Etherification of glycerol by isobutylene: Tuning the product composition

Article abstract of DOI:10.1016/j.apcata.2010.10.014

Etherification of glycerol by isobutylene was conducted in a batch mode using acidic and partially neutralized Amberlyst-15 ionic resin, p-toluenesulfonic acid, silicotungstic acid, cesium salt of silicotungstic acid, and ionic liquid containing sulfonic acid groups. All the catalysts are comparable in terms of the initial rate of glycerol conversion into mono-ether (except cesium salt of heteropolyacid), but differ substantially with respect to the yields of higher ethers of glycerol. Ionic liquid and heteropolyacid are immiscible/insoluble in higher ethers of glycerol. As a result they have unique capability of suppressing the formation of tri-ether during the initial stage of glycerol conversion. Acidic Amberlyst-15 in the form of fine powder has the highest activity per unit weight for glycerol etherification and relatively high activity in isobutene oligomerization. Partial exchange of acidic protons with cations Na⁺, Ag⁺, Mg²⁺, and Al³⁺ decreases the rates of all the processes, but isobutylene oligomerization is suppressed more efficiently. Ag⁺- and Al³⁺-modified Amberlyst demonstrates higher yields of tri-ether and has specific pattern of interaction with gaseous isobutene distinctive to other metal-substituted Amberlysts.

Full text of DOI:10.1016/j.apcata.2010.10.014

Applied Catalysis A: General 390 (2010) 235–244
Contents lists available at ScienceDirect
Applied Catalysis A: General
journal homepage: www.elsevier.com/locate/apcata
Etherification of glycerol by isobutylene: Tuning the product composition
Hee Jong Leea, Doyoung Seunga, Kwang S. Junga, Hwayong Kimb, Igor N. Filimonova,∗
a GS Caltex Corporation, Research Center, 104-4 Moonji-dong, Yuseong-gu, Daejeon, 305-380, Republic of Korea
b School of Chemical and Biological Engineering, Seoul National University, Seoul 151-744, Republic of Korea
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 8 June 2010
Received in revised form 11 October 2010
Accepted 15 October 2010
Available online 10 November 2010
Etherification of glycerol by isobutylene was conducted in a batch mode using acidic and partially neu-
tralized Amberlyst-15 ionic resin, p-toluenesulfonic acid, silicotungstic acid, cesium salt of silicotungstic
acid, and ionic liquid containing sulfonic acid groups. All the catalysts are comparable in terms of the
initial rate of glycerol conversion into mono-ether (except cesium salt of heteropolyacid), but differ
substantially with respect to the yields of higher ethers of glycerol.
Ionic liquid and heteropolyacid are immiscible/insoluble in higher ethers of glycerol. As a result they
have unique capability of suppressing the formation of tri-ether during the initial stage of glycerol
conversion.
Acidic Amberlyst-15 in the form of fine powder has the highest activity per unit weight for glyc-
erol etherification and relatively high activity in isobutene oligomerization. Partial exchange of acidic
protons with cations Na+, Ag+, Mg2+, and Al3+ decreases the rates of all the processes, but isobutylene
oligomerization is suppressed more efficiently. Ag+- and Al3+-modified Amberlyst demonstrates higher
yields of tri-ether and has specific pattern of interaction with gaseous isobutene distinctive to other
metal-substituted Amberlysts.
Keywords:
Etherification
tert-Butylation
Isobutylene
Glycerol
Ion-exchange resin
Amberlyst
Partial neutralization
p-Toluenesulfonic acid
Heteropolyacid
Silicotungstic acid cesium salt
Ionic liquid
© 2010 Elsevier B.V. All rights reserved.
Pulse chemisorption
1. Introduction
usage as fuel components and have been under investigation for
the last decade in several laboratories, both academic and industrial
Components of engine fuel from sustainable sources became a
business area in the last decades. With the introduction of biodiesel
to the market substantial “overproduction” of glycerol resulted in
the corresponding downtrend in the price of crude glycerol making
the latter an attractive platform chemical.
A number of value-added products derived from glycerol are
already proposed and some of them are commercialized [1–6]. A
well-known possibility for the upgrade of glycerol to fuel additives
is its transformation into higher (di- and tri-) tertiary butyl ethers.
Di- and tri-tert-butyl ethers of glycerol have good potential for the
[7–13].
t-Butylation of glycerol may be realized via the contact of glyc-
erol and isobutylene in the batch-type reactor under the modest
heating (50–100 ◦C) and pressure in the presence of a suitable acidic
catalyst. Ethers of glycerol (GE) are formed as a result of 3 consecu-
tive steps and currently the following reaction scheme is accepted:
G + IB ꢀ MTBGE
(1)
(2)
(3)
MTBGE + IB ꢀ DTBGE
DTBGE + IB ꢀ TTBGE
Two side reactions give rise to the formation of tert-butyl alcohol
(TBA) and oligomers of isobutylene (mainly dimers – DIB).
Abbreviations: IB, isobutylene; HSiW, silicotungstic acid H4SiW12O40
; G,
glycerol; Cs2.5SiW, cesium salt Cs2.5H1.5SiW12O40 DIB, diisobutylene (mixture
;
IB + H2O ꢀ TBA
2IB → DIB
(4)
(5)
of 2,4,4-trimethyl-1- and 2,4,4-trimethyl-2-pentenes); IL-6, 1-(4-sulfobutyl)-3-
hexylimidazolium trifluoromethanesulfonate; GTBE, glycerol tert-butyl ethers;
TsOH, para-toluenesulfonic acid; MTBGE, mono-tert-butyl glycerol ether; TBA, tert-
butyl alcohol; DTBGE, di-tert-butyl glycerol ether; Na(50)Amb, 50% ion-exchanged
Amberlyst-15 sodium form; TTBGE, tri-tert-butyl glycerol ether; H-Amb, acidic form
of Amberlyst-15.
Assuming that MTBGE and DTBGE may have two isomers, in
general case a mixture of five different glycerol ethers is obtained.
Higher ethers (di- and tri-) of glycerol are preferred as fuel com-
ponents due to their miscibility with conventional fuel and high
octane and cetane numbers. It is important to note that octane num-
∗
Corresponding author. Tel.: +82 42 866 1634; fax: +82 42 866 1649.
