Lewis acid zeolites for tandem Diels-Alder cycloaddition and dehydration of biomass-derived dimethylfuran and ethylene to renewable p-xylene

Article abstract of DOI:10.1039/c5gc02164b

Lewis acid zeolites including Zr-, Sn-, and Ti-BEA were examined for tandem [4 + 2] Diels-Alder cycloaddition of 2,5-dimethylfuran (DMF) and ethylene to oxanorbornene with subsequent dehydration to produce biorenewable p-xylene. Zr-BEA (Si/Zr = 168) exhibited superior performance with improved recalcitrance to deactivation, which was attributed to its low activity for the hydrolysis of DMF to 2,5-hexanedione and subsequent condensation. Zr-BEA also achieved the highest selectivity to p-xylene of 90% at 99% conversion of DMF. For low catalyst loading within a three-phase reactor, the reaction rate to form p-xylene was linearly proportional to the number of Lewis acid sites, while high catalyst loading exhibited zero order dependence on Lewis acid sites. A maximum achievable reaction rate was shown to be consistent with a transition in rate-limiting reactions from dehydration of oxanorbornene, the Diels-Alder product, to the Diels-Alder cycloaddition of DMF and ethylene.

Full text of DOI:10.1039/c5gc02164b

Green Chemistry
View Article Online
View Journal
PAPER
Lewis acid zeolites for tandem Diels–Alder
cycloaddition and dehydration of
biomass-derived dimethylfuran and
ethylene to renewable p-xylene†
Cite this: DOI: 10.1039/c5gc02164b
Chun-Chih Chang,a,b Hong Je Cho,a,b Jingye Yu,b,c Ray J. Gorte,b,c
Jason Gulbinski,b,d Paul Dauenhauerb,e and Wei Fan*a,b
Lewis acid zeolites including Zr-, Sn-, and Ti-BEA were examined for tandem [4 + 2] Diels–Alder cyclo-
addition of 2,5-dimethylfuran (DMF) and ethylene to oxanorbornene with subsequent dehydration to
produce biorenewable p-xylene. Zr-BEA (Si/Zr = 168) exhibited superior performance with improved
recalcitrance to deactivation, which was attributed to its low activity for the hydrolysis of DMF to 2,5-
hexanedione and subsequent condensation. Zr-BEA also achieved the highest selectivity to p-xylene of 90%
at 99% conversion of DMF. For low catalyst loading within a three-phase reactor, the reaction rate to form
p-xylene was linearly proportional to the number of Lewis acid sites, while high catalyst loading exhibited
zero order dependence on Lewis acid sites. A maximum achievable reaction rate was shown to be con-
sistent with a transition in rate-limiting reactions from dehydration of oxanorbornene, the Diels–Alder
product, to the Diels–Alder cycloaddition of DMF and ethylene.
Received 12th September 2015,
Accepted 7th October 2015
DOI: 10.1039/c5gc02164b
www.rsc.org/greenchem
alternative routes to produce p-xylene from renewable biomass
resources.5–14 For example, Lyons et al. demonstrated that
Introduction
p-Xylene is an important commodity chemical for the syn- p-xylene can be synthesized from ethylene (from dehydration of
thesis of terephthalic acid, which is a monomer for the manu- bio-ethanol) by a multi-step reaction pathway including trimer-
facture of polyethylene terephthalate (PET). PET has been ization of ethylene, the Diels–Alder reaction of 2,4-hexadiene
widely utilized to commercially manufacture synthetic fibers with ethylene and dehydrogenation of 3,6-dimethyl-
and plastic bottles due to its outstanding mechanical and cyclohexene; Shiramizu and Toste also reported that p-xylene
chemical properties.1–3 As the third largest chemical by can be selectively produced from 2,5-dimethylfuran (DMF,
volume in the world produced from petroleum (catalytic from hydrogenolysis of cellulosic biomass-derived 5-hydroxy-
reforming of naphtha) with increasing demand (6–8% per methylfurfural (HMF)) and acrolein (from glycerol, a biodiesel
year),4 there are now ongoing efforts worldwide for developing byproduct).10,12
In previous studies, we have reported a renewable reaction
pathway to produce p-xylene from lignocellulosic biomass over
solid Brønsted acid catalysts. The Diels–Alder dehydration
reaction includes the cycloaddition of biomass-derived di-
aDepartment of Chemical Engineering, University of Massachusetts Amherst, 686
N. Pleasant Street, Amherst, MA 01002, USA. E-mail: wfan@ecs.umass.edu;
http://www.efrc.udel.edu/; Fax: (+1) 413-545-1647; Tel: (+1) 413-545-1750
methylfuran (DMF) with ethylene, and subsequent de-
hydration of the formed Diels–Alder adduct to p-xylene
(Scheme 1).7,8,15–17 Competing reactions include hydrolysis of
DMF to 2,5-hexanedione and reactions between the cyclo-
adduct, DMF (oligomers) and ethylene (alkylated products).16
The proposed process to produce p-xylene can be completely
based on biomass resources. DMF can be produced from the
hydrolysis of cellulose followed by dehydration of glucose or
fructose to HMF, and hydrodeoxygenation of HMF to DMF.18
Ethylene can be produced from dehydration of bio-
ethanol.19,20 We have demonstrated that 90% selectivity to
p-xylene can be achieved by conducting the tandem Diels–Alder/
bCatalysis Center for Energy Innovation, a U.S. Department of Energy – Energy
Frontier Research Center, University of Delaware, 150 Academy Street, Newark,
Delaware 19716, USA
cDepartment of Chemical and Biomolecular Engineering, University of Pennsylvania,
220 S. 33rd St., Philadelphia, PA 19104, USA
dDepartment of Chemical Engineering, University of Delaware, 150 Academy Street,
Newark, Delaware 19716, USA
eDepartment of Chemical Engineering and Materials Science, University of
Minnesota, 421 Washington Ave. SE, Minneapolis, MN 55455, USA
†Electronic supplementary information (ESI) available: XRD patterns and
diethyl ether TPD data for Lewis acid zeolites, reaction kinetics from metal
oxides, GC/MS data for the cycloadduct of DMF and ethylene, data for de-
hydration of cyclohexanol, and DMF and byproduct distribution of 2,5-hexane-
dione added reactions for Zr-BEA and Al-BEA. See DOI: 10.1039/c5gc02164b
This journal is © The Royal Society of Chemistry 2015
Green Chem.

