Catalysis by framework zinc in silica-based molecular sieves

Article abstract of DOI:10.1039/c5sc03889h

Microporous and mesoporous zincosilicates (e.g., CIT-6, VPI-8, Zn-MFI, and Zn-MCM-41) synthesized in the presence of alkali cations contain two broad types of Zn sites: one that is a dication analog of the monocation ion-exchangeable Al-site in aluminosilicates, while the other resembles isolated Zn sites on amorphous silica. The ratio of these sites varies, depending on the synthesis conditions of the zincosilicate. Post-synthetic strategies based on ion-exchange can alter the site distribution towards either population. Furthermore, post-synthetic introduction of isolated Zn sites of the latter type is possible for materials possessing silanol nests. Both types of sites behave as Lewis acid centers in probe-molecule IR spectroscopy, but have very different catalytic properties. Due to the unusually high adsorption energies of Lewis bases on such materials, Lewis acid catalysis is difficult at low temperatures and in solvents bearing Lewis basic functionality. However, at high temperatures, in hydrocarbon solvents, CIT-6 (Zn-beta) is able to selectively catalyze the Lewis-acid-catalyzed Diels-Alder cycloaddition-dehydration reactions of ethylene with methyl 5-(methoxymethyl)furan-2-carboxylate, a furan that can be derived quantitatively by partial oxidation of biomass-based 5-hydroxymethylfurfural. Additionally, zinc in silica-based molecular sieves is shown here to enable chemistries previously not accessible with framework Sn-, Ti- and Zr-based Lewis acid sites, e.g., the direct production of dimethyl terephthalate by Diels-Alder cycloaddition-dehydration reactions of ethylene and the dimethyl ester of furan-2,5-dicarboxilic acid.

Full text of DOI:10.1039/c5sc03889h

Chemical
Science
View Article Online
View Journal | View Issue
EDGE ARTICLE
Catalysis by framework zinc in silica-based
molecular sieves†
Cite this: Chem. Sci., 2016, 7, 2264
*
Marat Orazov and Mark E. Davis
Microporous and mesoporous zincosilicates (e.g., CIT-6, VPI-8, Zn-MFI, and Zn-MCM-41) synthesized in
the presence of alkali cations contain two broad types of Zn sites: one that is a dication analog of the
monocation ion-exchangeable Al-site in aluminosilicates, while the other resembles isolated Zn sites on
amorphous silica. The ratio of these sites varies, depending on the synthesis conditions of the
zincosilicate. Post-synthetic strategies based on ion-exchange can alter the site distribution towards
either population. Furthermore, post-synthetic introduction of isolated Zn sites of the latter type is
possible for materials possessing silanol nests. Both types of sites behave as Lewis acid centers in probe-
molecule IR spectroscopy, but have very different catalytic properties. Due to the unusually high
adsorption energies of Lewis bases on such materials, Lewis acid catalysis is difficult at low temperatures
and in solvents bearing Lewis basic functionality. However, at high temperatures, in hydrocarbon
solvents, CIT-6 (Zn-beta) is able to selectively catalyze the Lewis-acid-catalyzed Diels–Alder
cycloaddition–dehydration reactions of ethylene with methyl 5-(methoxymethyl)furan-2-carboxylate,
a furan that can be derived quantitatively by partial oxidation of biomass-based 5-hydroxymethylfurfural.
Additionally, zinc in silica-based molecular sieves is shown here to enable chemistries previously not
accessible with framework Sn-, Ti- and Zr-based Lewis acid sites, e.g., the direct production of dimethyl
terephthalate by Diels–Alder cycloaddition–dehydration reactions of ethylene and the dimethyl ester of
furan-2,5-dicarboxilic acid.
Received 14th October 2015
Accepted 4th January 2016
DOI: 10.1039/c5sc03889h
www.rsc.org/chemicalscience
some instances, be coupled with other catalytic chemistries in
“one-pot” strategies.1,6–15 The catalytic performance of these
1. Introduction
materials appears to depend strongly on the heteroatom type,
the framework in which they are located, and the particular
reaction conditions utilized. No single heterogeneous Lewis
acid catalyst has been shown to offer optimal performance for
the broad range of reactions that are believed to involve Lewis
acid activation. Thus, discovery and characterization of cata-
lytically active sites for given reactions is necessary to guide
process optimization, and must be performed on an individual
basis. Expanding the number of members in this library of
zeotypic catalysts is an ongoing effort in the eld, with attempts
being made to both increase the number of usable zeolitic
frameworks, and enable the incorporation of catalytically
pertinent metal centers in a controlled manner.