E-mail addresses: C1L031@gscaltex.co.kr, infilimonov@gmail.com
(I.N. Filimonov).
0926-860X/$ – see front matter © 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.apcata.2010.10.014

236
H.J. Lee et al. / Applied Catalysis A: General 390 (2010) 235–244
were used for the estimation of the degree of ion exchange. The
notation of modified Amberlyst follows the example: Na(50)Amb
means 50% replacement of acidic protons by Na+ cations.
Cesium salt of silicotungstic acid, Cs2.5H1.5SiW12O40 was
obtained according to the procedure of Okuhara et al. [17]. Aqueous
solution of Cs2CO3 (0.2 M) was added to the appropriate amount of
H4SiW12O40 (0.08 M) at a pump rate of 0.1 ml/min under stirring.
Resulting precipitate was dried at 85 ◦C overnight. The concentra-
tion of heteropolyacid in the solution was estimated by the titration
procedure described in Ref. [18].
2.3. Characterization of samples
The distribution of metals in Amberlyst matrix was checked by
energy-dispersive X-ray (EDX) spectroscopy using Quanta FEG 250
scanning electron microscope.
Fig. 1. Structure of catalysts used in the study: (A) p-toluenesulfonic acid (TsOH); (B)
Amberlyst 15 (H-Amb); (C) 1-(4-sulfobutyl)-3-hexylimidazolium trifluoromethane-
sulfonate (IL-6); (D) Keggin unit of silicotungstic acid H4SiW12O40 (HSiW).
The interaction of the solid samples with isobutene was esti-
mated using pulse technique in Autochem II 2920 automated
catalyst characterization system (Micromeritics). Before analysis
samples were pretreated in helium flow for 1 h at 110 ◦C. A train
of 10 pulses of isobutene (0.5 ml each in helium flow 20 ml/min)
was applied at 0 ◦C, 30 ◦C, 60 ◦C and the response was followed by
TCD. All condensable materials in the effluent gas were trapped
at 0 ◦C. The data on catalyst characterization are presented in
Supplementary Materials.
bers of di- and tri-ethers are very close. On the other hand, glycerol
is relatively cheap and renewable component, while isobutylene
is more expensive and is not renewable. This simple consideration
reveals the necessity of manufacturing preferentially di-ether pool
of GTBE.
Recently we have tried to find the ways for a more selective
production of individual ethers in GTBE process.
This publication is devoted to the comparison of different types
of catalysts; applicability of the general reaction scheme (reactions
(1)–(3)) for the catalysts under study; and evaluation of the cat-
alytic activity of partially neutralized Amberlyst-15.
2.4. Catalytic tests
Etherification of glycerol was performed in stainless steel stirred
autoclaves (80 ml each), equipped with thermocouples, cooling
coil and sampling valve. Each autoclave was equipped with 6-
blade mechanical stirrer (100–1200 rpm). Before the experiment
the reactor was flushed with nitrogen and adjusted to 20 bar
when introducing liquid isobutylene. In typical runs conditions
of Klepácˇová et al. [9] were used: 11.50 g (0.125 mol) of glycerol,
0.86 g dry catalyst (7.5 wt%/glycerol), 28.2 g (0.503 mol) isobuty-
lene; 60 ◦C; 1200 rpm (above 1000 rpm the product distribution
does not depend on stirring speed). These conditions will be termed
as standard throughout the paper. All variations of conditions with
respect to catalyst, feed composition, temperature, etc. are reflected
in the text.
2. Experimental
2.1. Materials
Glycerol (99%), isobutylene (99%), p-toluenesulfonic acid,
Amberlyst-15 beads were obtained from Sigma–Aldrich; silico-
tungstic acid hydrate was obtained from Fluka. All the chemicals
were used as received. Supplementary solvents (methanol, diox-
ane, and others) were analytical grade purity.
2.2. Catalysts
Since reaction system initially separates into two liquid phases
(glycerol/isobutylene) and typically becomes visibly homogeneous
within 1–2 h, the most part of the data in this paper represent
homogeneous case. In certain cases e.g., when it is important to ana-
lyze initial conversions or slow reactions, the process was stopped
and the autoclave was quickly cooled to the room temperature and
depressurized. The phases were separated, their weight, volume
and composition were determined separately.
Ionic liquid with Brønsted acidity: 1-(4-sulfobutyl)-3-
hexylimidazolium trifluoromethanesulfonate was synthesized
according to the known protocols [14,15]. In brief, N-
hexylimidazole was prepared from imidazole and freshly distilled
1-bromohexane using ethanol solution of sodium ethoxide as a
base [16]. Equimolar amounts of N-hexylimidazole and 1,4-butane
sultone were stirred solvent free at room temperature for 3 days.
The solid mass formed was washed three times with toluene and
dried under vacuum. Thus obtained 3-hexyl-1-(butyl-4-sulfonate)-
imidazolium betaine was reacted with trifluoromethanesulfonic
acid for 2 h at 40 ◦C under stirring. Viscous colored ionic liquid thus
obtained was washed with toluene and ether to remove non-ionic
residues, and dried under vacuum at 80 ◦C, ∼10 mm Hg. Structure
of IL-6 is presented in Fig. 1 together with the other types of
catalysts used in this study for a comparison.
2.5. Analysis
The samples of the reaction mixture were analyzed by GC in
two ways. Standard analysis of all the components was performed
in Agilent 7890 System equipped with Agilent DB-WAX column
(30 m × 0.32 mm × 0.25 ␮m). 1-Octanol was used as internal stan-
dard.
