Vous êtes sur la page 1sur 9

ASIA-PACIFIC JOURNAL OF CHEMICAL ENGINEERING

Asia-Pac. J. Chem. Eng. 2015; 10: 598–606


Published online 9 June 2015 in Wiley Online Library
(wileyonlinelibrary.com) DOI: 10.1002/apj.1909

Special theme research article


Enhancing allyl alcohol selectivity in the catalytic conversion
of glycerol; influence of product distribution on the
subsequent epoxidation step
Luke Harvey, Gizelle Sánchez, Eric M. Kennedy and Michael Stockenhuber*
PRCfE, Chemical Engineering, School of Engineering, The University of Newcastle, Callaghan, New South Wales 2308, Australia

Received 30 November 2014; Revised 22 April 2015; Accepted 24 April 2015

ABSTRACT: Conversion of glycerol to allyl alcohol in a flow reactor and subsequent conversion to glycidol in a batch
reactor are discussed in this paper. To increase reaction selectivity in the glycerol to allyl alcohol reaction, the catalyst
was modified with alkali metal salts. The treatment of the catalyst (Fe/Al2O3) with potassium, caesium and rubidium salts
enhanced the allyl alcohol yield when a 35 wt% glycerol solution was used as feed. Furthermore, the addition of organic
reductants or hydrogen to the glycerol feed increased the allyl alcohol yield. Optimisation of allyl alcohol selectivity was
achieved by a combination of alkali metal cation modifiers and the addition of reductants. The influence of acid, base and
redox properties on selectivity is presented, focusing on the effect of these properties on product distribution. The effect
of by-products from the conversion of glycerol to allyl alcohol on the epoxidation reaction was examined to identify the ideal
reaction conditions for maximising the glycidol yield. Epoxidation of allyl alcohol by 30 wt% hydrogen peroxide was
conducted in a batch reactor at 333K using titanium silicate-1 catalyst. Conversion of allyl alcohol was high for all
conditions; however, hydrolysis of glycidol to glycerol reduces selectivity. © 2015 Curtin University of Technology and
John Wiley & Sons, Ltd.

KEYWORDS: glycerol conversion; allyl alcohol; glycidol; iron oxide; acrolein; acid-base properties

INTRODUCTION Mechanisms where glycerol is dehydrated to


acrolein, which is subsequently hydrogenated to allyl
Carbon dioxide emissions and depleting fossil fuel alcohol, have been reported over iron oxide catalysts
resources have been major drivers for the successful with high surface areas,[6] while alternative pathways
industrial scale-up of the transesterification process also leading to the selective formation of allyl alcohol
for the manufacture of fatty acid methyl esters. One have been proposed to occur over catalysts with low
of the primary concerns with respect to this alternative strength acid sites as zirconia–iron oxide.[7] In contrast,
route to synthetic fuels is the generation of reaction side production of acrolein at yields of up to 70% is usually
products.[1–3] In spite of the various applications of catalysed by solid acids.[8] This suggests that the
glycerol in other industries, the enormous production product distribution of the reaction depends on the
rates and high cost of purification[4] threaten the financial acid–base properties of the catalyst, which can be
viability of biodiesel manufacture. For every 10 kg of modified by, for example, the deposition of alkali metal
biodiesel being produced, roughly 1 kg of crude glycerol salts on the catalyst.[9] A multitude of basic compounds
is generated.[5] Conversion of glycerol to high-value has been used to modify the nature of the catalyst
products like allyl alcohol, acrolein or glycidol is seen support. Nevertheless, under certain conditions,
as a viable option for using glycerol as a raw product. treatment can even increase the acidity of the surface,
In this paper, we report how various methods of depending on the concentration of the metal cations
improving the selectivity from glycerol to allyl alcohol deposited on the support.[10] Konaka et al. deposited
impact the subsequent epoxidation reaction to glycidol. alkali metals on zirconia–iron oxide catalysts for
glycerol conversion,[11] while we previously reported
correlations between acidity/basicity and allyl alcohol
production.[12] In this contribution, the focus is to link
*Correspondence to: Michael Stockenhuber, PRCfE, Chemical the effects produced by alkali catalyst modification in
Engineering, School of Engineering, The University of Newcastle,
Callaghan, New South Wales 2308, Australia. E-mail: michael. the first reaction to the epoxidation of glycidol, because
stockenhuber@newcastle.edu.au the formation of certain impurities (promoted/impeded