View Article Online
Paper
Green Chemistry
p-xylene. The Zr-BEA catalyst, in particular, exhibited a high
selectivity of 90% to p-xylene and slower deactivation due to
the low hydrolysis rate for the by-product, 2,5-hexanedione. It
was also evidenced that Zr-BEA can moderately enhance the
formation of the cycloadduct from DMF and ethylene, result-
ing in the highest measured p-xylene production rate.
Experimental
Preparation of catalysts
Solid Lewis acid zeolite catalysts employed in this study were
prepared according to the literature and our previous report
Scheme 1 Simplified reaction scheme for synthesis of p-xylene from
DMF and ethylene. 1: DMF; 2: cycloadduct; 3: p-xylene; 4: 2,5-hexane-
dione; 5: representative alkylated product; 6: representative oligomer.
based on a seeded growth method developed for Sn-BEA.32–34
.
The detailed procedure for the preparation of seed suspension
is available in our previous study.33
Preparation of Zr-BEA. Typically, 23.72 g of tetraethyl-
dehydration reactions over Brønsted acid zeolite catalysts orthosilicate (TEOS, 99%, Alfa Aesar) was added to 27.39 g of
(i.e. Al-BEA zeolite, Si/Al = 12.5) in n-heptane solvent.7 Toluene and tetraethylammonium hydroxide solution (TEAOH, 35 wt%,
benzene can also be produced with similar chemistry by start- SACHEM) and magnetically stirred in a plastic beaker. After a
ing with 2-methylfuran and furan, respectively.21 The compu- homogeneous solution was obtained (about one hour), 0.3 g of
tational studies including density functional theory (DFT) and zirconium(IV) oxychloride octahydrate (ZrOCl2·8H2O, Sigma-
quantum mechanics/molecular mechanics (QM/MM) revealed Aldrich) was dissolved in 1.92 g water and added dropwise into
that the formation of the oxanorbornene cycloadduct inter- the silicate solution. The mixture was kept stirring in a fumed
mediate by the addition of ethylene to DMF is likely not cata- hood until the desired amount of water and ethanol (gener-
lyzed by Brønsted acid sites within zeolite catalysts.8,22,23 ated from hydrolysis of TEOS) was evaporated. To this solution,
However, Brønsted acids can effectively catalyze the de- 1.712 mL of dealuminated zeolite BEA seed suspension
hydration of the cycloadduct intermediate and decrease the (0.163 g of seed crystal per mL; corresponding to 4.0 wt% seed
activation energy from 58–60 kcal mol−1 to 14–19 kcal mol−1
,
with respect to silica) was added and mixed with a plastic
spatula. Finally, 2.36 mL of hydrofluoric acid (49%, Alfa Aesar)
leading to a significantly increased overall reaction rate.8,15,22
Although Brønsted acid zeolite catalysts exhibit promising was added and mixed with a plastic spatula, and a hard gel
performance, the overall reaction rate is limited due to the was formed. The final composition of the gel was SiO2 : 0.008
uncatalyzed Diels–Alder reaction. There are thus persistent ZrO2 : 0.56TEAOH : 0.56HF : 7.5H2O. The hard gel was transferred
needs for improving the overall reaction rate by developing to a Teflon lined stainless steel autoclave and heated in a pre-
catalysts which can promote the Diels–Alder reaction and more heated convection oven at 140 °C for four days with 2.0 rpm
selectively dehydrate to reduce side product formation. Pre- rotation. The solids were washed with 1.0 L of deionized water by
vious studies have suggested that Lewis acids can coordinate filtration and dried in an 80 °C oven overnight. The organic
with dienes to catalyze the Diels–Alder reaction.10,24–27 DFT template was removed by calcining the as-made catalyst at 550 °C
studies of the production of p-xylene from DMF have also with a ramping rate of 1.0 °C min−1 for 12 h under flowing dry
suggested that Lewis acids can coordinate with ethylene to air (Grade 0.1, Middlesex Gases & Technologies, Inc.).