Zinc is a common metal center in a number of homoge-
neous, Lewis acid catalysts (both in synthetic complexes and
naturally-occurring enzymes).16–18 While Zn2+ ions exchanged
onto aluminosilicate zeolites have been considered for
a number of catalytic applications, to the best of our knowledge,
evidence of heterogeneous catalysis performed by framework
Zn sites in otherwise pure-silica molecular sieves is sparse.19,20
For instance, Zn2+ exchanged onto Al-beta preferentially titrates
paired Al sites, and the resulting material can be used as
a catalyst for hydroamination reactions.19 Such Zn2+ sites have
Heterogeneous catalysts consisting of isolated Lewis acid
centers on silica-based supports have been investigated for
a wide range of reactions for the conversion of biomass into
valorized chemicals. Generation of framework, Lewis acid sites
in crystalline, pure-silica molecular sieves by the isomorphic
substitution of Si by Sn, Ti, Zr, or Hf is particularly interesting
because such sites are located in pores that have diameters
comparable to those of substrates, thus giving rise to the
possibility of shape-selective catalysis and support-induced
stabilization of transition states or intermediates.1–5 These
materials can also exhibit higher acid site stability, with lower
tendency for thermal ion migration and sintering into bulk
oxides, than analogous sites on amorphous supports. The
crystalline materials have been shown to be catalytically active
in alkane oxidation, alkene epoxidation, aromatics hydroxyl-
ation, Baeyer–Villiger (BV) oxidation, Meerwein–Ponndorf–Ver-
ley–Oppenauer (MPVO), sugar isomerization, retro-aldol, and
Diels–Alder cycloaddition–dehydration reactions and can, in
Chemical Engineering, California Institute of Technology, Pasadena, 91125, USA.
E-mail: mdavis@cheme.caltech.edu
† Electronic supplementary information (ESI) available. See DOI:
10.1039/c5sc03889h
2264 | Chem. Sci., 2016, 7, 2264–2274
This journal is © The Royal Society of Chemistry 2016

View Article Online
Edge Article
Chemical Science
been extensively characterized and were demonstrated to have (Aldrich) was dissolved in distilled water and the pH was
very strong interactions with Lewis basic probe molecules such increased to 6 by addition of ammonia solution (Mallinckrodt).
as acetonitrile and pyridine, with characteristic desorption The white precipitate that formed (presumably Zn(OH)2) was
temperatures considerably higher than those observed in our ltered, washed thoroughly with distilled water, and dried.
previous studies of Lewis acidic zeotypes containing Ti, Zr, or Sodium hydroxide (Mallinckrodt) and TPAOH (Acros Organics)
Sn.19,21 However, in all such samples, residual Brønsted acidity, dissolved in distilled water were added to the precipitate and
originating from isolated Al sites that are not readily exchanged stirred until the precipitate dissolved. Ludox AS-40 was added to
by Zn2+, is observed.19 These results, as well as literature on the solution with stirring. The resulting homogeneous gel of
amorphous silica materials bearing isolated Zn sites,22–24 have composition 1SiO2/0.067ZnO/0.105TPAOH/0.107Na2O/14.6H2O
prompted us to investigate the properties of CIT-6, an easily- was charged into a Teon-lined, stainless steel autoclave and
synthesized zincosilicate analog of zeolite beta, whose Lewis heated in a rotating oven at 175 ꢀC for 4 days under autogenous
and Brønsted acidity have not been sufficiently characterized, pressure.
despite its rst reported synthesis dating back to 1999.25,26
2.1.4. Zn-MCM-41 synthesis. Zn-MCM-41 was synthesized
Instead of direct catalysis by Zn sites, to date, CIT-6 has been according to the method reported by Takewaki et al.30 Tetra-
mainly used as a support for other metal centers, e.g., Ni2+ or ethylorthosilicate (Aldrich), zinc acetate dihydrate, cetyl-
Pt2+ ions exchanged onto the zincosilicate, or aluminum trimethylammonium bromide (Aldrich), and sodium hydroxide
inserted into the silanol nests of its de-zincated form.25,27,28 were mixed at ambient temperature for 2 h to form a gel of
Here, we report our characterization of CIT-6 and other similar composition
1SiO2/0.02Zn(OAc)2/0.61C16TMABr/0.5NaOH/
zincosilicates by probe-molecule FTIR spectroscopy and eval- 4EtOH/30H2O. The gel was charged into a Teon-lined, stain-
uate their catalytic properties in the context of Lewis acid less steel autoclave and heated statically at 105 ꢀC for 3 days
mediated reactions, namely: isomerization of glucose to fruc- under autogenous pressure.
tose, MPVO reactions of cyclohexanone and 2-butanol, and
2.1.5. SSZ-33 synthesis. The borosilicate SSZ-33 was
Diels–Alder cycloaddition–dehydration reactions of partially synthesized according to the method reported by Dartt and
oxidized variants of 5-hydroxymethylfurfural (5-HMF). We show Davis.9 Fumed silica (Cab-O-Sil), boric acid (J. T. Baker), N,N,N-
that such Zn zeotypes have exceedingly strong interactions with trimethyltricyclo[5.2.1.02,6] decaneammonium–hydroxide (R–
Lewis basic substrates. While such high interaction strengths OH) (provided by Dr Stacey I. Zones of Chevron Energy Tech-
can limit the conditions where catalysis is feasible, under nology Company), and sodium hydroxide were mixed for 1 h to
appropriate conditions, these materials may enable chemistries form a synthesis gel of composition 1SiO2/0.0125B2O3/0.2ROH/
that previously were effectively inaccessible, e.g., Diels–Alder 0.1NaOH/40H2O. The gel was charged into a Teon-lined,
cycloaddition–dehydration reactions of the dimethyl ester of stainless steel autoclave and heated in a rotating oven at 160 ꢀC
furan-2,5-dicarboxylic acid.
for 10 days under autogenous pressure.