© 2015 Curtin University of Technology and John Wiley & Sons, Ltd.
Curtin University is a trademark of Curtin University of Technology
Asia-Pacific Journal of Chemical Engineering INFLUENCE OF GLYCEROL CONVERSION ON GLYCIDOL EPOXIDATION 599

by changes in acidity/basicity) has found to influence similar conditions, with 100 mol% selectivity to glycidol
the epoxidation reaction.[13] The unsaturated alcohol and 87.7 mol% conversion of allyl alcohol at 293 K,
produced from glycerol may be epoxidised to produce equimolar allyl alcohol/H2O2 ratio.[29]
glycidol. Traditionally, organic hydroperoxides were Ally alcohol in particular has shown to be a
used as the oxygen source for this reaction in an aprotic challenging substrate for epoxidation reactions.
organic solvent such as benzene, catalysed by a range Substitution of the allyl alcohol pi bond by electron
of transition metal oxides with or without supports, in withdrawing groups significantly decreases the
particular silica-supported molybdenum or tungsten susceptibility of the bond to attack by an electrophilic
oxides.[14] The catalytic activity of many of these oxidising agent (in this case, hydrogen peroxide).[20,30]
transition metal catalysts was, however, attributed to This is most effectively illustrated by the relative
rapid leaching of metal ions from the catalyst surface, reactivity of allyl compounds featuring various
essentially producing a homogeneous catalyst, the substituents alpha to the allyl group, in the order of
parent catalyst being unable to be reactivated.[15] The methyl > chloro > hydroxyl;[30] however, steric factors
first truly heterogeneous catalyst for the selective have also shown to be key, in the case of cis-
epoxidation of unsaturated alcohols was silica- enantiomers and trans-enantiomers of crotyl alcohol.[31]
supported titania, initially developed by Shell[16] and In addition, the epoxide thus formed is highly
was found to not exhibit the same leaching problems vulnerable to ring-opening reactions and nucleophilic
as with earlier transition metal catalysts.[17] All silica- attack to yield glycerol or ether diols in the case of attack
supported metal oxide catalysts were hindered severely by water or alcohol solvents, respectively.
by water, because of competition with peroxide for the In this study, the effects of catalyst modification on
active sites and as such, oxidising agents were limited product distribution in the production of allyl alcohol
to hazardous organic hydroperoxides, in particular from glycerol are examined. Then, the direct effect of
tert-butyl hydroperoxide. The development of titanium those product distributions on the epoxidation of allyl
silicate with MFI structure (TS-1) in 1983 by Taramasso alcohol by TS-1 is studied.
et al.[18] led to the development of a heterogeneous
catalyst that was substantially less adversely affected by
water,[19–21] and as such, safer and more cost-effective MATERIALS AND METHODS
hydrogen peroxide could be used as an oxygen source.
In recent years, a variety of titanium-substituted A 13% iron oxide on alumina catalyst was synthesised
zeolites applicable to the epoxidation of unsaturated adopting the widely used solid impregnation
alcohols have been developed. Zeolite topologies technique.[12] Alumina spheres (Sasol) were added to
containing titanium such as Ti-MWW,[22] Ti-BEA[23] a solution of Fe(NO3)3 (Sigma-Aldrich) in methanol.
or Ti-SBA-15[24] have since been synthesised to The catalyst was calcined for 4 h at 673 K in air. This
address the limitation of TS-1 in that the MFI was followed by alkali treatment, which consisted of
accommodates only smaller substrates such as linear washing the catalyst with a 0.1-M solution of
olefins. Ti-MWW in particular has been shown to be, potassium hydroxide in methanol, caesium nitrate in
under certain conditions, superior to TS-1 in the water and rubidium nitrate in water (each solution for
epoxidation of allyl alcohol[25,26] although has attracted a fresh batch of the iron catalyst as prepared).
less attention industrially. This study focuses upon Following modification, the resultant catalysts were
catalytic systems with a view specifically to large-scale dried at 383 K for 2 h and then calcined for 4 h at
application, and as such, TS-1 has been used in this 673 K in air flow. Changes in the preparation of the
case because of its higher commercial significance. standard iron oxide catalyst were implemented in order
Other zeolite topologies such as Ti-MWW are to be to reduce the water content in the iron nitrate alcoholic
the subject of future work. solution. Catalytic reactions were carried out in a
Selectivity of the epoxidation reaction over TS-1 is vertical stainless steel plug flow reactor at atmospheric
highly influenced by the acidity of the catalyst. Post- pressure and 613 K with nitrogen as carrier gas. The
synthetic treatment of the catalyst with base enhances reactor was of cylindrical shape (60 × 1.5 cm internal
the selectivity to glycidol. Conversely, the Brønsted diameter × 1.9 cm outer diameter) using 10 g of catalyst
acid sites have been shown to enhance subsequent for every set of experiments. In experiments where
ring-opening reactions (solvolysis and/or hydrolysis) hydrogen (Coregas 99.99%) was employed as reducing
on the epoxide ring.[27] agent, the gas was introduced at 20 mL min1 using of
Reaction parameters optimising the epoxide yield (in a Brooks 5850E mass flow controller. The feed
this case for 2-methyl-2-propen-1-ol methallyl alcohol) (glycerol 35 wt% aqueous solution) was pumped at
are lower reaction temperatures (~293 K), equimolar 0.5 mL min1 and varying concentrations of reductant
substrate/H2O2 ratio, 70 wt% solvent concentration and additives (ammonia propanoic acid, acetic acid, formic
3 wt% catalyst concentration, as found by Wróblewska acid and oxalic acid). Liquid samples were taken every
et al.[28] TS-2 reaches optimum performance under very 30 min, and after dilution, product analysis was carried
© 2015 Curtin University of Technology and John Wiley & Sons, Ltd. Asia-Pac. J. Chem. Eng. 2015; 10: 598–606
DOI: 10.1002/apj
600 L. HARVEY ET AL. Asia-Pacific Journal of Chemical Engineering