Amberlyst-15 beads were washed with deionized water (Milli-
Q), methanol and dried at 70 ◦C overnight. The content of acidic
protons was checked by potentiometric titration (Titrino 721
NET, Metrohm AG). Ion exchange was performed using diluted
(0.2–0.5 mol/L) solutions of metal nitrates or chlorides at ambient
temperature overnight under gentle stirring. Modified Amberlyst
was washed with deionized water, dried overnight at 70 ◦C, and
vacuum dried at 60 ◦C. Acid content of modified Amberlyst was
rechecked by potentiometric titration and the results of titration
In order to increase the sensitivity and reliability of determina-
tion of glycerol and its derivatives a modified procedure of Plank
and Lorbeer [19] was used. Glycerol and its mono- and di- ethers
were transformed to trimethylsilyl derivatives. In brief, the sam-
ples were 2–3-fold diluted with 1,4-dioxane (for homogeneous
non-viscous samples this step may be omitted); 20 ␮L of pyridine
with internal standard (either 1,2,4-butanetriol or 1,3-propanediol
were used as internal standards); 0.5–1.5 ␮L of the sample and
10 ␮L MSTFA (N-methyl-N-trimethylsilyltrifluoroacetamide) were

H.J. Lee et al. / Applied Catalysis A: General 390 (2010) 235–244
237
Table 1
Properties of the catalysts.
a
Catalyst (abbreviation)
H0 [reference]
Acidity, meq/g
Conversion of
glycerol within
5 hb
Selectivity to
DIB within 5 hb
p-Toluenesulfonic acid (TsOH)
Amberlyst-15 (H-Amb)
H4SiW12O40 (HSiW)
Cs2.5H1.5SiW12O40 (Cs2.5SiW)
−2.6 [20]
5.3
4.8
1.4
0.5
2.3
99
99
94
50
76
0.2
22
17
37
0.5
−2.2 [21]
−3.4 [20]
<−8.2 [22]
0.23c; −0.02d [23]
1-(4-Sulfobutyl)-3-hexylimidazolium trifluoromethanesulfonate (IL-6)
a
Hammet acidity values taken from the literature.
Catalytic run under the standard conditions.
Hammet acidity for 3-methylimidazolium analogue.
Hammet acidity for 3-decylimidazolium analogue.
b
c
d
placed into 1.5 mL GC vial, capped and left for 30 min at room tem-
perature for the completion of silylation. 1 mL of heptane was added
to the vial and thus obtained solution was analyzed using Agilent
HP-5 column (35 m × 0.32 mm × 0.25 ␮m).
are more pronounced for the Amberlyst with high density of the
acidic sites and cesium salt of heteropolyacid. It is necessary to note
that acid content of cesium salt per weight base is by the order of
magnitude lower than that of Amberlyst. Therefore, when normal-
ized by the acid content, Cs2.5SiW should have the highest affinity
for isobutene during the pulse experiment. This is a clear indication
that the content of the strong acid sites is not a key factor in the
process of isobutene interaction with a catalyst.
Density of acidic sites and nature of the metal cation seem to be
important. Despite the fact that we used increased amount of par-
tially neutralized Amberlyst to compensate the change of acidity,
Fig. 2 clearly shows the difference between partially neutralized
Amberlyst and unmodified Amberlyst. Analysis of the shape of
responses suggests that the effect of counter-cations cannot be
excluded. Ag(62)Amb shows very broad responses and almost com-
plete consumption of isobutene during the first pulse. Al(57)Amb
shows somewhat reduced amplitude of responses compared to the
other samples. Li-, Ba-, Mg- and Na-modified samples show com-
parable responses though the shape of the first pulse shows certain
dependence on the nature of metal cation as well.
According to Fig. 2 we can point out certain qualitative differ-
ences for the Amberlyst samples having comparable acid content
but different metal cations used for the ion-exchange. Quantita-
tive difference is presented in Fig. 3. For the sake of brevity we will
term isobutene consumed by the sample during the first pulse as
“consumption by the clean sample”, while the amount of isobutene
consumed during the 10th pulse as “consumption by the saturated
sample”. Among partially neutralized Amberlyst samples silver-
modified catalysts have the highest consumption by the clean
sample and the lowest consumption by the saturated sample. Al-
The response factors for all the individual ethers were deter-
mined using home-made high-purity samples. Mono- and di-ether
were prepared via epichlorohydrin route. Tri-ether was made via
excessive t-butylation of DTBGE by isobutylene and subsequent
distillation over metallic sodium under rough vacuum.
3. Results and discussion
3.1. Characterization of catalysts
Some characteristics of the catalysts are summarized in Table 1;
structural types of the catalysts are presented in Fig. 1. The distri-
bution of metal cations within the matrix of Amberlyst according
to EDX does not depend on the nature of cation (see Supplementary
Materials).
Interaction of isobutene with the solid samples was investi-
gated using a pulse technique. Typical TCD responses are presented
in Fig. 2. Absence of responses was observed for the original
Amberlyst 15 in acidic form and cesium salt of heteropolyacid
indicating very high affinities of these catalysts to isobutene. Inter-
action of isobutene with catalysts is exothermic reaction and clear
exo-effects are observed (see Supplementary Materials). Isobutene
consumption is a result of chemisorption of isobutene followed by
the secondary processes like oligomerization and these processes
Na(45)
Mg(52)
clean
0.20
0.15
0.10
0.05
0.00
saturated
Li(44)
Ba(58)
Al(57)
Ag(62)
Cs 2.5 SiW
H-Amb
0
10
20
30
40
Time, min
Al(57)
Li(44)
Mg(52) Na(45) Ba(58) Ag(62)
Fig. 2. Pulse responses for a train of 0.5 mL pulses of isobutene at 0 ◦C. Cs2.5SiW and
H-Amb taken in the amount of 0.4 g, all the other samples taken in the amount of
0.9 g.
Fig. 3. Comparison of isobutene consumption for the 1st pulse (“clean sample”) and
the 10th pulse (“saturated sample”).

238
H.J. Lee et al. / Applied Catalysis A: General 390 (2010) 235–244
0.25
0.20
0.15
0.10
A
Ag(94)
Ag(62)
clean
saturated
Al(97)
Al(57)
Na(69)
Na(45)
0.05
0.00
0.25
0
20
40
60
80
Ag(62) Ag(94) Al(57)
Al(97) Na(45) Na(69)
Time, min
Fig. 5. Effect of the metal loading (Ag, Al, and Na) on the shape of responses. Solid
lines – lower amount of metal, dashed lines – higher amount of metal.