2.1.6. Zr-beta synthesis. Zr-beta was synthesized in uoride
media according to the methods adopted by Pacheco and
Davis.13 Tetraethylorthosilicate was partially hydrolysed in
a solution of tetraethylammonium hydroxide for 30 min, fol-
2. Experimental
2.1. Microporous materials synthesis
The syntheses of microporous and mesoporous materials used lowed by the addition of zirconium(IV) propoxide (Aldrich) in
in this study are standard and are reported elsewhere, but are ethanol. The alkoxides were allowed to hydrolyse overnight and
briey outlined below. In each case, as-synthesized solids were excess water and alcohols were evaporated. Finally hydrouoric
recovered by centrifugation, washed thoroughly with distilled acid (HF) (Aldrich) was added to form a synthesis gel of
water and acetone (Fisher Scientic), dried at 100 ꢀC, and composition 1SiO2/0.1ZrO2/0.54TEAOH/0.54HF/6.75H2O. Deal-
calcined in 100 mL minꢁ1 owing air (Air Liquide, breathing uminated Al-beta seeds (prepared according to the method re-
grade) at 580 C (ramped up at 1 C minꢁ1) for 6 h.
ported by Chang et al.)31 in water were dispersed in the Zr-beta
ꢀ
ꢀ
2.1.1. CIT-6 synthesis. CIT-6 was synthesized according to gel prior to crystallization (at 4 wt% loading of SiO2). The gels
the method reported by Takewaki et al.25 Colloidal silica (Ludox was charged into a Teon-lined, stainless steel autoclave and
AS-40), zinc acetate dihydrate (Aldrich), tetraethylammonium heated in a rotating oven at 140 ꢀC for 7 days under autogenous
hydroxide (Aldrich), and lithium hydroxide monohydrate pressure.
(Aldrich) were mixed to form a clear synthesis gel of composi-
2.1.7. Generation of silanol nests by heteroatom removal.
tion 1SiO2/0.03Zn(OAc)2/0.65TEAOH/0.05LiOH/30H2O. The gel Zn was inserted into some materials that contained silanol
was charged into a Teon-lined, stainless steel autoclave and nests generated by removal of Zn or B by treatment of calcined
ꢀ
heated statically at 140 C for 7.5 days under autogenous pres- zeolite powders with 1 M aqueous H2SO4 (Macron) (at 0.1 g
sure. A single large batch (8 g SiO2 in gel) of the material was solid/10 mL solution) at ambient temperatures for 12 h. The
synthesized and used throughout the study.
nal solids were recovered by centrifugation, thoroughly
2.1.2. VPI-8 synthesis. VPI-8 was synthesized from the same washed and calcined prior to further use.
ꢀ
gel as CIT-6, above, but at a higher temperature (150 C) and
longer crystallization times (14 days).
2.1.8. Post-synthetic Zn insertion. The zinc insertion
procedure was adapted from Kozawa.24 Material possessing
2.1.3. Zn-MFI synthesis. Zn-MFI was synthesized according silanols were contacted with an aqueous solution of 0.1 M ZnCl2
to the method reported by BP.29 Zinc sulfate heptahydrate (EM Science), and 2.0 M NH4Cl (Mallinckrodt) (adjusted to
This journal is © The Royal Society of Chemistry 2016
Chem. Sci., 2016, 7, 2264–2274 | 2265

View Article Online
Chemical Science
Edge Article
desired pH by NH4OH or HCl), using 1 g solid/25 mL solution. cryoprobe. Carbohydrate analysis was performed via high
The dispersed solids were stirred in such solutions for 12 h at performance liquid chromatography on an Agilent 1200 system
ambient temperature.
equipped with refractive index and evaporative light scattering
Na/Zn/Al-beta and Zn/Al-beta were generated by adapting the detectors. An Agilent Hi-Plex Ca column at 80 ꢀC was used with
Zn-exchange procedure of Penzien et al.19 0.06 M Zn(OAc)2 ultrapure water as the mobile phase (ow rate of 0.6 mL minꢁ1).
aqueous solution was contacted with calcined Na–Al-beta or H– Quantitative GC-FID analysis of MPVO and DA cycloaddition
Al-beta (Tosoh), at 80 ꢀC, for 24 h, using 1 g solid/25 mL dehydration reactions was performed on an Agilent 7890B GC
solution.
system equipped with a ame ionization detector and an Agi-
Aer the exchange procedures, the solids were recovered by lent HP-5 column. Qualitative GC-MS analysis of products was
centrifugation, washed twice with distilled water (1 g solid/25 performed on an Agilent 5890 GC system with an Agilent 5970
mL water), dried, and calcined, as previously described.
mass spectrometer and an Agilent DB-5 column.