out using a Hewlett Packard 5890A gas chromatograph 823 K for 4 h and heated with a ramp rate of 2 K per
equipped with a flame ionisation detector. Catalysts minute. X-ray diffraction using a Philips X’pert Pro
were analysed by nitrogen adsorption at 77 K. diffractometer with Cu Kα incident radiation
Isopropanol decomposition took place in a custom- (λ = 1.5406 Å) and that diffraction pattern have been
made apparatus,[32] where products of the reaction compared with samples synthesised by others[18] and
between 303 and 1073 K were monitored using ion have been published elsewhere.[13]
currents at m/z = 58 (acetone) and m/z = 42 (propene). The epoxidation reaction was conducted in a 25-mL
Peak fitting was performed assuming Gaussian three necked round-bottomed flask fitted with a
distribution. Isopropanol conversion was estimated condenser, thermocouple and septum. Feed solutions
using m/z = 60. Calculations for the heat of adsorption representative of the product streams obtained from
of ammonia were based on Eqn (1), assuming first- the glycerol conversion reaction with a range of
order kinetics:[33] modified iron oxide catalysts were reproduced from
neat chemicals (reactant compositions and suppliers in
T 2p E 1 Table 1).
In ¼ (1)
β R TP All epoxidation reactions were carried out at 333 K
in ethyl acetate solvent for 4 h, as per previous
where Tp is the temperature at which the desorption is experiments,[13] with equimolar amounts of allyl
maximum and β is the heating rate. The ion current alcohol and hydrogen peroxide (30 wt%, aqueous).
used for ammonia desorption was m/z = 16, while The quantity of TS-1 catalyst was kept constant relative
m/z = 44 was used for carbon dioxide desorption. to total reaction volume. First, TS-1 was added to the
Titanium silicate-1 catalyst was synthesised as per reaction flask, followed by ethyl acetate solvent and
Taramasso et al.[18] and reported previously by us[13] the feed mixture. Hydrogen peroxide was added last.
where 37.99 g (0.182 mol) of tetraethylorthosilicate Seventy-five microlitres of aliquots were taken at zero
(Merck) and 0.61 g (0.003 mol) tetratethylorthotitanate and 10 min after the addition of hydrogen peroxide,
(Merck) were mixed at 35 °C. The mixture was then then every hour following, until a total reaction time
cooled in an ice bath, and 39.54 g (0.194 mol) of of 4 h.
tetraethylammonium hydroxide template (40 wt% in The progress of the reaction was monitored using a
water, Merck) was added dropwise using a burette. Shimadzu 2014 gas chromatograph, fitted with an Rtx
The mixture was then heated to 353 K to evaporate wax column (30 m, with 1-m guard column) and flame
ethanol formed by hydrolysis of tetraethylorthosilicate, ionisation detector. Samples (1 μL) were injected at
and the volume returned to its initial level using 308 K with a split ratio of 10 : 1 and a carrier gas (He)
distilled water. The mixture was added to a 300-mL flow rate of 13 cm min1.[3] The injection temperature
Teflon-lined autoclave and heated to 175 °C for 2 days was held for 5 min, followed by a ramp of 10 K min1
with gentle stirring. The fine, white product was to 473 K and held for 6 min.
recovered and washed by centrifugation and dried The fractional conversion of allyl alcohol, XAA, was
overnight at 393 K before being calcined in air at calculated according to Eqn (2):

Table 1. Reaction feed composition and total allyl alcohol conversion for different catalyst treatments over 4 h
reaction time.

Iron oxide catalyst modification


Oxalic acid Formic acid Rb Ammonia Hydrogen Nil
Concentration (wt%)
Allyl alcohol (Sigma-Aldrich) 2.66 2.74 1.76 1.23 1.72 1.91
Acrolein (Sigma-Aldrich) 0.51 0.34 0.45 0.36 0.10 0.50
Ethanal (Merck) 1.28 0.73 0.20 1.18 0.61 0.50
Hydroxyacetone (Sigma-Aldrich) 1.07 0.88 2.27 1.43 0.45 0.90
Acetic acid (Univar) 1.13 1.08 0.40 1.13 0.81 0.40
Propanoic acid (Ajax Finechem) 1.69 1.32 0.81 1.28 1.01 1.31
Glycerol (Univar) 0.05 0.49 1.66 0.87 0.30 0.15
Watera 42.8 41.3 45.3 43.8 45.4 45.2
Solvent (EtOAc) (Fisher Scientific) 0.52 0.53 0.35 0.24 0.33 0.37
H2O2 (30 wt%)b (Chem Supply) 4.28 4.50 4.53 4.55 4.54 4.52
Titanium silicate-1 0.09 0.09 0.09 0.09 0.09 0.09
XAA 0.67 0.70 0.73 0.76 0.81 0.88
a
Does not include water contributed by hydrogen peroxide solution.
b
Total percentage of 30 wt% hydrogen peroxide solution added.
© 2015 Curtin University of Technology and John Wiley & Sons, Ltd. Asia-Pac. J. Chem. Eng. 2015; 10: 598–606
DOI: 10.1002/apj
Asia-Pacific Journal of Chemical Engineering INFLUENCE OF GLYCEROL CONVERSION ON GLYCIDOL EPOXIDATION 601
12
½AA0  ½AAt Allyl alcohol
X AA ¼ (2) Acrolein
½AA0 10