catalyze the Diels–Alder reaction to form the cycloadduct inter-
Preparation of Sn-BEA. Typically, 20.94 g of TEOS was
mediate, oxanorbornene.15,28 In addition, experimental and added to 23.02 g of TEAOH solution and magnetically stirred
theoretical studies have shown that solid Lewis acid catalysts in a plastic beaker. After a homogeneous solution was
are capable of catalyzing the dehydration of the cycloadduct obtained (about one hour), 0.3 g of tin(IV) chloride hydrate
intermediate and other alcohols.15,28–30 Using
a
Fe3+- (SnCl4·xH2O, 34.4 wt% of tin) was dissolved in 1.96 g water
exchanged montmorillonite K10 clay as a Lewis acid catalyst and added dropwise into the silicate solution. The mixture was
for the aromatization of oxabicyclic derivatives has also been kept stirring in a fumed hood until the desired amount of
reported.27,31 Lewis acid zeolite catalysts, in particular Sn-BEA water and ethanol was evaporated. To this solution, 1.541 mL
and Zr-BEA, have also shown catalytic activity for the synthesis of dealuminated zeolite BEA seed suspension (corresponding
of terephthalic acid via Diels–Alder reactions with oxidized to 4.0 wt% seed with respect to silica) was added and mixed
variants of 5-hydroxymethylfurfural.3
with a plastic spatula. Finally, 1.941 mL of hydrofluoric acid
Motivated by the existing experimental and theoretical (49%, Alfa Aesar) was added and mixed with a plastic spatula,
studies, Lewis acid catalysts were employed in this work for and a hard gel was formed. The final composition of the gel
producing p-xylene from DMF and ethylene through the was SiO2 : 0.008SnO2 : 0.54TEAOH : 0.54HF : 7.5H2O. The hard
tandem Diels–Alder/dehydration reaction pathway. We demon- gel was transferred to a Teflon lined stainless steel autoclave
strate that Lewis acid zeolite catalysts are active for producing and heated in a preheated convection oven at 140 °C for four
Green Chem.
This journal is © The Royal Society of Chemistry 2015

View Article Online
Green Chemistry
Paper
days with 2.0 rpm rotation. The solids were washed with 1.0 L Reaction experiments
of deionized water by filtration and dried in an 80 °C oven
The reactor system schematic is available in our previous
work.8 In typical reactions with DMF (Alfa Aesar), 100 mL of
1.35 M of DMF in n-heptane (Alfa Aesar) with 0.08 M n-tri-
decane (Alfa Aesar) was enclosed in a 160 mL bench-top reactor
(Parr), and the reaction vessel was purged with nitrogen. The
catalyst acid concentration present in the system was defined
as the total number of acid sites of the catalyst divided by the
reactant volume. The mixture was then stirred at 1000 rpm
with a gas entrainment impeller to ensure facile mass transfer.
The system was heated to 250 °C and controlled with a 4848
control unit (Parr). The reactor was then pressurized with
38 bar (partial pressure) of ethylene gas (Airgas), and the total
pressure of the system was maintained at 62 bar over the reac-
tion period. Liquid was withdrawn by using a double block
sampling system at designated reaction times. The compo-
sition of the sample was then analyzed with an Agilent 6890A
gas chromatograph with a Rtx-VMS (Restek) column equipped
with a flame ionization detector. The products (DMF, p-xylene,
2,5-hexanedione) were identified by comparing the retention
times with standard chemicals. Based on the reaction
scheme,16 the concentration of the alkylated products was esti-
mated by using the response factor (RF) for 1-methyl-4-propyl-
benzene, and the concentration of the oligomers was
estimated using the additive RF of p-xylene and DMF. The oxa-
norbornene cycloadduct of DMF and ethylene was identified
with a GC-MS system (5890 GC/5972 series mass selective
detector, HP) and by comparing the electron ionization
spectrum with the literature.16 The concentration of the cyclo-
adduct was estimated using the RF for p-xylene. The conver-
sion of DMF and selectivity to products were defined as
follows:
overnight. The organic template was removed by calcining the
as-made catalyst at 550 °C with a ramping rate of 1.0 °C min−1
for 12 h under flowing dry air.
Preparation of Ti-BEA. Typically, 24.66 g of TEOS was added
to a mixture of 26.76 g of TEAOH solution and 4.24 g of hydro-
gen peroxide solution (H2O2, 30%, Fisher) and magnetically
stirred in a plastic beaker. After a homogeneous solution was
obtained (about one hour), 0.45 g of titanium(IV) ethoxide
(Aldrich) was added dropwise into the silicate solution. The
mixture was kept stirring in a fumed hood until the desired
amount of water and ethanol was evaporated. To this yellow
clear solution, 1.708 mL of dealuminated zeolite BEA seed sus-
pension (corresponding to 4.0 wt% seed with respect to silica)
was added and mixed with a plastic spatula. Finally, 2.306 mL
of hydrofluoric acid (49%, Alfa Aesar) was added and mixed
with a plastic spatula, and a hard gel was formed. The final
composition of the gel was SiO2 : 0.008TiO2 : 0.54TEAOH : 0.54
HF : 7.5H2O. The hard gel was transferred to a Teflon lined
stainless steel autoclave and heated in a preheated convection
oven at 140 °C for eight days with 2.0 rpm rotation. The solids
were washed with 1.0 L of deionized water by filtration
and dried in an 80 °C oven overnight. The organic template
was removed by calcining the as-made catalyst at 550 °C
with a ramping rate of 1.0 °C min−1 for 12 h under flowing
dry air.
Preparation of Al-BEA. Aluminosilicate zeolite BEA (CP814E,
Si/Al = 12.5) was obtained from Zeolyst. The catalyst was cal-
cined at 550 °C with a ramping rate of 1.0 °C min−1 for 12 h
under flowing dry air prior to the reaction.
Characterization of the catalysts
C
ꢁ CDMF;t
DMF;t0
The crystalline phase of the Lewis acid catalysts was examined
by powder X-ray diffraction (PXRD) on a diffractometer (X’Pert
Pro, PANalytical) operated at a voltage of 40 kV and a current
of 45 mA with an X’cellerator detector scanning from 4° to 40°.