2.3.1. Glucose isomerization reactions. Reactions of
glucose (Aldrich) (1 wt% in water or methanol (EMD Millipore)
solvent) catalysed by CIT-6 were performed in 10 mL thick-
walled crimp-sealed glass reactors (VWR) that were heated in
a temperature-controlled oil bath. Reactions were performed at
100 ꢀC, with a 1 : 50 Zn : glucose initial molar ratio. Aliquots
(ꢂ100 mL) were extracted at indicated times, mixed with
a mannitol solution (external standard), ltered with a 0.2 mm
PTFE syringe lter, and analysed by HPLC. To determine the
mechanism of fructose formation, glucose 13C-enriched at the
C1 position (13C-C1-glucoꢀse) (Cambridge Isotope) was reacted
with CIT-6 in D2O at 100 C for 1 h. The product solution was
ltered and analysed by liquid NMR directly.
2.3.2. MPVO reactions. Reactions of cyclohexanone
(Aldrich) and 2-butanol (Fisher Scientic) catalysed by CIT-6
were performed in 10 mL thick-walled crimp-sealed glass reac-
tors (VWR) that were heated in a temperature-controlled oil
bath. Reactions were performed at 100 ꢀC, in cyclohexane
solvent. The cyclohexanone concentration was xed at 0.1 M for
all reactions, and the initial ratio of Zn : cyclohexanone was
1 : 100. Naphthalene was used as an internal standard, and
reactions were analysed by GC-FID. The turnover frequency at
a given 2-butanol concentration was calculated from the initial
rate of formation of cyclohexanol.
2.2. Characterization of solids
Scanning electron microscopy (SEM) with energy dispersive X-
ray spectroscopy (EDS) measurements were recorded on a LEO
1550 VP FE SEM at an electron high tension (EHT) of 15 kV. The
crystalline structures of zeolite samples were determined from
powder X-ray diffraction (XRD) patterns collected using
a Rigaku Miniex II diffractometer and Cu Ka radiation. Ther-
mogravimetric analysis (TGA) under an air atmosphere was
performed on a PerkinElmer STA 6000 with a ramp of 10 ꢀC
minꢁ1 up to 900 ꢀC. Probe molecule IR spectroscopy experi-
ments were performed on a Nicolet Nexus 470 Fourier trans-
form infrared (FTIR) spectrometer with a liquid N2 cooled Hg–
Cd–Te (MCT) detector. Spectra in 4000ꢁ650 cmꢁ1 range were
acquired with 2 cmꢁ1 resolution. Self-supporting wafers (10ꢁ20
mg cmꢁ2) were pressed and sealed in a heatable quartz vacuum
cell with removable KBr windows. The cell was purged with air
(60 mL minꢁ1, Air Liquide, breathing grade) while heating to
500 ꢀC (1 ꢀC minꢁ1), where it was held for 1 h, followed by
evacuation at 500 ꢀC for >2 h (<0.01 Pa dynamic vacuum; oil
diffusion pump), and cooling to 35 ꢀC under a dynamic vacuum.
CD3CN (Sigma-Aldrich, 99.8% D atoms) or pyridine (EMD Mil-
lipore) was degassed by three freeze (liquid N2), pump, thaw
ꢀ
cycles, then dosed to the sample at 35 C until the Lewis acid
2.3.3. Diels–Alder reactions. The procedure for Diels–Alder
reactions was adapted from Pacheco and Davis,13 but was
modied for quantication by GC-FID. Reactions of methyl 5-
(methoxymethyl)furan-2-carboxylate (MMFC) (enamine) or
dimethyl 2,5-furandicarboxylate (DMFDC) (Matrix Scientic)
with ethylene (Matheson) were carried out in a 50 mL high-
pressure stainless steel batch reactor (Parr Series 4590) equip-
ped with a magnetic stirrer (operated at 200 rpm) and heater.