Yield (%)
where [AA] is the concentration of allyl alcohol and
glycidol (µg μL1) initially and at a given time 6

(indicated by subscripts 0 and t, respectively). 4

0
RESULTS AND DISCUSSION Rb K Cs Fe

Preliminary catalytic measurements using the alumina Figure 1. Effect of alkali metal catalyst treatment on allyl
support were conducted at 613 K at 1 atm for reaction alcohol and acrolein yield at 613 K. Reactant: 35 wt%
times as long as 5 h. Following 3 h of catalytic tests, glycerol. GHSV = 1190 h1. This figure is available in
acrolein is produced in moderate yields, while allyl colour online at www.apjChemEng.com.
alcohol production is low (Table 2). Similar molar
carbon selectivities to acrolein and allyl alcohol have
been reported previously at low conversions (17%).[34] adsorption) and base sites (determined by CO2
Glycerol conversion was close to 100% and remained adsorption) that directly correlates with selectivity
constant as a function on time on stream for 5 h. towards allyl alcohol and acrolein.[12] In accordance
Figure 1 shows allyl alcohol and acrolein yields with the literature, isopropanol can be used for both
following 3 h of catalytic measurements for the alkali acid and base site characterisation.[36] Over acid sites,
metal-treated and alkali metal-unmodified 13% iron dehydration to propene is observed, whereas base sites
oxide catalysts. Details on physicochemical properties dehydrogenate i-propanol to acetone.
of the iron catalyst have been provided elsewhere.[12] The addition of alkali metal cations to the catalysts
The presence of iron resulted in a product distribution resulted in reduced acidity as evident by the decreased
distinctly different to that listed on Table 2, obtained propene yields (Table 3). A drop in the heat of
operating at the same conditions over alumina. The desorption for the ammonia experiment over modified
increase in the rate of formation of allyl alcohol over catalysts (Table 4) suggests that less energy is
the iron catalyst suggests that redox active sites are a required in the desorption process, implying that the
requirement for the formation of the desired product. surface is less acidic than unmodified iron and
The decrease in the rate of formation of acrolein could alumina. Moreover, an increase in basicity was found
be explained by a reduction in the number of acid sites in the modified catalyst as observed by increased
due to iron deposition. If the catalyst is modified by an acetone yields when desorbing i-propanol (Table 3)
alkali salt, a further increase in the yield of allyl alcohol as well as increased CO2 desorption rate (not shown).
is observed (30–60%), presumably because of an We also observed variations of the acid site strength
increase in the surface basicity of alumina, as observed with the type of countercation, but the correlations are
by other authors.[35] The acid sites of the alumina are not straightforward. Enhanced catalyst activities appear
poisoned by the alkali metal cation; therefore, to be related to the larger atomic radius of the metal
selectivity towards acrolein is reduced, which is also used to modify the support. While we observed a
evident in Figure. general trend of increasing allyl alcohol yield with the
To test the hypothesis-proposed importance of size and thus the polarisability of the countercation,
acidic and basic sites on the catalyst, the modified the highest allyl alcohol yield was observed with Rb
catalysts were characterised using temperature- and not Cs. The acid site density and concentration
programmed desorption of ammonia, carbon dioxide were also affected in a similar manner, whereas the
and isopropanol.[12] There is a relationship between base site concentration was highest in K-modified
the number of acid sites (determined by ammonia materials.[12] Thus, it is suggested that the optimum

Table 2. Product distribution from the conversion of glycerol over alumina at 613 K.

Time on Allyl alcohol Acrolein Ethanal Hydroxyacetone Conversion


stream (min) yield (%) yield (%) yield (%) yield (%) (%)
120 2.5 19.1 10.1 5.2 99.9
180 1.4 14.4 4.5 6.6 99.9
300 1.5 15.3 4.6 8.6 99.8

