
5.2
5.2
7.6
7.8
9.3
14.5
15.8
0.0
1.9
3.3
4.5
5.8
7.0
5.9
8.6
29.7
48.2
9.0
10.3
11.5
10.5
15.9
19.4
17.3
1.4
gical observations, requires in-depth physico-chemical and
mechanistic studies, supported by computational modelling.
These studies are currently under way in our group.
At this stage, however, it can be stated that there is a quali-
tative difference between AlIII-based LCCs and other AlIII based
catalysts; the introduction of a non-halide ligand altered the
way in which the catalyst affects product distribution. This
suggests that LCCs may promote the formation of different
classes of PAOs than previously reported Al(III)-based catalysts,
which led to high-viscosity, heavy products.
Influence of the reaction conditions was investigated by
varying reaction temperature and amount of LCC catalyst.
The Ur–AlCl3 (χGaCl = 0.60) system was used to catalyse
3
oligomerisation of 1-decene at a wide range of temperatures,
from 40 to 160 °C. Conversions and product distributions are
compared in Table 6; conversions and yields of selected frac-
tions are plotted against reaction temperature in Fig. 5.
Conversions were independent of temperature, and levelled
at 79 4%, except for the reaction carried out at the lowest
temperature, 40 °C, which reached only 70.5% conversion.
As expected for carbocationic process, distribution of oligo-
mers depended strongly on the reaction temperature: as the
temperature increased, the distribution shifted towards lighter
products. Because termination rate increases with temperature
more rapidly than propagation rate, oligomeric chains are
shorter at higher temperatures. From 80 to 160 °C, yields in all
fractions change gradually, but for <80 °C the amount of heavy
oligomers increases rapidly. At high reaction temperatures the
amount of heavy products is negligible, but the amount of C20
increases to nearly 50% of the sample. Consequently, tempera-
tures around 120 °C appear to offer a good balance of heavy
oligomers (<5%) and relatively low level of the C20 fraction,
whilst maintaining the highest possible yield of the middle
fractions.
Fig. 5 Distributions of 1-decene oligomers, obtained using Ur–AlCl3,
χAlCl = 0.60, at a range of temperatures (1 h, 1 wt% of LCC).
3
Table 7 Conversions and distributions of 1-decene oligomers, obtained
using two LCC catalysts: L-AlCl3, χAlCl = 0.60, where L = P888O or Ur, at
3
different quantities, expressed as mmol of AlCl3 eq. (120 °C, 1 h)
Product distribution (%)
Catalyst amount Conversion
(AlCl3 mmol)
(%)
C20 C30 C40 C50 C60 C70+
P888O–AlCl3, χAlCl = 0.60
3
0.86
1.71
3.41
5.12
75.5
82.5
88.5
85.5
45.6 34.9 14.8
39.3 34.4 18.4
25.1 35.4 24.0 10.3 4.6
4.7
6.1 1.8
—
—
—
0.6
0.6
28.4 36.7 22.5
9.5 2.4
Ur–AlCl3, χAlCl = 0.60
3
1.71
3.41
5.12
79.0
86.5
91.5
36.5 30.1 15.4
33.3 37.4 18.1
29.3 37.0 19.9
9.0 4.5
7.6 2.3
8.8 2.8
4.5
1.2
2.2
The influence of catalyst amount was investigated using
two catalytic systems: P888O–AlCl3, χAlCl = 0.60, and Ur–AlCl3,
3
χAlCl = 0.60. The amount of former LCC was varied from 0.86
3
to 5.12 mmol of AlCl3 eq.; however, for <1.00 mmol of AlCl3
eq., small amount of catalyst hindered reproducibility of the
For P888O–AlCl3, conversion was slightly lower for the
results. Therefore, the latter system was used at 1.71 to minimum catalyst amount, but for 1.71 to 5.12 mmol of AlCl3
5.12 mmol of AlCl3 eq. Conversions and product distributions eq., it oscillated at ca. 85%. In contrast, for Ur–AlCl3, a signifi-
are compared in Table 7.
cant increase in conversion as a function of catalyst loading
This journal is © The Royal Society of Chemistry 2015
Green Chem.

View Article Online
Paper
Green Chemistry
Fig. 6 Distributions of 1-decene oligomers, obtained using two LCC catalysts: L-AlCl3, χAlCl = 0.60, where L = P888O (left) or Ur (right), at different
3
quantities, expressed as mol% of AlCl3 (120 °C, 1 h).
was observed, up to 91.5%. Yields of all fractions are plotted
against the catalyst loading in Fig. 6. For both LCCs, there was
a small but observable decrease in C20 content, in favour of
C40, with increasing catalyst amount, which levelled out for
higher contents of the P888O–AlCl3 system.