B
clean
saturated
0.20
0.15
0.10
0.05
0.00
100
80
60
40
20
0
Ag(62) Ag(94) Al(57)
Al(97) Na(45) Na(69)
Fig. 4. Effect of the metal loading (Ag, Al, and Na) on the isobutene consumption by
“clean sample” and “saturated sample” at 0 ◦C (A) and 60 ◦C (B).
modified Amberlyst has the highest consumption of isobutene by
the saturated sample.
0
5
10
15
20
Time, h
Effects of temperature and metal content on the isobutene con-
sumption are presented in Fig. 4(A) and (B). Silver and aluminium
tend to increase consumption by the clean sample compared to
sodium-modified sample. Increase of the silver content does not
have suppressing effect on the consumption by the clean sample.
On the other hand, increase of metal content in the case of sodium
and aluminium clearly leads to the suppression of the interaction
of isobutene with Amberlyst. Partially Al-exchanged Amberlyst has
the highest consumption of isobutene by the saturated sample
among all the modified Amberlysts and the increase of temperature
induces this effect.
The shape of the responses in pulse experiments is sensitive to
the degree of ion-exchange for all the ions except silver (Fig. 5).
Ag(62)Amb and Ag(94)Amb show almost identical responses.
Taken all together, Ag- and Al-modified samples have certain
unique properties with respect to the interaction with isobutene
compared to the other samples. This leads to certain specific fea-
tures in catalysis as well (see lower).
Fig. 6. Conversion of glycerol under the standard conditions (IB:G = 4:1, 7.5% weight
of catalyst with respect to glycerol); empty symbols – heterogeneous catalysts, filled
symbols – homogeneous catalysts H-Amb beads 1 mm ( ), H-Amb powder (ꢀ),
TsOH (᭹), IL-6 (ꢁ), HSiW (ꢂ), Cs2.5SiW (♦) taken in the amount of 2 g.
Table 2
Selectivity to diisobutenes under standard conditions.
Catalyst
SDIB (2 h of reaction)
SDIB (22 h of reaction)
TsOH
0.3
14.7
0.7
14.3
33.8
0.6
5.0
1.0
0
0.9
H-Amb (powder)
H-Amb (beads)
HSiW
39.5
31.8
32.9
44.8
1.6
11.3
8.9
1.9
Cs2.5SiWa
IL-6
H-Amb (powder)b
Al(57)Amb
Mg(52)Amb
Ag(62)Amb
Na(45)Amb
1.0
0.8
11.0
3.2
3.2. Conversion of glycerol and oligomerization of isobutylene
Figures on DIB formation are presented in Supplementary Materials.
a
Catalyst taken in the amount of 2 g.
b
Catalyst taken in the amount corresponding to 10% of the standard.
Fig. 6 and Tables 1–2 show conversions of glycerol and selec-
tivities of DIB formation. In general, conversion exceeding 80% is

H.J. Lee et al. / Applied Catalysis A: General 390 (2010) 235–244
239
reached within 2 h under the standard conditions with two excep-
tions: Amberlyst-15 in the form of the beads (1 mm in diameter)
and insoluble cesium salt of silicotungstic acid. Amberlyst-15 in
the form of the beads shows a lag period due to severe inter-
nal diffusion limitations. This period of a very slow reaction may
be explained by a slow transport of reactants within the cross-
linked matrix of Amberlyst-15. After 5 h of reaction, when sufficient
amount of mono-ether (MTBGE) is accumulated, the rate of ether-
ification increases due to the solvent effect. MTBGE is a very good
solvent both for glycerol and isobutylene [11] and facilitates the
transport of reactants within the pores. The more MTBGE is accu-
mulated, the higher is the reaction rate, and after 7 h of reaction
almost all glycerol is converted by the beads of Amberlyst.
Particle size of Amberlyst is of crucial importance for the fast
glycerol conversion. When beads of Amberlyst-15 are ground into
the fine powder, internal diffusion control becomes negligible and
complete conversion of glycerol occurs within 30 min (Fig. 6). Pow-
dered Amberlyst-15 has superior rate of conversion per weight
unit among all the catalysts studied and considerably more active
than its homogeneous analogue, p-toluenesulfonic acid (Fig. 6). This
effect may be attributed to (i) higher local concentration of the
acidic sites (discussed lower) and (ii) to the increased concentration
of isobutylene within the hydrophobic backbone of Amberlyst-
15. Both effects lead to the enhanced formation of isobutylene
oligomers. Indeed, according to Table 2 Amberlyst-15 demon-
strates much higher rate of isobutylene oligomerization compared
to p-toluenesulfonic acid.
therefore we expect high probability of isobutene oligomerization
over H-Amb. For homogeneous acids with moderate acidity (TsOH
and IL-6) lower concentration of alcohol is sufficient for the sup-
pression of isobutylene oligomerization. Hydrophobic interaction
is an additional factor that may play some role in the oligomer-
ization and probably needs some speculation. Addition of cesium
induces hydrophobicity of the surface of the heteropolyacid salt
[31], as well as hydrophobic divinylbenzene matrix of Amberlyst
may favor adsorption of isobutene and may promote the formation
of DIB.
Finally, our pulse experiments on the interaction of isobutene
with catalysts (Fig. 2) revealed the highest affinity of isobutene to
Cs2.5SiW and H-Amb among all the catalysts studied and provide a
clear interpretation of catalytic data.
3.3. Product distribution
The distribution of products depends on the ratio of isobutylene
to glycerol (Fig. 7A). Mostly tri- and di-ethers are formed at high
IB/G ratio and mostly mono- and di-ethers are formed at low IB/G
ratio. It is difficult to obtain single ether at the high level of glycerol
conversion. If the equilibrium in reactions ((1)–(3)) is established,
then the distribution of glycerol ethers should not be dependent on
the nature ofthe catalyst. Wecan see thatafter20 h of reaction there
is still considerable difference in the product distribution (Fig. 7B).
Therefore, the system is not in the equilibrium in general even after
long reaction runs.