The quantity of Lewis acid sites within the zeolites was deter-
mined by diethyl ether temperature programmed desorption
coupled with thermogravimetric analysis (TPD–TGA), with the
exception of Ti-BEA. Ti-BEA interacted weakly with the probe
molecules, and did not provide characterization of Lewis acid
sites. Temperature Programmed Desorption (TPD) and
Thermogravimetric Analysis (TGA) experiments were carried
out using a CAHN 2000 microbalance mounted within a
vacuum chamber that could be evacuated with a diffusion
pump.35 Prior to exposing them to a few hundred Pa of the
diethyl ether, the samples were heated in a vacuum to 500 °C
and then cooled to room temperature in vacuo. During TPD–
TGA measurements, the sample weights were recorded con-
tinuously using the microbalance and the desorbing species
XDMFð%Þ ¼ 100 ꢀ
ð1Þ
ð2Þ
ð3Þ
C
DMF;t0
Ci ꢁ Ci;t
0
P
Selectivityið%Þ ¼ 100 ꢀ
Yieldið%Þ ¼ 100 ꢀ
ðCi ꢁ Ci;t
Þ
0
i
Ci
C
DMF;t0
where t is the duration of time after ethylene was added to the
reactor, and Ci is the concentration of the reactant and pro-
ducts, i. For a reaction carried out with 2,5-hexanedione, 0.2 M
2,5-hexanedione (Alfa Aesar) was added with 1.35 M DMF solu-
tion in n-heptane. The reactor was purged with N2, heated to
250 °C and held at 250 °C for one hour prior to addition of
ethylene (62 bar).
Results and discussion
were monitored using an SRI quadrupole mass spectrometer To demonstrate the catalytic activity of solid Lewis acid cata-
(RGA100) at a ramping rate of 10 °C min−1. Elemental analyses lysts for the Diels–Alder dehydration reaction, siliceous zeolite
were carried out in Analytical Geochemistry Lab, Department BEA with substituted metal atoms (Zr-, Sn-, and Ti-BEA) at
of Earth Sciences in the University of Minnesota using framework positions were prepared according to the methods
ICP-OES (iCAP 6500 Duo View, Thermo Scientific).
reported in the literature and applied in the reaction of DMF
This journal is © The Royal Society of Chemistry 2015
Green Chem.

View Article Online
Paper
Green Chemistry
and ethylene. Other conventional metal oxide catalysts includ-
ing ZrO2, SnO2, and TiO2 were also tested in the reaction for
comparison. The corresponding XRD patterns for the Lewis
acid zeolites used in this study are shown in Fig. S1.† Typical
diffraction peaks for zeolite BEA were observed for all samples,
and no impurity phase (i.e., metal oxides) was detected, indi-
cating that high-quality crystals were obtained. Acid site con-
centrations of the Lewis acid zeolites were determined by the
Si/M ratio obtained from elemental analysis measurement
(ICP-OES), and further confirmed by diethyl ether TPD–TGA
measurement (Table 1). It has been shown that diethyl ether
forms a 1 : 1 complex pair with Lewis acid sites (Sn center) in
the Sn-BEA catalyst, and it desorbs from the Sn site at the
temperatures from 350 K to 450 K. Measuring the weight
change between the temperatures can provide the quantity of
Lewis acid sites in the sample.29 The diethyl ether TPD–TGA
results obtained from Sn-BEA, Zr-BEA and Ti-BEA catalysts are
shown in Fig. S2.† As listed in Table 1, the TPD–TGA measure-
ments for Sn- and Zr-BEA indicate good agreement with the
elemental analysis, while the measurement for Ti-BEA shows a
lower acid site density than the Ti concentration in the sample
determined by elemental analysis (40 μmol g−1 vs. 129 μmol g−1).
This may be due to the weak interaction between diethyl
ether and the Ti site in the Ti-BEA sample, resulting in deso-
rption of the probe molecule from the sample in vacuo prior to
heating. For Al-BEA, isopropylamine TPD–TGA was utilized to
obtain the Brønsted acid site concentration.21 The quantity of
Brønsted acid sites in Al-BEA (620 μmol g−1) was lower than
the Al concentration (1230 μmol g−1), indicating the presence
of an extra framework Al species in the sample. Contribution
from the extra framework Al species was neglected due to low
activity from Al2O3 according to our previous report.7 For reac-
tion rate calculations, the acid site concentrations determined
by TPD–TGA analysis were used. Due to weak adsorption of
diethyl ether on Ti-BEA, the Ti concentration determined by
elemental analysis was used in the reaction rate calculations.
Fig. 1 depicts the temporal profiles of yield and selectivity
in the reaction to p-xylene with respect to DMF conversion over
different catalysts at the same acid site concentration (1.0 mM)
in the reaction system (1.35 M DMF in n-heptane and 62 bar of
ethylene) at 250 °C. Surprisingly, all considered Lewis acid zeo-
lites are capable of catalyzing the reaction to produce p-xylene.
This activity was attributed to Lewis acidity from the isolated
Fig. 1 (a) Kinetics of DMF conversion at 250 °C. (b) Selectivity and
(c) yield to p-xylene as a function of DMF conversion for the reaction of
DMF and ethylene at 1.0 mM acid loading of Sn-BEA, Zr-BEA, Ti-BEA,
and Al-BEA.
Table 1 Acid site density of the zeolites
Acid site density (μmol g−1
)
Si/T
ICP
Catalyst
ICP
Chemical TPDb
metal atoms located in the zeolite frameworks, as metal oxides
(ZrO2, SnO2, and TiO2) and pure siliceous BEA did not show
significant catalytic activity (Fig. S3†).8 Among the tested cata-
lysts, Zr-BEA exhibited superior performance with 80% selecti-
vity to p-xylene at 80% conversion of DMF (Fig. 1b) in 24 h.
Sn-, Ti-, and Al-BEA, however, deactivated noticeably faster,
Zr-BEA
Sn-BEA
Ti-BEA
Al-BEA
168
126
128
98
130
129
95
120
40
12.5a
1230
620c
a Zeolyst. b Diethyl ether TPD except as otherwise noted.
c Isopropylamine TPD for Brønsted acid sites.