In conclusion, optimising the reaction temperature is the
key to manipulating the distribution of oligomers. Increasing
the catalyst amount could also aid to increase conversion and
slightly decrease C20 amount, but tripling the catalyst amount
may be more expensive solution than recycling an additional
few per cent of the feedstock.
Table 8 Conversions and distributions of 1-decene oligomers,
obtained using Ur–AlCl3, χAlCl = 0.60, as a function of reaction time
3
(100 °C, 1 wt% LCC)
Product distribution (%)
Reaction time Conversion
(min)
(%)
C20 C30 C40 C50 C60 C70+
5
55.0
66.0
68.0
69.0
69.0
69.0
71.0
72.0
72.0
72.0
74.0
74.0
78.0
78.0
37.0 38.9 14.8 5.6
40.0 36.9 12.3 6.2
38.8 37.3 13.4 6.0
39.7 35.3 14.7 5.9
38.2 36.8 14.7 5.9
38.2 36.8 14.7 5.9
38.6 35.7 15.7 5.7
38.0 36.6 14.1 7.0
38.0 36.6 14.1 7.0
36.6 38.0 14.1 7.0
37.0 37.0 15.1 6.8
37.0 37.0 15.1 6.8
36.4 36.4 15.6 6.5
35.1 37.7 15.6 6.5
1.9
3.1
3.0
2.9
2.9
2.9
2.9
2.8
1.4
1.4
1.4
1.4
2.6
2.6
1.9
1.5
1.5
1.5
1.5
1.5
1.4
1.4
2.8
2.8
2.7
2.7
2.6
2.6
10
15
20
25
30
35
40
45
50
55
60
100
120
Scale-up
In order to test physical properties, such as kinematic viscosity
(Kv), viscosity index (VI) and pour point (PP), it was necessary
to scale-up the reaction. The products could then be distilled
into fractions, and blended with the aim to achieve marketable
PAO4 or PAO6 (viscosity brackets of 4 or 6 cSt, respectively).
Scale-up reaction conditions were chosen to strike balance
between achieving optimum conditions and project economy.
Urea was selected as a very cheap ligand, offering a balanced
oligomers distribution: limited amounts of C20 and heavies.
AlIII-based LCC was chosen over the GaIII equivalent due to
much lower price. Reaction temperature of ca. 120 °C was
chosen to balance C20 and heavies in the product.
reaction time, in a reaction catalysed with Ur–AlCl3, χAlCl
=
3
0.60, carried out for 2 h (Table 8 and Fig. 7).
The initial reaction rate was very high, with 55% conversion
after 5 min and 66% after 10 min; afterwards the conversion of
C10 increased slowly to plateau at 78% after 100 min. Product
distribution was remarkably independent from reaction time
and from conversion.
In conclusion, it has been demonstrated that the oligomeri-
sation could be carried out using as little as 0.25 wt% of the
catalyst, achieving >75% conversion and favourable product
distribution (Fig. 7). Alternatively, the catalyst amount could
be increased, which would increase the conversion and
decrease the volume of the recycled fraction: C10 and C20
(Fig. 6). To make an informed decision, techno-economical
evaluation of the process would be required. Moreover, at
0.25 wt%, considering the low catalyst price and benign pro-
ducts of hydrolysis (urea and oxo/hydroxoaluminium species,
Reproducibility of scale-up reactions was superior to that in
the small scale. Products prepared at the same conditions con-
tained slightly less heavies. The catalyst amount was reduced
to 0.25 wt%, whilst maintaining good conversion.
Reaction kinetics. Preliminary small-scale studies suggested
that after 1 h the reaction did not progress. However, frequent
sampling of the reaction mixture at that scale introduced sig-
nificant amounts of moisture, compromising data quality.
This problem has been eliminated at the scale-up. Conversion
and products distribution were studied as a function of the
Green Chem.
This journal is © The Royal Society of Chemistry 2015

View Article Online
Green Chemistry
Paper
mildly acidified water), recycling of the catalyst is not required.