Reactant: 35 wt% glycerol. Gas hourly space velocity (GHSV) = 1190 h1.
© 2015 Curtin University of Technology and John Wiley & Sons, Ltd. Asia-Pac. J. Chem. Eng. 2015; 10: 598–606
DOI: 10.1002/apj
602 L. HARVEY ET AL. Asia-Pacific Journal of Chemical Engineering
Table 3. Isopropanol temperature-programmed salts of Li+, Na+ and K+.[37] However, with the
desorption between 298 and 1073 K. deposition of the alkali metal, an increase in surface
i-propanol Propene Acetone area was observed, which was attributed to the removal
conversion yield yield of pore blocking material on the alumina. Differences
Catalyst (%) (%) (%) between the current findings and those from others[37]
Iron oxide on alumina 98.5 97.8 0.7 were explained by differences in catalyst preparation.
Rb-modified iron oxide 89.3 86.1 3.2 While Perrichon et al. deposited the alkali metal salts
on alumina through impregnation, the current catalysts were
modified by washing. Even though there is a
substantial surface loss after reaction using the
modified catalyst, this value is still 4% higher in
Table 4. Heat of desorption of ammonia. modified used catalysts than untreated used iron oxide
alumina catalysts.
Energy of desorption
Catalyst of ammonia (kJ mol) Deactivation of the iron on alumina catalyst was
observed, and conversion dropped to 76% following
Alumina 28.8 3 h of time on stream with an untreated catalyst. In
Iron oxide on alumina 28.0
Cs-modified iron oxide on alumina 27.9 contrast, for each of the three treated catalysts,
conversion levels remained at 87% following 3 h of
catalytic measurements (Fig. 3). This suggests that
acidity influences not only product distribution but also
catalyst stability. It is well known that coke can be
acid and base site concentration and strength subtly formed by interaction with acid sites, which affects the
influence the selectivity and yield to allyl alcohol and dehydrogenation route of glycerol to form acrolein.
acrolein. Consequently, deactivation reduces the rate of acrolein
In addition to the chemical effects, alkali treatment formation (yield decreased significantly with time on
influenced surface area of the catalyst. Figure 2 stream for the untreated catalysts, from approximately
discloses the nitrogen adsorption isotherms of alumina, 5% in the first hour of catalytic measurements to 3% in
unmodified iron catalyst and caesium-treated iron the following 2 h of test).
catalyst. In the absence of iron, an enhanced uptake A further increase in the selectivity and yield of allyl
of nitrogen was observed, particularly in the range alcohol was achieved following the addition of a
P/Po = 0.6–0.95. It is generally accepted that metal reductant to the reaction. From a purely stoichiometric
cation deposition on a support decreases its surface perspective, the reaction of glycerol formally requires
area, which was observed following impregnation of two hydrogen atoms (under the assumption of water
the alumina with the iron salt. Other authors found a formation) to be able to form allyl alcohol (Figure 4).
reduction on surface area when doping alumina with It is suggested that hydrogen is generated in situ, via
the dehydrogenation of intermediates formed at high
temperatures. Indeed, some hydrogen was detected in
400 the gas phase, in concentrations of approximately
Cs modified iron oxide on alumina
Iron oxide on alumina
0.8% v/v. We have thus investigated the addition of
350 -alumina
Volume adsorbed (cm3/g)

300 100
90
250 80
Conversion (%)

70
200
60
150 50
40
100
30
20
50
10
0 0
0 0.2 0.4 0.6 0.8 1 Rb K Cs Fe
P/Po
Figure 3. Glycerol conversion at 613 K over different
Figure 2. Nitrogen adsorption isotherms of alumina, catalysts following 3 h of catalytic measurements.
unmodified iron catalysts and caesium-treated iron Reactant: 35 wt% glycerol in water. GHSV = 1190 h1.
catalysts. This figure is available in colour online at This figure is available in colour online at www.
www.apjChemEng.com. apjChemEng.com.
© 2015 Curtin University of Technology and John Wiley & Sons, Ltd. Asia-Pac. J. Chem. Eng. 2015; 10: 598–606
DOI: 10.1002/apj
Asia-Pacific Journal of Chemical Engineering INFLUENCE OF GLYCEROL CONVERSION ON GLYCIDOL EPOXIDATION 603

Figure 4. Reaction mechanism for the transformation of glycerol to allyl alcohol. This figure is
available in colour online at www.apjChemEng.com.