Green Chem.
This journal is © The Royal Society of Chemistry 2015

View Article Online
Green Chemistry
Paper
and only achieved about 60% DMF conversion after 24 hours
of reaction with lower selectivity to p-xylene. Furthermore, Zr-
BEA yielded ∼33% more p-xylene than Al-BEA at 60% conver-
sion (Fig. 1c). It is worth noting that the activity of Sn-MFI
was considerably lower (8 mM h−1; Table 2) under identical
conditions, as it converted less than 20% of DMF after 24 h
of reaction, which highlights the uniqueness of the zeolite
BEA structure for this reaction. The result is consistent with
our previous report which suggested that the small micro-
pore structure of ZSM-5, an aluminum analog of Sn-MFI, is
not favorable for the reaction.7 This implies that either the
Diels–Alder reaction could not be catalyzed within the frame-
work or the formed cycloadduct intermediate could not
easily diffuse within the micropore structure of the MFI
zeolite.
To further understand the high selectivity to p-xylene over
Zr-BEA, the byproduct distributions for the reaction of DMF
and ethylene over Lewis acid zeolite catalysts and Al-BEA were
measured and are shown in Fig. 2. Sn-BEA preferentially cata-
lyzes the side reactions and produces a more hydrolysis
product, 2,5-hexanedione, and oligomers at lower conversion,
which explains the low selectivity to p-xylene for Sn-BEA. Zr-
BEA, however, is able to minimize the hydrolysis of DMF, with
low 2,5-hexanedione yield at all extents of conversion (Fig. 2a).
Zr-BEA also minimizes the formation of oligomers (Fig. 2b)
but promotes the reaction between cycloadduct and ethylene,
giving the highest yield of alkylated products (Fig. 2c). Al-BEA,
on the other hand, produces only small amounts of oligomeric
products and more 2,5-hexanedione and alkylated products. At
low conversion of DMF, strong Brønsted acidity of Al-BEA can
effectively catalyze the hydrolysis reaction of DMF, which
explains the even lower p-xylene selectivity at high catalyst
loading shown later (Fig. 3). All results suggest that using Zr-
BEA is moderately beneficial for the production of p-xylene
compared with Al-BEA due to its low activity for catalyzing side
reaction of the hydrolysis of DMF. The reaction results reveal
that the characteristics of the acid type can manipulate the
product distributions by affecting the reaction rates between
the competing reaction pathways, while the selectivity towards
p-xylene for all catalysts is not significantly different at high
conversion of DMF.
Table 2 Summary of the reaction results for the catalystsa
Fig. 2 Yield of (a) 2,5-hexanedione, (b) oligomers and (c) alkylated side
products as a function of DMF conversion for the reaction of DMF in
heptane with ethylene (62 bar) at 250 °C with Sn-BEA, Zr-BEA, and
Ti-BEA as well as Al-BEA catalysts at 1.0 mM acid concentration.
Rate of p-xylene
DMF conversion
after 24 h
p-Xylene yield
after 24 h
Catalyst
production (mM h−1
)
Al-BEA
Sn-BEA
Zr-BEA
Ti-BEA
Sn-MFI
SnO2
122 11.0
113 6.8
121 4.9
65 11.2
8
0.6
0.7
55 3.7
63 0.1
83 1.6
60 3.5
27
2
2
30 0.5
43 1.8
61 1.5
41 4.5
17
0.3
0.5
To understand the effect of the active site concentration, Zr-
BEA and Al-BEA catalysts were compared at high catalyst
loading (acid site concentration of 3.5 and 3.1 mM, respect-
ively). As shown in Fig. 3, it was found that the kinetics of
DMF conversion and overall p-xylene yield were similar
between Zr-BEA and Al-BEA, but the selectivity to p-xylene for
ZrO2
TiO2
1.1
3
1.4
a Reaction conditions: 1.0 mM acid (0.8 g for metal oxides), 1.35 M
DMF in n-heptane, 62 bar ethylene, at 250 °C.
This journal is © The Royal Society of Chemistry 2015
Green Chem.

View Article Online
Paper
Green Chemistry
highest selectivity to p-xylene (∼90%) at full conversion of
DMF at the acid site concentration of 3.5 mM, which was
similar to the previously reported selectivity to p-xylene
achieved with Al-BEA.7
Fig. 4 shows the rate of production of p-xylene versus the
acid site concentration for different catalysts. The reaction
rates were deliberately presented on a volumetric basis of mM h−1
(rather than TOF), due to the previous work which indi-
cated that the overall reaction was limited by either Diels–
Alder cycloaddition or dehydration depending on the active
site concentration.7,8,22 At a low acid site concentration, it was
found that the rate of p-xylene production is proportional to
the acid concentration and reaches a maximum when the acid
site concentration is higher than ∼2.5 mM for both Brønsted
acid, Al-BEA, and Lewis acid zeolites, Sn-BEA and Zr-BEA. This
is consistent with our recent report on the kinetic study for the
same system using a Brønsted acid zeolite, zeolite Y.22 The
kinetic behavior and rate dependence on the acid site concen-
tration can be interpreted as a transition in the rate controlling
reaction. The overall reaction from DMF and ethylene to
p-xylene is a sequence of Diels–Alder cycloaddition of ethylene
and DMF followed by dehydration of the cycloadduct. Similar
to the Brønsted acid zeolite catalysts, at low Lewis acid site
concentrations, the overall rate might be controlled by the
dehydration reaction due to insufficient quantity of catalytic
sites; at high Lewis acid site loadings, the reaction could be
limited by the cycloaddition reaction, which results in the
plateau (i.e. a maximum reaction rate) of p-xylene production.