Insoluble cesium salt of heteropolyacid (Cs2.5SiW) has the poor-
est performance among all the catalysts studied (Fig. 6). Conversion
of glycerol is never complete even when high catalyst loading
is used. The reaction mixture remains all the time biphasic. The
lower phase is comprised mostly of viscous glycerol-rich media
with particles of catalyst, and the upper phase is transparent and
is comprised mostly of higher ethers of glycerol and oligomers of
isobutylene. Cs2.5SiW provides the highest activity in side reaction
of isobutylene oligomerization according to Table 2. GC analy-
sis revealed as well formation of many unidentified peaks that
probably belong to the condensed molecules of glycerol. Evidently
the introduction of Cs+ cations leads to the creation of basic sites
that are facilitating condensation reactions of glycerol. These high
molecular weight products probably deactivate the catalyst within
ca. 1 h of catalytic run.
In contrast to Cs2.5SiW, parent homogeneous silicotungstic acid
(HSiW) is very active and glycerol conversion is over 90% after 1 h
of catalytic run. HSiW shows much higher yields DIB in contrast to
the other homogeneous catalysts: TsOH and IL-6 (Table 2). Accord-
ing to theoretic and experimental studies [24,25] there is quite
high affinity of isobutylene for Keggin anion due to the protonation
of isobutylene and formation of the alkoxy state bound to Keggin
anion. These bound alkoxy-species may couple with a next isobuty-
lene molecule [26] producing dimeric species. Another route for
dimerization is through ␲-bound state of olefin found for the cat-
alysts with strong acid sites [27]. According to Table 1 HSiW and
Cs2.5SiW have the strongest acidity in terms of Hammett function,
and thus tend to promote isobutylene oligomerization.
Figs. 8 and 9 show time-dependent yields of mono-, di-, and tri-
ethers of glycerol. It is important to note that experimental patterns
in general do not follow the simplified kinetic model developed by
Behr and Obendorf [11] for the same process. According to reactions
(1)–(3), the following rates may be written:
r1 = k1CG · CIB − k−1 · CMTBGE
r2 = k2CMTBGE · CIB − k−2 · CDTBGE
r3 = k3CDTBGE · CIB − k−3 · CTTBGE
(6)
(7)
(8)
Concentration profiles of ethers then should be found according
to the system of equations:
dCG
= −r1
(9)
(10)
(11)
(12)
(13)
dt
dCMTBGE
dt
= r1 − r2
= r2 − r3
= r3
dCDTBGE
dt
dCTTBGE
dt
dCIB
= −r1 − r2 − r3
dt
Fig. 10 shows example of satisfactory fit in the case of TsOH
and example of unsatisfactory fit demonstrated by IL-6. Powdered
Amberlyst shows clearly deviations from the expected patterns
(Fig. 8). In this case almost immediate formation of the reaction
mixture (∼10 mol% MTBGE, ∼60 mol% DTBGE and ∼30 mol% TTBGE)
occurs followed by the gradual decay of TTBGE and corresponding
rise of MTBGE concentration. After 20 h of reaction the composi-
tion of the reaction mixture changes to ∼30 mol% MTBGE, ∼55 mol%
DTBGE and ∼15 mol% TTBGE. This behavior may be attributed to a
very high activity of powdered Amberlyst in the target and side
reaction. We observed a temperature spike from ca. 55 ◦C to ca.
65 ◦C during initial 15 min of catalytic run when the autoclave is
heated from room temperature to 60 ◦C. This temperature spike is
short (about 5–10 min) and is observed only in the case of pow-
dered Amberlyst. All the other catalysts show smooth temperature
Amberlyst has lower acid strength compared to heteropoly-
acid, nevertheless it shows elevated yields of DIB. According to [28]
isobutene dimerization is a second order reaction with respect to
the Amberlyst acid concentration. Spatially fixed sulfonic groups
of Amberlyst provide higher local concentration of acidic protons
compared to homogeneous case and thus the probability of IB
dimerization is increased. Higher local concentration of acidic sites
may be attributed to Cs2.5SiW where acidic centers are confined.
Alcohols are suppressing oligomerization of olefins [28,29], but
complete suppression of DIB formation in the case of Amberlyst
is achieved at ratio olefin/alcohol < 0.6 [30]. In our study IB/G = 4,

240
H.J. Lee et al. / Applied Catalysis A: General 390 (2010) 235–244
100
80
60
40
20
0
A
A
60
40
20
0
B
4:1
3:1
2:1
1.5:1
60
IB:G ratio
S(MTBGE)
S(DTBGE)
S(TTBGE)
G conv. %
55
50
45
40
35
100
80
60
40
20
0
B
C
40
30
20
10
0
H-Amb
TsOH
HSiW Cs2.5HSiW
IL-6
Catalyst
Fig. 7. (A) Conversion of glycerol and selectivity to glycerol ethers after 1 h of reac-
tion; H-Amb powder-catalyst, variable ratio of IB:G; standard conditions (P = 20 bar,
7.5% weight of catalyst with respect to glycerol). (B) Conversion of glycerol and
selectivity to glycerol ethers after 20 h of reaction, ratio of IB:G = 4:1; standard con-
ditions: P = 20 bar, 7.5% weight of catalyst (with exception of Cs2.5SiW taken in the
amount corresponding 17% weight with respect to glycerol).
0
5
10
Time, h
15
20
profile during the onset of the catalytic run. This is a clear indica-
tion that reaction starts violently and finishes within 10 min when
it is driven by powdered Amberlyst. As a result the system arrives
to the equilibrium (or overshoots the equilibrium) very fast but
then, due to the steady process of isobutylene removal in the form
of DIB, the equilibrium is perturbed to the left in a more gradual
manner. This is indicated by the decay of TTBGE and slow accumu-
lation of MTBGE (Fig. 8). In contrast to H-Amb, TsOH with higher
content of acidic sites of comparable strength shows the concentra-
tion profiles that agree with the expected simplified kinetic scheme
presented in Fig. 10A. There is no decay of TTBGE since isobutylene
removal in the form of DIB is negligible for TsOH-driven reaction
(Table 2).
Fig. 8. Yields of glycerol ethers under the standard conditions (P = 20 bar, IB:G = 4:1,
7.5% weight of catalyst with respect to glycerol); empty symbols – heterogeneous
catalysts, filled symbols – homogeneous catalysts H-Amb powder (ꢀ), TsOH (᭹).