For MMFC reactions, 10 mL of a 0.1 M diene solution in
heptane (Aldrich) and 100 mg catalyst were loaded into the
reactor. DMFDC is poorly soluble in heptane at low tempera-
tures, so this diene, along with 100 mg catalyst were loaded
directly into the reactor, and 10 mL of heptane was added to
give a nominal concentration of 0.33 M of diene. Decane was
also added as an internal standard for GC-FID quantication. At
the start of a reaction, the head space of the reactor was purged
with helium gas with a ll/vent cycle (10 times). Next, the reactor
was pressurized to 35 bar with ethylene gas at ambient
temperature. The reactor was heated to reaction temperature
sites were saturated, at which point physisorbed and gas-phase
species were observed. The cell was evacuated down to 13.3 Pa,
and the rst spectrum was recorded. Then, the cell was evacu-
ated under a dynamic vacuum at 35 ꢀC for 24 h, aer which the
second spectrum was acquired. Subsequent heating to
temperatures specied in plots of spectra was performed at 5 ꢀC
minꢁ1, with a subsequent hold time of 0.5 h or greater (as
specied in gure legends) prior to acquisition of the next
spectrum. The resulting spectra were baseline-corrected by
subtracting the spectrum of the pellet at same temperature
prior to adsorption of the probe molecule. The spectra are not
normalized by the number of Lewis acid sites or framework
vibrations. Spectral artifacts known as “interference fringes”
were apparent in some spectra and were removed using
a computational method based on digital ltering techniques
and Fourier analysis.32
2.3. Catalytic testing
Liquid 1H and 13C NMR spectra were acquired on a Varian 500 while the pressure increased autogenously (ꢂ60ꢁ80 bar).
MHz spectrometer equipped with an auto-x pfg broadband Reaction time was started when the contents of the vessel
probe and
a Bruker 400 MHz with Prodigy broadband reached desired temperature, and aer a specied time, the
2266 | Chem. Sci., 2016, 7, 2264–2274
This journal is © The Royal Society of Chemistry 2016

View Article Online
Edge Article
Chemical Science
reactor was quenched with water and allowed to cool to ambient
temperature. At this point, the reactor gases were carefully
vented. Solution aliquots that were collected for GC analysis
were ltered with a 0.2 mm PTFE syringe lter. The MMFC
reaction solutions were analysed directly, while for the DMFDC
system, 20 mL of acetone were added to solubilize components
not readily soluble in heptane. In both cases, aliquots taken for
NMR studies were ltered, rotavaped, and redissolved in
acetone-d6 (Cambridge Isotope). Samples of post-reaction
catalysts intended for TGA were isolated from reaction solution
by centrifugation, washed twice with either heptane (for MMFC
reactions) or acetone (for DMFDC reactions), and dried at 100
ꢀC overnight. TGA of corresponding washed, unreacted catalysts
was used to account for any strongly-retained solvent.
Fig. 1 Baseline corrected IR spectra of pyridine adsorbed on CIT-6 at
35 ꢀC. Different colors (indicated in legend) correspond to different
subsequent desorption temperatures carried out for 1 h. Peaks cor-
responding to pyridine coordinated to Lewis acid (L), Brønsted acid (B),
and hydrogen-bonding (H) sites are marked.
Catalyst recycle experiments were performed for the Diels–
Alder cycloaddition–dehydration reaction of DMFDC and
ethylene at 210 C, with CIT-6-reZn-pH ¼ 6.9. First two reruns
ꢀ
were performed with samples that were triply washed with
acetone and dried at 100 ꢀC overnight. The third rerun was
performed on recalcined catalyst recovered aer the second
rerun. In all instances reagent and solvent ratios were adjusted
to keep constant ratio to inorganic content, as determined by
TGA. SEM-EDS analysis of recovered and acetone-washed cata-
lysts was performed. The catalyst used in the third rerun was
also analysed by XRD and CD3CN adsorption tracked by IR.
and to protonated pyridine on strong Brønsted acid sites
(C5H5NH+ species)33,34 (Fig. S1†). Furthermore, all such adsor-
ꢀ
bed pyridine desorbs upon evacuation at 300 C (Fig. S1†).
Adsorption of CD3CN reveals the presence of at least two
Lewis acid sites in CIT-6 (Fig. 2), with deconvoluted bands
appearing at 2311 cmꢁ1 and 2290 cmꢁ1. The frequency of the
2311 cmꢁ1 band of CIT-6 suggests an extent of polarization
comparable to that generated by Sn-MCM-41, Sn-MFI, and Zr-
beta (ca. 2309–2312 cmꢁ1), but lower than that generated by the
“open” Sn site of Sn-beta (2315 cmꢁ1) or Al Lewis acid sites in
various zeolites ($2320 cmꢁ1).33,35,36 This band is intermediate
between the 2314 cmꢁ1 band reported for the Zn-exchanged Al-
beta19 and the 2305 cmꢁ1 band measured for Zn sites dispersed
on amorphous silica (Fig. S2†). The lower frequency of the band
measured for the amorphous-supported Zn sites is consistent
with the trend observed for Sn in Sn-beta and Sn-MCM-41.36,37
3. Results and discussion
3.1. Probe molecule FTIR spectroscopy
The presence and character of Lewis and Brønsted acid sites in
solid catalysts may be probed by following the adsorption and
desorption behavior of Lewis basic molecules through FTIR
spectroscopy. Pyridine and deuterated acetonitrile (CD3CN)
have been routinely used as Lewis bases for this purpose.33,34
While the signicantly different vibrational modes of pyridine
coordinated to a Lewis acid center and pyridinium ion gener-
ated from protonation of pyridine by a Brønsted acid allow for
easy determination of the presence of the two kinds of sites,
more subtle differences that differentiate one kind of Lewis acid
center from another are harder to discern. Thus, CD3CN, whose
CN stretching frequency tends to increase with the strength of
the interaction with a Lewis acid center, can be used as
a complimentary probe molecule to qualitatively compare the
site of interest with other previously studied sites.33 Addition-
ally, desorption temperatures of such molecules reect the
strength of interactions a given functional group may be ex-
pected to have with the site of interest.