The activity of the catalysts was compared using the turn-
over frequencies (TOF) calculated within the dehydration-
limited region (linear dependent region, <2.5 mM acid sites in
Fig. 4) using the production rate of p-xylene normalized to the
amount of acid sites of the catalysts (only Brønsted acid sites
were considered for Al-BEA). Interestingly, Zr-BEA exhibited
Fig. 3 (a) Kinetics of conversion of DMF, (b) selectivity, and (c) yield to
p-xylene as a function of DMF conversion for the reaction of 1.35 M
DMF in heptane and 62 bar of ethylene at 250 °C with high acid loading
of Zr-BEA (3.5 mM) and Al-BEA (3.1 mM).
Zr-BEA was significantly higher at low conversion (80% vs.
60%). Differences in selectivity to p-xylene were larger when
compared at low acid site loading (acid site concentration of
1.0 mM), which appears due to the low activity of Zr-BEA for
Fig. 4 p-Xylene production rate versus the active site concentration in
n-heptane for Zr-BEA, Sn-BEA, Ti-BEA and Al-BEA for the reaction of
DMF with ethylene (62 bar) at 250 °C. Initial catalytic rates were calcu-
lated at <20% conversion of DMF. The dashed line is provided as a guide
catalyzing the hydrolysis reaction of DMF. Zr-BEA exhibited the to the eyes.
Green Chem.
This journal is © The Royal Society of Chemistry 2015

View Article Online
Green Chemistry
Paper
higher TOF (121 4.9 h−1) than Sn-BEA (113 6.8 h−1) and Ti- ticular, Zr-BEA produced the most cycloadduct intermediate
BEA (65 11 h−1), and was similar to Al-BEA (122 11 h−1) as throughout the reaction while exhibiting the same initial pro-
shown in Table 2. The results suggest that Zr-BEA is an duction rate of p-xylene as Al-BEA (Fig. 3). The results suggest
efficient catalyst for the sequential reaction to produce that the Zr center is moderately effective for the cycloaddition
p-xylene. The result was unexpected since it has been known reaction and less efficient for the dehydration reaction com-
that Lewis acid zeolites are not efficient dehydration catalysts pared to Al-BEA.
for several alcohols compared with Brønsted acid zeolites.35
In addition to the moderately improved production rate of
Our control experiment for dehydration of cyclohexanol to p-xylene, Zr-BEA also exhibited slower deactivation compared
cyclohexene using Al-BEA and Zr-BEA also indicates that Al- to other Lewis acid zeolites and Al-BEA as shown in Fig. 1a. In
BEA is generally superior for dehydration reactions (Fig. S4†). the plot of selectivity to p-xylene (Fig. 1b), it was observed that
A further study on the catalytic activity of Zr-BEA for the dehy- both Lewis and Brønsted acid zeolite catalysts exhibit counter-
dration of the cycloadduct to p-xylene is necessary in order to intuitive behavior of increasing selectivity to p-xylene with
further improve the catalytic system.
DMF conversion. With the Zr-BEA catalyst, the selectivity to
Despite significant catalytic differences under dehydration- p-xylene at the DMF conversion of 20% is slightly lower than
controlling reaction conditions, only subtle differences were 80%, while it increases to 80% at higher conversion. This
observed between Lewis acid catalysts and Brønsted acid zeo- phenomenon has been observed with the Brønsted acid
lites within the “plateau region” (catalyst independent region, zeolite, Al-BEA, and was attributed to the equilibrium-driven
>2.5 mM acid sites in Fig. 4) where the Diels–Alder reaction formation of DMF from its hydrolysis byproduct, 2,5-hexane-
controls the overall rate. Comparing the formation rates of dione, followed by the formation of the thermodynamically
p-xylene, Zr-BEA exhibited about 10–20% higher production stable final product, p-xylene.7 It has been suggested by
rate than Sn-BEA and Al-BEA, which might indicate a slight Do et al. that 2,5-hexanedione formed from the hydrolysis of DMF
enhancement in catalysis of the Diels–Alder reaction. However, can cause catalyst deactivation by forming insoluble surface
the enhancement of the overall reaction with Lewis acids is condensates via acid catalyzed aldol-type reactions.16 In order
minor; variation of the Lewis acid site concentration for either to further understand the effects of 2,5-hexanedione on the de-
Sn-BEA or Zr-BEA (3.5–5.0 mM acid sites in Fig. 3) did not activation of zeolite catalysts in the reaction, 2,5-hexanedione
result in measurable differences in the overall rate.
(0.2 M) was intentionally added in the reaction mixture as a
To confirm the limited catalytic activity of Zr-BEA for the reactant with Zr-BEA and Al-BEA catalysts (at 1.0 mM acid site
Diels–Alder reaction, the concentration of the cycloadduct pro- concentration). The reaction mixture was well mixed and
duced from the Diels–Alder reaction during the reaction was heated at 250 °C under nitrogen for one hour prior to the
monitored at low catalyst loadings. The formation of the DMF/ addition of ethylene. The changes in the concentration of 2,5-
ethylene cycloadduct was confirmed by GC/MS (Fig. S5†). As hexanedione and p-xylene with the reaction time are shown in
shown in Fig. 5, the Lewis acid zeolites are moderately more Fig. 6. It was found that with increasing temperature, the
active to catalyze the Diels–Alder reaction compared to Al-BEA,
and they produce a slightly higher cycloadduct intermediate
concentration during the tandem reaction than Al-BEA. In par-
Fig. 6 The concentration evolution of 2,5 hexanedione (2,5-hex)
and p-xylene with time for the deactivation-simulated reaction with
Fig. 5 Yield of cycloadduct versus DMF conversion (1.35 M DMF in
heptane, 250 °C, 62 bar ethylene). The concentration of the cycloadduct
was estimated by using the response factor of p-xylene. The acid con-
centration is 1.0 mM for all.