reductants to the reaction and were able to increase hydrogen transfer reactions that led to the development
selectivity to allyl alcohol.[38] The addition of oxalic of a single-step reaction leading to the formation of
acid, a known reductant, doubled the yield of allyl allyl alcohol from glycerol.[6] The effect of added
alcohol over an unmodified iron catalyst. Formic acid hydrogen on both allyl alcohol yield and glycerol
had a similar effect, although propanoic and acetic acid conversion was studied in a molar ratio of the additive
did not have any effect on allyl alcohol yield. with respect to glycerol of 0.14 : 1. As evident in
Furthermore, the addition of ammonia, known for its Table 4, in the presence of hydrogen, conversion
reductive properties, had a positive effect on the yield remained constant following 3 h on stream. Moreover,
of allyl alcohol (a 35% increment). selectivity to coke decreased from 6.77 wt% over the
When molecular hydrogen is co-fed over bimetallic alumina-supported iron catalyst in the absence of
(nickel–copper) catalysts supported on alumina, as additives to 0.42 wt% when hydrogen was added.
well as over palladium, rhodium and copper-based Gandarias et al. reported previously lower rates of
catalysts supported on alumina, zinc oxide and carbon[39] catalyst deactivation for glycerol conversion in the
can produce 1,2-propanediol and 1,3-propanediol, presence of hydrogen.[41] A significant enhancement
respectively, from glycerol at high pressures. Noble in the yield of allyl alcohol is also observed when
metals are required for hydrogen activation in the compared with the absence of additives. The product
hydrogenolysis reactions of glycerol.[40] In some cases, distribution is summarised in terms of a product carbon
noble metals are employed in combination with an yield basis (Table 5). Hydroxyacetone and 3-
acid, or base or as promoters of metal oxide hydroxyprop-2-enal, which are formed by dehydration
catalysts.[40] Iron oxide has been reported to catalyse of glycerol, were hydrogenated to 1,2-propanediol and
© 2015 Curtin University of Technology and John Wiley & Sons, Ltd. Asia-Pac. J. Chem. Eng. 2015; 10: 598–606
DOI: 10.1002/apj
604 L. HARVEY ET AL. Asia-Pacific Journal of Chemical Engineering
Table 5. Product distribution for glycerol conversion little, if any, distinguishable effect. Similarly, acetic
in the presence of hydrogen over the alumina- acid and ethanal were found to enhance the conversion
supported iron catalyst, carbon yield basis.
and reaction rate; however, this effect was not observed
Allyl alcohol 12.45 here. The low concentration of reactants and impurities
Acrolein 4.42 most likely diminishes the effect additives that may
Acetaldehyde 4.25 have on allyl alcohol conversion.
Hydroxyacetone 6.50 In line with that observation, the concentration of
Acetic acid 3.72
Propanoic acid 6.91 glycidol, in all cases, decreases substantially after
Acrolein (gas phase) 0.91 60–120 min reaction time as hydrolysis kinetics begin
Acetaldehyde (gas phase) 0.78 to dominate the reaction and the glycerol concentration
Carbon dioxide 3.85 increases (Figs 5 and 6, respectively).
Carbon monoxide 0.89 No other by-products, most notably the ethers and
Unconverted glycerol 0.61
Coke 0.15 ether diols, were formed during the reaction, and this
Other compoundsa 45.45 has been explained previously by surmising that aprotic
solvents do not readily engage in ring-opening
a
Other compounds include 1,2-propanediol and 1,3-propanediol. reactions with glycidol.[13] The water, present in the
feed stream in high concentrations, does engage in
1,3-propanediol, and its contribution to the carbon ring-opening reactions on either of the highly strained
balance is accounted together with other compounds epoxide carbon atoms.
in Table 5. It is evidence that allyl alcohol yield can
be improved by the inclusion of reductants in the feed
and that the redox activity of the catalyst appears to
be responsible for the increase in yield. At the same
time, the addition of some reductants such as ammonia
reduces acrolein yield, which is important for the
subsequent reaction to glycidol (see the following
discussions). Without iron in the catalyst, very little
selectivity to allyl alcohol was observed, which
suggests that the redox activity of the transition metal
is responsible for the generation and/or reaction of
hydrogen with the intermediates. The combination of
the suitable acid and base sites as well as the redox
properties of the catalysts significantly alters the Figure 5. Glycidol concentration over course of
selectivity to allyl alcohol, and the yield can be reaction for reaction mixtures as produced by each
optimised by variation of these parameters. glycerol conversion catalyst modification. This figure
Previous work has shown that certain by-product is available in colour online at www.apjChemEng.
functionalities (such as aldehydes), including those com.
produced as by-products in the glycerol conversion
reaction, are detrimental to the epoxidation reaction,
whereas others may act as reaction promoters,[13] and
a possible mechanism to explain these observations