(A) Yield of MTBGE; (B) yield of DTBGE; (C) yield of TTBGE.
accumulation of DTBGE, the TTBGE profile shows a relatively
long lag-period in the beginning of the run. Though DTBGE con-
centration reaches maximum within 2 h, formation of TTBGE is
suppressed to ∼5 mol% for initial 7 h of reaction. Only after 20 h
the concentration of TTBGE increases up to 35 mol%.
Ionic liquid demonstrates the same feature of suppressed TTBGE
formation during the first 7 h of reaction. DTBGE and MTBGE pro-
files have plateaus during initial 3 h of the run (Fig. 9). These
lag-periods and plateaus probably correspond to the solvent effects
Heteropolyacid exhibits concentration profiles that cannot be
fitted by the simple kinetic scheme. TTBGE profile in this case
has certain deviation from the expected pattern. Despite the fast

H.J. Lee et al. / Applied Catalysis A: General 390 (2010) 235–244
241
in higher ethers of glycerol (Fig. 11). Plateaus observed in MTBGE
and DTBE profiles in the case of IL-6 are less clear for the inter-
pretation. Solubility of IL-6 in MTBGE is good therefore, the solvent
effect is excluded. It is clear that initial conversion of glycerol is fast
until 60% of glycerol is converted. Then etherification of the second
hydroxyl-group of glycerol becomes slower according to the slow
rate of MTBGE decay (Fig. 9A) demonstrated by both IL-6 and HSiW.
We may speculate on the possible changes of the system of
hydrogen bonds when bulky t-butyl groups appear near the active
centers. According to the common views on the mechanism of
etherification of alcohols by olefins, concerted action of several pro-
tons is required resulting in the very complicated ensemble in the
transition state [30,32]. When t-butyl group is inserted into the
molecule of glycerol it may distort the ensemble of the transition
state and change the rate constant. Probably the same changes in
the structure of “active ensemble”, when ca. 60 mol% of MTBGE is
converted into DTBGE, lead to the corresponding sigmoid shape of
TTBGE profile.
A
60
40
20
0
B
60
Product distribution in the case of insoluble Cs2.5SiW salt should
be treated with caution (or in a semi-quantitative manner) since
in this case we get considerable errors due to the formation of
a big number of unidentified products. Excessive oligomerization
of isobutylene may promote glycerol etherification by a variety
of IB oligomers. Glycerol condensation to polyglycerols cannot be
excluded as well. We expect that these high molecular weight prod-
ucts lead to the fast deactivation of the catalyst.
50
40
30
20
3.4. Recycling of the catalyst
Amberlyst-15 may be recycled several times without any loss of
activity. This catalyst may be recovered by filtration, washing with
methanol and drying.
p-Toluenesulfonic acid is more difficult to recycle. For exam-
ple, it may be extracted by glycerol from the product mixture, but
still traces of p-toluenesulfonic could not be removed. The latter
deteriorates the yields of ethers during the distillation step [8].
Catalytic activity of Cs2.5SiW vanishes after a single catalytic run
indicating severe deactivation.
Heteropolyacids and ionic liquids are easily separated from the
products due to their low solubility in higher ethers of glycerol.
They form liquid lower phase or solid sediment in the case of HSiW,
which may be removed via simple decantation (Fig. 11). Though
phase separation is easily obtained, gradual loss of the catalyst
occurs during multiple runs. This gradual loss is attributed to the
small solubility of these acids in the product mixture. Treatment of
the products with glycerol greatly improves the recovery of these
catalysts.
10
40
C
30
20
10
0
3.5. Partially neutralized Amberlyst-15
0
5
10
15
20
Fine powder of Amberlyst-15 provides the best activity and may
be easily separated from the products. Unfortunately, it gives ele-
vated levels of side products due to isobutylene oligomerization
(Table 2). Behr et al. [8] also reported higher levels of IB oligomer-
ization for heteropolyacid and Amberlyst compared to TsOH.
We tried to modify the density of acidic sites in order to sup-
press the formation of DIB and other oligomers. Ion exchange using
aqueous solutions containing single-charged cations (Na+, Ag+),
Time, h
Fig. 9. Yields of glycerol ethers under the standard conditions (P = 20 bar, IB:G = 4:1,
7.5% weight of catalyst with respect to glycerol); empty symbols – heterogeneous
catalysts, filled symbols – homogeneous catalysts IL-6 (ꢁ), HSiW (ꢂ), Cs2.5SiW (♦)
taken in the amount of corresponding to 17% with respect to glycerol. (A) Yield of
MTBGE; (B) yield of DTBGE; (C) yield of TTBGE.
and cannot be fitted according to the simplified kinetic scheme
(Fig. 10B). Ionic liquid has good solubility in glycerol and MTBGE,
but poor solubility in higher ethers of glycerol (Fig. 11). Thus in the
beginning, the reaction of glycerol etherification is fast and ∼75%
of glycerol is converted into MTBGE and DTBGE. DTBGE is expelled
from the polar phase rich in catalyst resulting in a very slow reaction
of TTBGE formation. The same explanation holds for heteropolyacid
since it is as well soluble in glycerol and MTBGE, but not soluble
double-charged cations (Mg2+), and triple charged cations (Al3+
)
diminished the number of acidic sites and resulted in the loss of
catalytic activity. Conversions of glycerol catalyzed by partially
neutralized Amberlyst samples are above 80% within 2 h and above
90% within 4 h of reaction (see Supplementary Materials for the
illustration). All the catalysts with 50% reduced acidity show com-
parable rate of conversion. This rate of conversion is equivalent to
ca. 10% of the standard amount of H-Amb. Ag(62)Amb has the high-

242
H.J. Lee et al. / Applied Catalysis A: General 390 (2010) 235–244
2.0
1.5
1.0
0.5
0.0
2.0
A
B
1.5
1.0
0.5
0.0
0
5
10
15
20
0
5
10
15
20
Time, h
Time, h
Fig. 10. Experimental concentrations of glycerol (᭹), MTBGE (♦), DTBGE (ꢃ),and TTBGE (ꢁ) fitted by the simplified kinetic model (Eqs. (6)–(13)). (A) Satisfactory fit in case
of TsOH. (B) Unsatisfactory fit in case of IL-6.
est rate of etherification compared to the other partially neutralized
samples of Amberlyst.