2,5 hexanedione added with DMF as the reactant for Al-BEA and Zr-BEA.
The reaction mixture was heated at 250 °C under N2 for 1 h before
ethylene (62 bar of total pressure) was introduced into the reactor. The
acid concentration is 1 mM in the system.
This journal is © The Royal Society of Chemistry 2015
Green Chem.

View Article Online
Paper
Green Chemistry
concentration of 2,5-hexanedione quickly decreased over both BEA can prevent fast catalyst deactivation but does not signifi-
Zr-BEA and Al-BEA catalysts. After one hour at 250 °C, the con- cantly enhance the production rate of toluene (70 vs. 68 mM h−1
centration of 2,5-hexanedione was 0.05 M for the Al-BEA cata- toluene production rate for Zr-BEA and Al-BEA, respect-
lyst and 0.1 M for the Zr-BEA catalyst, indicating a faster ively). In the reaction with furan, Zr-BEA only produced ∼5%
reaction rate of 2,5-hexanedione over the Brønsted acid zeolite of benzene after the 24 h reaction (∼70% conversion), reveal-
catalyst. One hour after the introduction of ethylene into the ing that the Lewis acid zeolite (Zr-BEA) is not effective in
experiment, p-xylene was produced by both catalysts with Zr- improving the yield of benzene. Overall, the performance of
BEA exhibiting a significantly higher p-xylene formation rate Zr-BEA was slightly better in furan and MF reactions in terms
than Al-BEA (56 mM h−1 vs. 35 mM h−1). However, the pro- of catalyst life, but they exhibited no significant difference
duction rates of p-xylene were significantly lower than the rate with regard to selectivity to benzene and toluene when com-
achieved without adding 2,5-hexanedione at the beginning of pared with Al-BEA.
the reaction (120 mM h−1). In addition, it was observed that
during the reaction the concentration of DMF does not
increase with the decrease of 2,5-hexanedione, which indicates Conclusions
that 2,5-hexanedione did not undergo dehydration to form
DMF under the reaction conditions (Fig. S6a†).
The results clearly suggest that the addition of 2,5-hexane-
dione did indeed cause catalyst deactivation, in particular for
Solid Lewis acid zeolites including Zr-, Sn-, and Ti-BEA were
evaluated in p-xylene production from DMF and ethylene.
Lewis acid zeolites were shown to catalyze the Diels–Alder
addition/dehydration reactions. It was also found that Lewis
the Brønsted acid zeolite catalyst, Al-BEA. Zr-BEA, however,
acid zeolites exhibit longer lifetime than Al-BEA, in particular
deactivated slower under the same conditions, which could be
Zr-BEA shows the best performance with higher turnover fre-
quency and slower deactivation, which could be attributed to
due to weaker acid sites on Zr-BEA compared to Al-BEA that
are not efficient at catalyzing the condensation reactions of
its low selectivity to the hydrolysis product, 2,5-hexanedione,
2,5-hexanedione. The longer catalyst life (slower deactivation)
and weak acid site nature which prevents the polymerization
of Zr-BEA may therefore be attributed to: (1) a slower pro-
of 2,5-hexanedione. It was also demonstrated that the pro-
duction rate of 2,5-hexanedione (Fig. 2a), and (2) weaker acid
duction rate of p-xylene was dependent on the total amount of
sites that are not efficient for catalyzing condensation reac-
the Lewis acid catalyst at low loading and independent at high
tions of 2,5-hexanedione (Fig. 6).
loading due to the nature of the tandem reaction, which was
The cycloaddition/dehydrative aromatization chemistry can
similar to the behavior of Brønsted acid zeolite catalysts.
be extended to 2-methylfuran (MF) and furan to produce
Finally, it was revealed that Zr-BEA moderately promoted the
toluene and benzene.7,17,21 Due to the lack of protecting
Diels–Alder cycloaddition reaction by producing more oxanor-
bornene while being a less efficient dehydration catalyst; this
resulted in a mild enhancement of p-xylene production by
∼10–20% compared to Al-BEA. The current findings can serve
as a guide to design efficient and long-lasting catalysts for the
tandem Diels–Alder cycloaddition/dehydration reactions.
methyl groups on the furan α-carbon, the yields of toluene and
benzene were generally lower than the reaction for producing
p-xylene from DMF. As shown in Fig. 7, the conversion of MF
and yield of toluene achieved over the Zr-BEA catalyst were
slightly higher than Al-BEA after a 24 hour reaction. However,
the selectivity to toluene was similar, which indicates that Zr-
Acknowledgements
The authors acknowledge financial support from the Center
for Catalysis and Energy Innovation, an Energy Frontier
Research Center funded by the U.S. Department of Energy,
Office of Science, Office of Basic Energy Sciences under award
number DE-SC0001004.
References
1 R. A. F. Tomás, J. C. M. Bordado and J. F. P. Gomes, Chem.
Rev., 2013, 113, 7421–7469.
2 Y. Tachibana, S. Kimura and K.-i. Kasuya, Sci. Rep., 2015, 5,
8249.
3 J. J. Pacheco and M. E. Davis, Proc. Nat. Acad. Sci. U. S. A.,
2014, 111, 8363–8367.
Fig. 7 MF conversion, toluene yield, and toluene selectivity for the
reaction of MF and ethylene with Zr-BEA and Al-BEA after a 24 h reac-
4 P. Wantanachaisaeng and K. O’Neil, Capturing oppurtunities
for para-xylene production, UOP LLC, 2007.
tion. The reaction involves 1.35 M of MF in n-heptane with 62 bar of
ethylene at 250 °C. The acid concentration for both catalysts is 0.8 mM.