Upon increasing the catalyst amount, it may be beneficial to
attempt recycling: LCC would be purified from acid soluble
oils (ASO), which accumulate in the catalyst phase during car-
bocationic processes,20 and then returned to the reaction. This
methodology is currently studied in our group. However, the
recycle conditions (solvents, energy demand) may render the
decomposition and disposal of the catalysts a more sustain-
able option, as practiced in comparable industrial processes
(viz. BF3-based oligomerisation).4
Physical properties of PAOs. Scale-up reactions were per-
formed at four temperatures between 100 and 160 °C, with
several runs at each temperatures. C10 and C20 fractions were
removed at reduced pressure. C30 and some C40 were separ-
ated by fractionated distillation, leaving heavy products as a
residue. Then, these fractions were blended at various pro-
portions, with the aim to achieve marketable PAO grades: Kv100
= ca. 4 or 6 cSt, VI >120 and PP < −40 °C. The catalyst and
process conditions were different from the commercial ones,
hence different degree of branching could be anticipated. Con-
sequently, proportion of oligomers typically found in the com-
mercial samples could only be a rough guideline. N.B., unlike
the commercial samples, PAOs produced in this work have not
yet been hydrogenated, which is anticipated to slightly increase
Kv values, but should not affect other physical properties.
All tested samples had very low pour points, PP < −57 °C.
Other physical parameters of PAOs blended at various pro-
portions, from oligomers synthesised at different tempera-
tures, are listed in Table 9. Achieving Kv100 = ca. 4 or 6 cSt was
feasible by adjusting the ratio of oligomers. The main chal-
lenge lied in accessing VI > 120, which could not be obtained,
at any fractional composition, with products made at 140 or
160 °C. However, it was possible to obtain satisfactory VI for
PAO6 produced at 100 and 120 °C, and for PAO4 produced at
100 °C. This clearly suggests that the degree of branching in
the studied samples is higher than that in commercial
Fig. 7 SimDist GC chromatograms (top) and corresponding graph
(bottom), depicting conversions and distributions of 1-decene oligomers
as a function of reaction time (100 °C, 0.25 wt% of LCC: Ur–AlCl3,
χAlCl = 0.60).
3
Table 9 Physical properties of unhydrogenated PAO blends, produced using Ur–AlCl3, χAlCl = 0.60, at a range of temperatures (1 h, 0.25 wt% of
3
LCC)
Products distribution (%)
Reaction
Blend no.
temperature (°C)
C30
C40
C50+
Kv100 (cSt)
Kv40 (cSt)
VI (−)
1
2
3
4
5
6
7
8
160
160
160
160
160
140
140
120
120
120
100
100
100
86.9
47.0
48.5
29.3
14.1
77.8
47.0
46.5
42.0
42.0
93.9
45.5
77.0
13.1
36.0
35.4
45.5
54.5
22.2
35.0
31.3
36.0
37.0
6.1
—
4.11
5.84
5.68
6.72
7.79
4.13
6.39
5.22
5.79
5.74
3.10
5.27
4.31
20.28
35.16
32.32
43.87
53.93
19.88
39.61
26.88
31.46
31.35
12.58
26.19
20.08
102
108
115
106
109
109
111
127
126
126
105
137
124
17.0
16.2
25.3
31.3
—
18.0
22.2
22.0
21.0
—
9
10
11
12
13
32.3
16.0
22.2
7.0
This journal is © The Royal Society of Chemistry 2015
Green Chem.

View Article Online
Paper
Green Chemistry
This enabled the production of low-viscosity PAO4 and PAO6
base oils, in contrast to high-viscosity base oils accessible with
other AlIII-based catalysts.
After fractionated distillation and blending, it has been
possible to prepare PAO4 and PAO6 samples matching com-
mercial specifications. The combination of AlCl3 and inexpen-
sive ligands, such as urea or acetamide, makes a safer and
economically attractive alternative to the extant process, based
on poisonous and corrosive BF3 gas.
Acknowledgements
Petronas is kindly acknowledged for funding this project.
Cytec is acknowledged for providing trioctylphosphine and
phosphonium chloride salts.
Notes and references
Fig. 8 SimDist GC chromatograms of unhydrogenated PAO6 blend,
1 J. R. Dickey, Lubrication and Lubricants, in Kirk-Othmer
Encyclopedia of Chemical Technology, John Willey & Sons,
Inc., 2005, vol. 15, pp. 1–82.
produced using Ur–AlCl3, χAlCl = 0.60, compared to commercial PAO6
sample.
3
2 API 1509, Engine Oil Licensing and Certification System,
17th Ed., September 2012, Appendix E – API Base Oil Inter-
changeability Guidelines for Passenger Car Motor Oils and
Diesel Engine Oils, September 2011.
samples. Moreover, degree of branching can be controlled,
and decreased to a satisfactory level, by lowering the reaction
temperature.
3 R. Benda, J. Bullen and A. Plomer, J. Synth. Lubr., 1996, 13,
41.
4 S. Ray, P. V. Rao and N. V. Choudary, Lubr. Sci., 2012, 24,
23.