The effect of “overshooting” the equilibrium is more pro-
nounced for the Ag-modified Amberlyst. This effect is further
enhanced when the loading of catalyst is increased 4 times (see
Supplementary Materials). In this case the overall amount of acidic
sites is 2 times higher compared to the standard amount of unmod-
ified Amberlyst. According to the data on the consumption of
isobutene during the pulse experiments, silver-modified Amberlyst
catalysts show enhanced consumption of isobutene by the clean
sample and enhanced consumption of isobutene at high metal
loadings. While Al(97)Amb sample has no catalytic activity in the
target reaction, Ag(94)Amb sample has measurable catalytic activ-
ity despite very low content of acidic sites. Enhanced adsorption of
ethylene by Ag-modified Amberlyst 15 was revealed by Wu et al.
studied in equilibrium adsorption–desorption mode [37]. Ag+ ions
are often used in membranes for the separation of olefin/paraffin
mixtures [38,39] due to the specific interaction of olefins with Ag+.
Then it is possible to explain the mechanism of “overshooting”
the equilibrium due to the increased possibility of simultaneous
etherification of mono-ether by two isobutylene molecules produc-
ing tri-ether in one step. Since Ag+ ions have increased affinity to
isobutene, and then this effect is more evident for Ag(62)Amb sam-
ple. Probably the same mechanism operates for Al(57)Amb since
this sample also revealed enhanced interaction with isobutene in
pulse experiments. Al(57)Amb, unlike the other modified samples,
shows wide distribution of weak acidic sites. Probably these weaker
sites have certain affinity to olefin molecules.
According to Figs. 8, 9 and 12 there is certain possibility to vary
the yields of individual ethers using suitable catalysts or fine tun-
ing H-Amb by partial ion exchange. It is clear that certain variations
of the individual rates (i.e., rates of reactions (1)–(3)) are observed
according to different product distributions on different catalysts
with the same number of acidic sites. Different counter-cations
used for the neutralization of H-Amb are changing the reaction
rates in the individual manner. According to the literature, partial
neutralization of Amberlyst 31 revealed the influence of the nature
of amine, taken as neutralizing agent, on the catalytic activity in
esterification [40]. Therefore, the nature of the cation used for the
ion-exchange and the distribution of the remaining acidic sites are
important factors for fine tuning the catalytic activity.
Selectivity to DIB exhibited by partially neutralized Amberlyst
samples is presented in Table 2. All the samples show lower rates
of DIB formation compared to parent H-Amb. Ag(62)Amb and
Al(57)Amb have higher level of DIB formation compared to Na- and
Mg-samples. This catalytic effect clearly corroborates our data on
the specific features of isobutene interaction with these catalysts
during pulse experiments.
Yields of MTBE, DTBGE, and TTBGE exhibited by modified
Amberlyst are summarized in Fig. 12(A)–(C). Comparison with the
unmodified H-Amb indicates that 50% reduction of acidic sites
caused by ion-exchange corresponds to 90% reduction of produc-
tion rate of ethers. According to Panneman and Beenackers [33]
90% decrease of acidity caused by the neutralization of Amberlyst
leads to 200 times lower production rate of MTBE formation from
methanol and isobutene. Second order with respect to TsOH con-
centration was found by Kogelbauer et al. [34] for the acid catalyzed
hydrolysis of MTBE. Ancillotti et al. found higher than third order
reaction rate with respect to the acid content of Amberlyst for
alcohol etherification by isobutylene [35,36]. All these features in
general imply that associates of acidic sites act as catalytic center.
Analysis of the yields of tri-ethers (Fig. 12C) reveals that in the
case of Na(45)Amb and Mg(52)Amb the concentration of TTBGE is
gradually increasing until the probable equilibrium level. But in the
case of Al(57)Amb and especially Ag(62)Amb TTBGE concentration
“overshoots” the equilibrium level and after reaching maximum
it gradually decreases. The same but “mirrored” effect is clearly
observed for MTBGE (Fig. 12A). The yields of MTBGE go below the
equilibrium level for Al(57)Amb and Ag(62)Amb samples.
Cation exchange decreases the density of acidic sites in
Amberlyst and deteriorates the catalysis by associated acidic pro-
tons leading to the non-linear dependence of the reaction rates
upon the concentration of acid centers in Amberlyst. The rates of
Fig. 11. Liquid phase separation of products and catalysts: left – HSiW, right – IL-6.

H.J. Lee et al. / Applied Catalysis A: General 390 (2010) 235–244
243
80
60
40
20
4. Conclusions
A
Comparison of the performances of different solid and liquid
acids in the etherification of glycerol by liquid isobutylene in a batch
reactor was performed and revealed that equilibrium is not reached
in general case.
The most active catalyst per weight basis was fine powder of
Amberlyst-15. It may be “fine tuned” using partial exchange of
acidic protons with suitable metal cations. This tuning leads to the
suppression of isobutylene oligomerization and provides a possi-
bility to vary the product distribution of glycerol ethers.
Some liquid acids (heteropolyacid and ionic liquid with
Brønsted acidity) are immiscible/insoluble in higher ethers and
thus have unique feature of suppressing the formation of tri-ethers
of glycerol.
These catalytic characteristics give the possibility to tune within
certain limits the yields of mono-, di-, and tri-ethers of glycerol at
fixed IB/G ratios.
0
80
B
Appendix A. Supplementary data
60
40
20
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.apcata.2010.10.014.
References
[1] A. Behr, J. Eilting, K. Irawadi, J. Leschinski, F. Lindner, Green Chem. 10 (2008)
13–30.
[2] C.-H.C. Zhou, J.N. Beltramini, Y.-X. Fan, G.Q.M. Lu, Chem. Soc. Rev. 37 (2008)
527–549.
[3] M. Pagliaro, R. Ciriminna, H. Kimura, M. Rossi, C.D. Pina, Angew. Chem., Int. Ed.
46 (2007) 4434–4440.
[4] D.T. Johnson, K.A. Taconi, Environ. Prog. 26 (2007) 338–348.