Green Chem.
This journal is © The Royal Society of Chemistry 2015

View Article Online
Green Chemistry
Paper
5 M. N. Masuno, P. B. Smith, D. A. Hucul, K. Brune, 20 A. P. Kagyrmanova, V. A. Chumachenko, V. N. Korotkikh,
R. L. Smith, J. Bissell, D. A. Hirsch-Weil and E. J. Stark,
Methods of producing p-xylene from 2,5-hexanedione and
V. N. Kashkin and A. S. Noskov, Chem. Eng. Jpn., 2011,
176–177, 188–194.
ethylene in the presence of catalysts and solvents, US Patent 21 S. K. Green, R. E. Patet, N. Nikbin, C. L. Williams,
8889938B2, 2014.
6 B. Saha and M. M. Abu-Omar, ChemSusChem, 2015, 8,
1133–1142.
C.-C. Chang, J. Yu, R. J. Gorte, S. Caratzoulas, W. Fan,
D. G. Vlachos and P. J. Dauenhauer, Appl. Catal., B., 2016,
180, 487–496.
7 C.-C. Chang, S. K. Green, C. L. Williams, P. J. Dauenhauer 22 R. E. Patet, N. Nikbin, C. L. Williams, S. K. Green,
and W. Fan, Green Chem., 2014, 16, 585–588.
C.-C. Chang, W. Fan, S. Caratzoulas, P. J. Dauenhauer and
8 C. L. Williams, C.-C. Chang, P. Do, N. Nikbin,
D. G. Vlachos, ACS Catal.., 2015, 5, 2367–2375.
S. Caratzoulas, D. G. Vlachos, R. F. Lobo, 23 Y.-P. Li, M. Head-Gordon and A. T. Bell, J. Phys. Chem. C,
W. Fan and P. J. Dauenhauer, ACS Catal.., 2012, 2, 935–
939.
9 Y.-T. Cheng, Z. Wang, C. J. Gilbert, W. Fan and
2014, 118, 22090–22095.
24 M. V. Gómez, Á. Cantín, A. Corma and A. de la Hoz, J. Mol.
Catal. A: Chem., 2005, 240, 16–21.
G. W. Huber, Angew. Chem., Int. Ed., 2012, 51, 11097– 25 Y. Hayashi, M. Nakamura, S. Nakao, T. Inoue and M. Shoji,
11100. Angew. Chem., Int. Ed., 2002, 41, 4079–4082.
10 M. Shiramizu and F. D. Toste, Chem. – Eur. J., 2011, 17, 26 R. M. Dessau, J. Chem. Soc., Chem. Commun., 1986, 1167–
12452–12457. 1168.
11 D. Wang, C. M. Osmundsen, E. Taarning and 27 P. Vogel, J. Cossy, J. n. Plumet and O. n. Arjona, Tetra-
J. A. Dumesic, ChemCatChem, 2013, 5, 1–8. hedron, 1999, 55, 13521–13642.
12 T. W. Lyons, D. Guironnet, M. Findlater and M. Brookhart, 28 N. Nikbin, S. Feng, S. Caratzoulas and D. G. Vlachos,
J. Am. Chem. Soc., 2012, 134, 15708–15711. J. Phys. Chem. C, 2014, 118, 24415–24424.
13 M. W. Peters, J. D. Taylor, M. Jenni, L. E. Manzer and 29 S. Roy, K. Bakhmutsky, E. Mahmoud, R. F. Lobo and
D. E. Henton, Integrated process to selectively convert R. J. Gorte, ACS Catal., 2013, 3, 573–580.
renewable isobutanol to p-xylene, US Patent 20110087000, 30 S. Roy, G. Mpourmpakis, D.-Y. Hong, D. G. Vlachos,
2011. A. Bhan and R. J. Gorte, ACS Catal., 2012, 2, 1846–1853.
14 T. A. Brandvold, Carbohydrate route to para-xylene and 31 D. D. Claeys, C. V. Stevens, B. I. Roman, P. Van De Caveye,
terephthalic acid, US Patent 20100331568, 2010.
15 N. Nikbin, P. T. Do, S. Caratzoulas, R. F. Lobo,
M. Waroquier and V. Van Speybroeck, Org. Biomol. Chem.,
2010, 8, 3644–3654.
P. J. Dauenhauer and D. G. Vlachos, J. Catal., 2013, 297, 32 Y. Zhu, G. Chuah and S. Jaenicke, Chem. Commun., 2003,
35–43. 2734–2735.
16 P. T. M. Do, J. R. McAtee, D. A. Watson and R. F. Lobo, ACS 33 C.-C. Chang, Z. Wang, P. Dornath, H. J. Cho and W. Fan,
Catal.., 2013, 3, 41–46. RSC Adv., 2012, 2, 10475–10477.
17 D. Wang, C. M. Osmundsen, E. Taarning and J. A. Dumesic, 34 T. Blasco, M. A. Camblor, A. Corma, P. Esteve, A. Martinez,
ChemCatChem, 2013, 5, 2044–2050.
18 R. A. Sheldon, Green Chem., 2014, 16, 950–963.
19 A. Morschbacker, Polym. Rev., 2009, 49, 79–84.
C. Prieto and S. Valencia, Chem. Commun., 1996, 2367–2368.
35 J. Luo, J. Yu, R. J. Gorte, E. Mahmoud, D. G. Vlachos and
M. A. Smith, Catal. Sci. Technol.., 2014, 4, 3074–3081.
This journal is © The Royal Society of Chemistry 2015
Green Chem.