was developed. The effect of selected product
distributions, as obtained directly from the glycerol
conversion reaction, on the epoxidation of allyl alcohol
has been examined (Table 1).
The conversion of allyl alcohol was uniformly
relatively high, with the reaction feed mixture produced
by the unmodified catalyst being the most favourable,
at 88%. The total amount of allyl alcohol obtained from
the unmodified glycerol conversion catalyst, however,
is slightly less than half of that obtained from the
formic acid-modified catalyst. The total yield of
glycidol therefore is higher for the formic acid catalyst
feed mixture, despite the lower allyl alcohol Figure 6. Glycerol concentration over course of
reaction for reaction mixtures as produced by each
conversion. In our previous work,[13] acrolein was glycerol conversion catalyst modification. This figure
found to be exceptionally detrimental to allyl alcohol is available in colour online at www.apjChemEng.
conversion; however, in this case, there appears to be com.
© 2015 Curtin University of Technology and John Wiley & Sons, Ltd. Asia-Pac. J. Chem. Eng. 2015; 10: 598–606
DOI: 10.1002/apj
Asia-Pacific Journal of Chemical Engineering INFLUENCE OF GLYCEROL CONVERSION ON GLYCIDOL EPOXIDATION 605
[10] Fiedorow R, Dalla Lana IG. Interaction between hydroxides
CONCLUSIONS of alkali metals and acid centers on the surface of alumina.
J. Phys. Chem. 1980; 84(21): 2779–2782.
Catalyst acidity and basicity subtly influence allyl [11] Konaka A, Tago T, Yoshikawa T, Nakamura A, Masuda T.
alcohol and acrolein yields from glycerol. K, Rb and Conversion of glycerol into allyl alcohol over potassium-
supported zirconia-iron oxide catalyst. Appl. Catal. Environ.
Cs poison the catalyst acid sites and increase catalyst 2014; 146: 267–273.
basicity, improving selectivity towards allyl alcohol. [12] Sánchez G, Friggieri J, Keast C, Drewery M, Dlugogorski
The same treatment reduces catalyst deactivation and BZ, Kennedy E, Stockenhuber M. The effect of catalyst
modification on the conversion of glycerol to allyl alcohol.
increases surface area by 8%. Furthermore, the addition Appl. Catal. Environ. 2014; 152–153: 117–128.
of reductants can double the yield of allyl alcohol. [13] Harvey L, Kennedy E, Dlugogorski BZ, Stockenhuber M.
Catalysing the reaction of the precursors with added Influence of impurities on the epoxidation of allyl alcohol to
glycidol with hydrogen peroxide over titanium silicate
reductants is the main function of the redox site in these TS-1. Appl. Catal. Gen. 2015; 489: 241–246.
multifunctional oxide-based catalysts. [14] Sheldon RA. In Organic Peroxygen Chemistry. W Herrmann
The epoxidation of allyl alcohol to glycidol using (ed.). Springer: Berlin Heidelberg, 1993; Vol. 164, p 21–43.
[15] Trifirò F, Forzatti P, Preite S, Pasquon I. Liquid phase
feed directly obtained from the conversion of glycerol epoxidation of cyclohexene by tert-butyl hydroperoxide on
is feasible; however, the high water concentration leads a Mo-based catalyst. Journal of the Less Common Metals
to rapid hydrolysis of the target product once the 1974; 36(1–2): 319–328.
[16] Wulff HP. (Shell Oil Co., USA.). Epoxidation with improved
glycidol concentration is sufficiently high. It is heterogeneous catalyst. US3923843A, 1975.
recommended that allyl alcohol be removed from the [17] Kleemann A, Wagner RM. Glycidol: properties, reactions,
feed via distillation in order to mitigate glycidol applications, Hüthig, 1981.
[18] Perego G, Taramasso M, Notari B. (Snamprogetti SpA,
hydrolysis. Italy.). Porous crystalline synthetic material consisting of
silicon and titanium oxides. BE886812A1, 1981.
[19] Hutchings GJ, Lee DF, Minihan AR. Epoxidation of allyl
alcohol to glycidol using titanium silicate TS-1: effect of
the method of preparation. Catal. Lett. 1995; 33: 369–385.
Acknowledgements [20] Hutchings GJ, Lee DFARM. Epoxidation of allyl alcohol to
glycidol using titanium silicate TS-1: effect of the reaction
Financial support by AEL and the ARC is gratefully conditions and catalyst acidity. Catal. Lett. 1996; 39:
83–90.
acknowledged. Luke Harvey and Gizelle Sanchez [21] Wróblewska A, Fajdek A. Epoxidation of allyl alcohol to
would like to thank the University of Newcastle for glycidol over the microporous TS-1 catalyst. J. Hazard.
their postgraduate research scholarships. Mater. 2010; 179(1–3): 258–265.
[22] Wu P, Tatsumi T. A novel titanosilicate with MWW
structure: III. Highly efficient and selective production of
glycidol through epoxidation of allyl alcohol with H2O2.
J. Catal. 2003; 214(2): 317–326.
[23] Adam W, Corma A, Martínez A, Mitchell CM, Indrasena
REFERENCES Reddy T, Renz M, Smerz AK. Diastereoselective epoxidation
of allylic alcohols with hydrogen peroxide catalyzed by
[1] Leung DYC, Wu X, Leung MKH. A review on biodiesel titanium-containing zeolites or methyltrioxorhenium versus
production using catalyzed transesterification. Applied stoichiometric oxidation with dimethyldioxirane: clues on
Energy 2010; 87(4): 1083–1095. the active species in the zeolite lattice. Journal of Molecular