[5] A. Corma, S. Iborra, A. Velty, Chem. Rev. 107 (2007) 2411–2502.
[6] N. Rahmat, A.Z. Abdullah, A.R. Mohamed, Renew. Sustain. Energy Rev. 14 (2010)
987–1000.
[7] R.S. Karinen, A.O.I. Krause, Appl. Catal. A 306 (2006) 128–133.
[8] A. Behr, L. Obendorf, Chem. Ing. Tech. 73 (2001) 1463–1467.
[9] K. Klepácˇová, D. Mravec, M. Bajus, Appl. Catal. A 294 (2005) 141–147.
[10] K. Klepácˇová, D. Mravec, A. Kaszonyi, M. Bajus, Appl. Catal. A 328 (2007) 1–13.
[11] A. Behr, L. Obendorf, Eng. Life Sci. 2 (2002) 185–189.
[12] J.A. Melero, G. Vicente, G. Morales, M. Paniagua, J.M. Moreno, R. Roldán, A.
Ezquerro, C. Pérez, Appl. Catal. A 346 (2008) 44–51.
C
40
30
20
10
0
[13] H. Noureddini, W.R. Daily, B.A. Hunt, Chem. Biomol. Eng. Res. Publ. 13 (1998)
121–129.
[14] A.C. Cole, J.L. Jensen, I. Ntai, K.L.T. Tran, K.J. Weaver, D.C. Forbes, J.H. Davis Jr., J.
Am. Chem. Soc. 124 (2002) 5962–5963.
[15] Y. Gu, F. Shi, Y. Deng, J. Mol. Catal. A 212 (2004) 71–75.
[16] P. Bonhote, A.P. Dias, N. Papageorgiou, K. Kalyanasundaram, M. Gratzel, Inorg.
Chem. 35 (1996) 1168–1178.
[17] T. Okuhara, T. Nishimura, H. Watanabe, M. Misono, J. Mol. Catal. 74 (1992)
247–256.
[18] S. Soled, S. Miseo, G. McVicker, W.E. Gates, A. Gutierrez, J. Paes, Catal. Today 36
(1997) 441–450.
[19] C. Plank, E. Lorbeer, J. Chromatogr. A 697 (1995) 461–468.
[20] R.S. Drago, J.A. Dias, T.O. Maier, J. Am. Chem. Soc. 119 (1997) 7702–7710.
[21] M.A. Harmer, Q. Sun, Appl. Catal. A 221 (2001) 45–62.
[22] Y. Izumi, M. Ogawa, K. Urabe, Appl. Catal. A 132 (1995) 127–140.
[23] Y. Zhao, J. Long, F. Deng, X. Liu, Z. Li, C. Xia, J. Peng, Catal. Commun. 10 (2009)
732–736.
0
5
10
15
20
Time, h
[24] M.J. Janik, R.J. Davis, M. Neurock, Catal. Today 105 (2005) 134–143.
[25] K.A. Campbell, M.J. Janik, R.J. Davis, M. Neurock, Langmuir 21 (2005)
4738–4745.
[26] S. Svelle, S. Kolboe, O. Swang, J. Phys. Chem. B 108 (2004) 2953–2962.
[27] H. Ishikawa, E. Yoda, J.N. Kondo, F. Wakabayashi, K. Domen, J. Phys. Chem. B
103 (1999) 5681–5686.
Fig. 12. Yields of glycerol ethers under the standard conditions (P = 20 bar,
IB:G = 4:1) for powered Amberlyst based catalysts: H-Amb, taken as 10% amount
with respect to standard conditions (ꢀ); Na(45)Amb (ꢄ); Ag(62)Amb (♦);
Mg(52)Amb (ꢃ); Al(57)Amb (ଝ). (A) Yield of MTBGE; (B) yield of DTBGE; (C) yield
of TTBGE.
[28] M.L. Honkela, A. Root, M. Lindblad, A.O.I. Krause, Appl. Catal. A 295 (2005)
216–223.
[29] V.J. Cruz, J.F. Izquierdo, F. Cunill, J. Tejero, M. Iborra, C. Fité, React. Funct. Polym.
65 (2005) 149–160.
[30] J. Tejero, F. Cunill, J.F. Izquierdo, M. Iborra, C. Fité, D. Parra, Appl. Catal. A 134
(1996) 21–36.
[31] T. Nakato, M. Kimura, S.-i. Nakata, T. Okuhara, Langmuir 14 (1998) 319–325.
[32] J. Tejero, F. Cunill, M. Iborra, J. Mol. Catal. 42 (1987) 257–268.
[33] H.-J. Panneman, A.A.C.M. Beenackers, Ind. Eng. Chem. Res. 34 (1995)
4318–4325.
individual reactions may have different dependence on the concen-
tration of acidic sites; and thus certain reactions (e.g., formation of
DIB) are suppressed more effectively. Solvent effect is another fac-
tor that plays important role in the catalysis by homogeneous acids.
Depending on the solubility in glycerol (starting material) and glyc-
erol ethers (product of reaction) different homogeneous catalysts
may provide different yields of higher ethers of glycerol.
[34] A. Kogelbauer, J. Reddick, D. Farcasiu, J. Mol. Catal. A 103 (1995) 31–41.

244
H.J. Lee et al. / Applied Catalysis A: General 390 (2010) 235–244
[35] F. Ancillotti, M. Massi Mauri, E. Pescarollo, L. Romagnoni, J. Mol. Catal. 4 (1978)
37–48.
[38] J.H. Kim, S.M. Park, J. Won, Y.S. Kang, J. Membr. Sci. 248 (2005) 171–
176.
[36] F. Ancillotti, M.M. Mauri, E. Pescarollo, J. Catal. 46 (1977) 49–57.
[37] Z. Wu, S.-S. Han, S.-H. Cho, J.-N. Kim, K.-T. Chue, R.T. Yang, Ind. Eng. Chem. Res.
36 (1997) 2749–2756.
[39] J.-S. Yang, G.-H. Hsiue, J. Membr. Sci. 138 (1998) 203–211.
[40] L. Holub, K. Jerˇábek, J. Mol. Catal. A 231 (2005) 21–26.