5 G. R. Lapin, L. H. Nemec, J. D. Sauer and D. Wagner,
Olefins, Higher, in Kirk-Othmer Encyclopedia of Chemical
Technology, John Wiley and Sons, Inc., 2010, vol. 17, pp.
709–728.
6 L. I. Kioupis and E. J. Maginn, J. Phys. Chem. B, 1999, 103,
10781.
Chromatogram of the unhydrogenated PAO6 sample pro-
duced in this work (Table 9, entry 10) was compared to the
chromatogram of a commercial PAO6 (Fig. 8). In the former
chromatogram, peaks corresponding to different oligomers
were slightly broader and shifted towards lower retention
times, which again suggested a higher degree of isomerisation
and more branching.
PAO samples synthesised using LCCs differ in degree of
branching from the commercial products. Nevertheless, they
were demonstrated to meet all major specifications for low vis-
cosity synthetic base oils, in both PAO4 and PAO6 grades.
7 S. S. Scheuermann, S. Eibl and P. Bartl, Lubr. Sci., 2011, 23,
221.
8 M. P. Atkins, K. R. Seddon and M. Swadźba-Kwaśny, Pure
Appl. Chem., 2011, 83, 1391.
9 E. T. Marquis, J. R. Sanderson and J. F. Knifton, US Pat,
5053569, 1991.
Conclusions
A new class of liquid Lewis acidic catalysts, liquid coordination
complexes, were successfully used as a sustainable and 10 F. Coleman, G. Srinivasan and M. Swadźba-Kwaśny, Angew.
inexpensive alternative to the industrially used BF3/nBuOH
catalyst.
Chem., Int. Ed., 2013, 52, 12582.
11 H. M. A. Abood, A. P. Abbott, A. D. Ballantyne and
K. S. Ryder, Chem. Commun., 2011, 47, 3523.
Oligomers distribution and physical properties of the pro-
ducts could be modified by altering reaction temperature. 12 (a) T. A. Zawodzinski Jr. and R. A. Osteryoung, Inorg. Chem.,
Lipophilicity of LCC ligand and the amount of the catalyst
used had lesser, but significant, effects on products distri-
1989, 28, 1710; (b) J. Estager, A. A. Oliferenko, K. R. Seddon
and M. Swadźba-Kwaśny, Dalton Trans., 2010, 39, 11375.
bution. Conversion could be influenced by changing reaction 13 (a) H. Aichaoui, J. H. Poupaert, D. Lesieur and
time and the amount of the catalyst.
J. P. Hénichart, Bull. Soc. Chim. Belg., 1992, 101, 1053;
(b) H. A. Derbala, J. Heterocycl. Chem., 2012, 49, 700.
It has been demonstrated that AlIII-based LCCs acted in the
oligomerisation process distinctly different to other AlIII-based 14 F. Fringuelli, R. Girotti, F. Pizzo and L. Vaccaro, Org. Lett.,
catalysts, such as AlCl3 or chloroaluminate(III) ionic liquids.
2006, 8, 2487.
Green Chem.
This journal is © The Royal Society of Chemistry 2015

View Article Online
Green Chemistry
Paper
15 F. Alam, F. Evans, G. Mani and J. R. Papcun, Boron 18 D. C. Apperley, C. Hardacre, P. Licence, R. W. Murphy,
Halides, in Kirk-Othmer Encyclopedia of Chemical Tech-
nology, John Wiley and Sons, Inc., 2003, vol. 4, pp.
138–168.
N. V. Plechkova, K. R. Seddon, G. Srinivasan, M. Swadźba-
Kwaśny and I. J. Villar-Garcia, Dalton Trans., 2010, 39, 8679.
19 (a) I. Majoros, A. Nagy and J. P. Kennedy, in Advances in
Polymer Science, Springer Berlin Heidelberg, 1994, vol. 112,
p. 1; (b) I. Majoros, T. M. Marsalko and J. P. Kennedy,
J. Polym. Sci., Part A: Polym. Chem., 1996, 34, 1675.
16 R. A. Smith, Fluorine Compounds, Inorganic, Hydrogen, in
Kirk-Othmer Encyclopedia of Chemical Technology, 2003,
vol. 13, pp. 1–26.
17 J. S. Wilkes, J. A. Levisky, C. L. Hussey and 20 (a) L. F. Albright, M. A. Spalding, C. G. Kopser and
M. L. Druelinger, Proc. Int. Symp. Molten Salts, 1980, 81–89,
245.
R. E. Eckert, Ind. Eng. Chem. Res., 1988, 27, 386;
(b) L. F. Albright, Ind. Eng. Chem. Res., 2002, 41, 5627.
This journal is © The Royal Society of Chemistry 2015
Green Chem.