[2] Wu X, Leung DYC. Optimization of biodiesel production Catalysis A: Chemical 1997; 117(1–3): 357–366.
from camelina oil using orthogonal experiment. Applied [24] Wróblewska A, Makuch E. The utilization of Ti-SBA-15
Energy 2011; 88(11): 3615–3624. catalyst in the epoxidation of allylic alcohols. Reaction
[3] Bournay L, Casanave D, Delfort B, Hillion G, Chodorge JA. Kinetics, Mechanisms and Catalysis 2012; 105(2): 451–468.
New heterogeneous process for biodiesel production: a way [25] Tatsumi T, Nakamura M, Negishi S, Tominaga H-o. Shape-
to improve the quality and the value of the crude glycerin selective oxidation of alkanes with H2O2 catalysed by
produced by biodiesel plants. Catalysis Today 2005; 106 titanosilicate. Journal of the Chemical Society, Chemical
(1–4): 190–192. Communications 1990; (6): 476–477.
[4] Hájek M, Skopal F. Treatment of glycerol phase formed by [26] Chen X, Fan Z, Quan X, Wei K. Epoxidation of allyl alcohol
biodiesel production. Bioresour. Technol. 2010; 101(9): to glycidol on Ti-MWW molecular sieves. Chinese Journal
3242–3245. of Catalysis 2006; 27(3): 285–290.
[5] Fan X, Burton R, Zhou Y. Glycerol (byproduct of biodiesel [27] Hutchings GJ, Lee DF. Control of product selectivity for the
production) as a source for fuels and chemicals – mini review. epoxidation of allyl alcohol by variation of the acidity of the
The Open Fuels & Energy Science Journal 2010; 3: 17–22. catalyst TS-1. Journal of the Chemical Society, Chemical
[6] Liu Y, Tuysuz H, Jia CJ, Schwickardi M, Rinaldi R, Lu AH, Communications 1994; (9): 1095–1096.
Schmidt W, Schuth F. From glycerol to allyl alcohol: iron [28] Wróblewska A, Ławro E, Milchert E. Technological
oxide catalyzed dehydration and consecutive hydrogen parameter optimization for epoxidation of methallyl alcohol
transfer. Chem. Commun. 2010; 46(8): 1238–1240. by hydrogen peroxide over TS-1 catalyst. Industrial &
[7] Yoshikawa T, Tago T, Nakamura A, Konaka A, Mukaida M, Engineering Chemistry Research 2006; 45(22): 7365–7373.
Masuda T. Investigation of reaction routes for direct [29] Wróblewska A. Optimization of the reaction parameters of
conversion of glycerol over zirconia-iron oxide catalyst. epoxidation of allyl alcohol with hydrogen peroxide over
Research on Chemical Intermediates 2011; 37(9): 1247–1256. TS-2 catalyst. Appl. Catal. Gen. 2006; 309(2): 192–200.
[8] Mota JAC, Pinto PB. Zeolite-catalyzed glycerol transformation. [30] Clerici MG, Ingallina P. Epoxidation of lower olefins with
Current Physical Chemistry 2012; 2(2): 211–216. hydrogen peroxide and titanium silicalite. J. Catal. 1993;
[9] Scokart PO, Amin A, Defosse C, Rouxhet PG. Direct probing 140(1): 71–83.
of the surface properties of alkali-treated aluminas by infrared [31] Hutchings GJ, Firth PG, Lee DF, McMorn P, Bethell D,
and x-ray photoelectron spectroscopy. J. Phys. Chem. 1981; Bulman Page PC, King F, Hancock F. In Studies in Surface
85(10): 1406–1412. Science and Catalysis, Grasselli RK, Oyama ST, Gaffney
© 2015 Curtin University of Technology and John Wiley & Sons, Ltd. Asia-Pac. J. Chem. Eng. 2015; 10: 598–606
DOI: 10.1002/apj
606 L. HARVEY ET AL. Asia-Pacific Journal of Chemical Engineering
AM, Lyons JE (eds.). Elsevier: Amsterdam, The Netherlands, [37] Perrichon V, Durupty M. Thermal stability of alkali metals
1997; 110: 535–544. deposited on oxide supports and their influence on the
[32] Bonati MLM, Joyner RW, Stockenhuber M. A temperature surface area of the support. Applied Catalysis 1988; 42(2):
programmed desorption study of the interaction of acetic anhydride 217–227.
with zeolite beta (BEA). Catalysis Today 2003; 81(4): 653–658. [38] Sanchez G, Friggieri J, Adesina AA, Dlugogorski BZ,
[33] Redhead P. Thermal desorption of gases. Vacuum 1962; Kennedy EM, Stockenhuber M. Catalytic conversion of
12(4): 203–211. glycerol to allyl alcohol; effect of a sacrificial reductant on
[34] Kim YT, Jung KD, Park ED. Gas-phase dehydration of the product yield. Catalysis Science & Technology 2014;
glycerol over silica-alumina catalysts. Appl. Catal. Environ. 4(9): 3090–3098.
2011; 107(1–2): 177–187. [39] Chaminand J, aurent Djakovitch L, Gallezot P, Marion P,
[35] Calvino-Casilda V, Martín-Aranda RM, López-Peinado AJ, Pinel C, Rosier C. Glycerol hydrogenolysis on heterogeneous
Sobczak I, Ziolek M. Catalytic properties of alkali metal- catalysts. Green Chemistry 2004; 6(8): 359–361.
modified oxide supports for the Knoevenagel condensation: [40] Nakagawa Y, Tomishige K. Heterogeneous catalysis of the
kinetic aspects. Catalysis Today 2009; 142(3–4): 278–282. glycerol hydrogenolysis. Catalysis Science & Technology
[36] Moens B, De Winne H, Corthals S, Poelman H, De Gryse R, 2011; 1(2): 179–190.
Meynen V, Cool P, Sels BF, Jacobs PA. Epoxidation of [41] Gandarias I, Requies J, Arias PL, Armbruster U, Martin A.
propylene with nitrous oxide on Rb2SO4-modified iron oxide Liquid-phase glycerol hydrogenolysis by formic acid over
on silica catalysts. J. Catal. 2007; 247(1): 86–100. Ni-Cu/Al2O3 catalysts. J. Catal. 2012; 290: 79–89.

© 2015 Curtin University of Technology and John Wiley & Sons, Ltd. Asia-Pac. J. Chem. Eng. 2015; 10: 598–606
DOI: 10.1002/apj

Vous aimerez peut-être aussi