Pyridine adsorption onto CIT-6 (Fig. 1) results in IR bands
characteristic of pyridine interacting with Lewis acid sites (1451,
1491, and 1610 cmꢁ1) and hydrogen bonded pyridine (ca. 1575
and 1446 cmꢁ1).23 No band characteristic of Brønsted acid sites
(ca. 1550 cmꢁ1) is observed. Pyrꢀidine remains adsorbed on the
Lewis acid sites up to 300–350 C. In contrast, pyridine dosed
onto bulk ZnO results in bands characteristic of Lewis acid sites
(1451 and 1610 cmꢁ1), hydrogen-bonded pyridine (1574 cmꢁ1),
and a broad band in the range previously assigned to both
Fig. 2 Baseline corrected IR spectra of CD3CN adsorbed on CIT-6 at
35 ꢀC (green solid) and desorbed at different times and temperatures:
24 h at 35 ꢀC (green, dashed), 0.5 h at 100 ꢀC (black, solid), 1.5 h at 100
ꢀC (black, dashed), 0.5 h at 200 ꢀC (orange, solid), 0.5 h at 300 ꢀC (red,
dissociatively-adsorbed pyridine on base sites (C5H4Nꢁ species) solid), and 0.5 h at 350 ꢀC (blue, solid).
This journal is © The Royal Society of Chemistry 2016
Chem. Sci., 2016, 7, 2264–2274 | 2267

View Article Online
Chemical Science
Edge Article
Silica-supported Zn sites have been previously characterized in counterparts. In fact, the persistence of coordinated CD3CN on
ꢀ
the context of alkane dehydrogenation. EXAFS data suggest that the CIT-6 sites to temperatures beyond 200 C under vacuum
the dehydrated Zn sites in this type of material datively coor- (Fig. 2) is consistent with the high temperatures of TPD
dinate an oxygen of a neighboring silanol (as shown in structure desorption peaks for Zn-exchanged Al-beta.19 The high interac-
Z0 in Scheme 1),22 but this polarization appears insufficient to tion strength of CD3CN with such sites (as inferred from high
generate a strong Brønsted acid site that is capable of proton- desorption temperatures) appears to conict with the relatively
ating pyridine.
low induced blue shis of the CN vibration. This disparity may
The 2290 cmꢁ1 band of CD3CN adsorbed on CIT-6 is likely stem from the difference in the energies of structural rear-
associated with Li-bearing Zn sites. We previously observed rangements for Zn sites vs. Sn, Ti, or Zr sites upon desorption of
a CD3CN band ca. 2292 cmꢁ1 in a Sn-beta material that was probe molecules.
exchanged with Li+ under basic conditions in order to cationate
CD3CN adsorbed on bulk ZnO generates spectroscopic
the neighboring silanol of the “open” Sn site.21 Similarly, Na+ signatures distinct from CIT-6 (Fig. S4†), with a broad band ca.
and K+ exchanges generate sites with characteristic CD3CN 2300 cmꢁ1 that can be attributed to Lewis or Brønsted acid sites,
bands ca. 2280 and 2273 cmꢁ1, respectively. VPI-8 (VET frame- as well as a multitude of bands below 2200 cmꢁ1, characteristic
work) is a zincosilicate with a higher framework density than of CD2CNꢁ and polyanions formed by deprotonation of CD3CN
CIT-6 that can be found as a minor phase impurity in CIT-6 by strongly basic surface oxygen species.34 While the 2300 cmꢁ1
powders and is difficult to detect at low concentrations (as band is not spectroscopically resolved from the bands observed
demonstrated by the data shown in Fig. S3†). VPI-8 crystallizes if in CIT-6, the lack of CD2CNꢁ signatures in the CIT-6 spectra
the CIT-6 gel (which contains Li+ ions) is aged beyond complete suggests absence of detectable amounts of extra-framework
CIT-6 crystallization38 and exhibits a CD3CN band primarily ca. ZnO in CIT-6 aer calcination.
2294 cmꢁ1. On the other hand, Zn-MCM-41 and Zn-MFI are
synthesized from gels containing Na+ ions and CD3CN bands ca.
2280 cmꢁ1 are observed for these materials. These data are
3.2. Catalysis with microporous zincosilicates
consistent with the presence of Zn sites possessing structures Z1
and/or Z2 in Scheme 1. Further support for this tentative
isolated Sn sites in SiO2-based materials and by SnOx particles
We have previously explored reactions of sugars catalyzed by
located in the pores of Si-beta.39 The isolated “open” Sn sites
that have an adjacent protonated silanol were found to promote
glucose–fructose isomerization through a 1,2-intramolecular
hydride shi mechanism, while SnOx particles behaved as base
catalysts by promoting the same isomerization through an
enolate mechanism that involves deprotonation of a-carbonyl
carbon (Scheme S1†).21,39 Data provided in Fig. 3 show that
under the same reaction conditions as we used previously, CIT-6
isomerizes glucose to fructose in both water and methanol
solvents, but does so with proton abstraction from the a-
carbonyl carbon (as evidenced by 13C NMR of isotopically
labeled glucose; Fig. S5†). While the reaction appears to proceed
catalytically (TOF > 1 in water and methanol), the reaction slows
assignment comes from ion exchange experiments that shi the
site distribution in CIT-6. A moderately basic Li+ exchange (1 M
LiNO3, and initial pH ¼ 10, set by LiOH) generates a material
(CIT-6-LiEx) possessing primarily a 2290 cmꢁ1 CD3CN band. On
the other hand, exchanging CIT-6 with a nearly-neutral 1 M
solution of N(CH3)4Cl, followed by calcination, produces
a material (CIT-6-Z0) possessing primarily a 2312 cmꢁ1 CD3CN
band, with a broad red-shied shoulder. It is important to note
that, within the measurement error of energy dispersive spec-
troscopy (EDS), CIT-6-LiEx (Si/Zn ¼ 10.9 ꢃ 2.1) has the same Zn
content as the parent material (Si/Zn ¼ 12.2 ꢃ 0.9), but CIT-6-Z0
loses nearly half of its Zn (Si/Zn ¼ 21.3 ꢃ 4.2). Materials with
higher total Zn contents, with CD3CN bands ca. 2312–2310
cmꢁ1, can be generated by an alternative post-synthetic strategy
(vide infra). IR spectra of CD3CN adsorbed on VPI-8, Zn-MCM-
41, Zn-MFI, and a number of post-synthetically Zn-modied
materials are shown in Fig. S2† (powder XRD data of these
microporous materials are included in Fig. S3†).
In our hands, CD3CN appears to interact much stronger with
the two discernable Lewis acid sites in calcined CIT-6, than with
any of the sites in Sn-, Ti-, or Zr-beta, or their alkali-exchanged
Fig. 3 Glucose isomerization reactions are catalyzed by CIT-6 in
water (left) and methanol (right) solvents. Glucose conversion (red
Scheme 1 Proposed framework Zn site structures in microporous squares), fructose yields (blue circles), and mannose yields (green
zincosilicates. M+ is a monovalent cation, such as alkali or alkyl triangles) are plotted as a function of reaction time. Reaction condi-
ammonium.
tions: 100 ꢀC, 1% (wt/wt) glucose, 1 : 50 Zn : glucose initial molar ratio.
2268 | Chem. Sci., 2016, 7, 2264–2274
This journal is © The Royal Society of Chemistry 2016

View Article Online
Edge Article
Chemical Science
prior to reaching an equilibrium distribution of sugars, sug-
gesting that catalyst deactivation occurs (Fig. S6†).
EDS analysis indicates a 35% decrease in Zn content of the
CIT-6 sample aer reaction in water. Recalcination of the
catalyst recovered and washed aer such a reaction does not
result in the recovery of isomerization activity. Additionally,
CD3CN adsorption on this material reveals a loss of the 2290
cmꢁ1 band that is present in the original CIT-6 sample, but
retention of the 2311 cmꢁ1 band (Fig. S7†). Though ZnO and
Zn(OH)2 can catalyze the isomerization glucose through the
enolate pathway, it is unlikely that the isomerization activity
observed in the original CIT-6 is attributable to such species, as
their presence is not observed in spectroscopic characteriza-
tion, and the permanent catalyst deactivation is inconsistent
with their presence. The catalytic data obtained from CIT-6 are
more consistent with involvement of Z1 or Z2 sites that we
hypothesize are correlated to the 2290 cmꢁ1 CD3CN band. The
enhanced basicity of framework oxygens whose charge is
Fig. 4 Initial TOF of MPVO reactions of cyclohexanone and 2-butanol
catalyzed by CIT-6 as a function of 2-butanol concentration. Reaction
conditions: 100 ꢀC, 0.2 M cyclohexanone, indicated concentration of
2-butanol in cyclohexane, and 1 : 100 Zn : cyclohexanone initial molar
ratio.
balanced by alkali cations may allow for participation of Z1 or Z2 catalytic activity of the Zn sites and their hindered desorption
sites in this base-promoted reaction. The generation of small rates, or if severe diffusion limitations induced by high
quantities of a-hydroxycarboxylic acid species (e.g., lactic acid), concentrations of strongly bound species also contribute to the
as ascertained by 1H and 13C NMR, suggests acid-catalyzed slow measured kinetics.
hydrolysis of Z1 and Z2 sites as the possible mode of
deactivation.
Sabatier's principle can be used to qualitatively rationalize
the poor catalytic performance observed for MPVO reactions
While CIT-6 is an active catalyst for the glucose–fructose promoted by CIT-6 relative to the performance of other Lewis
isomerization, this reaction is not proceeding via catalysis by acidic beta zeotypes known to catalyze such reactions,2,40 and to
the Lewis acid sites (as probed by base adsorption and IR). The infer reaction conditions where CIT-6 may behave as a catalyti-
pKa values of the conjugate acids of nitriles are higher than cally interesting material. The principle states that for a reaction
those of the conjugate acids of aldehydes, ketones, alcohols, with a given activation energy, an optimal enthalpy of desorp-
and water. Outside of solvation effects, this relative ranking tion of the substrate from the heterogeneous catalyst exists. For
implies that nitriles should have weaker interactions with Lewis MPVO reactions, Ti-beta appears to interact too weakly and CIT-
acid sites than the other Lewis basic species listed above. Data 6 too strongly with the carbonyl bearing substrates (as inferred
provide in Fig. 2 show that vaꢀcuum desorption of CD3CN from from desorption of cyclohexanone), with both cases resulting in
CIT-6 Lewis acid sites at 100 C is slow, with minimal desorp- low reaction rates. Sn-beta and Zr-beta have desorption rates
tion occurring over the course of an hour. Desorption of water, intermediate to the two extremes and result in signicantly
methanol, and cyclohexanone were also observed to be slow at higher reaction rates at a given temperature. Because these
100 ꢀC, and appreciable desorption rates only occurred at materials usually possess more than one kind of coordination
temperatures higher than 200 ꢀC. These low rates of desorption environment for the heteroatoms, quantitative statements of
of probe molecules are expected to translate to slow desorption these results is presently not possible due to a lack of
of these types of compounds present in solutions at reaction measurements of site-specic reaction rates and adsorption
conditions. Thus, the Lewis acid-mediated catalytic properties enthalpies. Another outcome of Sabatier's principle suggests
of Zn sites in CIT-6 at low temperatures are mitigated in that reactions with higher intrinsic activation barriers will
solvents possessing strongly Lewis basic functional groups. This generally correspond to higher optimal adsorption enthalpies.
interpretation is supported by the observed rate behavior of Thus, the Lewis acid sites of CIT-6 have the potential to catalyze
Lewis-acid mediated MPVO reactions of cyclohexanone and 2- high-temperature reactions more optimally than the Lewis acid
butanol catalyzed by CIT-6. Data in Fig. 4 show the measured sites in weaker binding materials such as in Ti-, Zr-, and Sn-
initial rates for this reaction as a function of 2-butanol beta.
concentration, for a xed concentration of cyclohexanone (0.2
The Diels–Alder (DA) cycloaddition–dehydration reactions of
M). The initial rate of the reaction increases with increasing 2- substituted furans with ethylene are a promising route to ter-
butanol concentration, but peaks at a concentration where the ephthalic acid made from renewable sources (Scheme 2). Such
ratio of the two reactants is stoichiometric (0.2 M), and reactions of dimethyl furan (R1 ¼ R2 ¼ CH3 in Scheme 2) can be
decreases with further increase in 2-butanol concentration. This catalyzed more efficiently by Brønsted acids than Lewis
reaction behavior is consistent with kinetically limiting acids,41,42 but furans with oxygenated side-groups that can be
desorption rates. The nominal TOF of CIT-6 for this reaction derived from 5-hydroxymethylfurfural without costly reduction
(based on total Zn) is 3–4 orders of magnitude lower than that of steps react on Brønsted acidic zeolites with negligible selectiv-
Sn-beta (based on total Sn) under similar reaction conditions.3 ities towards desired DA products under similar or milder
At this point, it is not clear if the low TOF reects the intrinsic conditions.13 Only the Lewis-acidic, Sn- and Zr-beta were
This journal is © The Royal Society of Chemistry 2016
Chem. Sci., 2016, 7, 2264–2274 | 2269

View Article Online
Chemical Science
Edge Article
framework Zn in CIT-6 that is catalytically active in this reac-
tion. CIT-6-LiEx is signicantly less active than the parent CIT-6,
suggesting that the CIT-6 sites associated with 2290 cmꢁ1
CD3CN IR band are unable to effectively catalyze these reac-
tions. CIT-6-Z0 primarily has sites associated with the 2311
cmꢁ1 CD3CN IR band and, despite its lower Zn content, results
in higher MMBC yields than the parent CIT-6 material. These
data implicate the Z0 sites as the catalytically active species in
such DA cycloaddition–dehydration reactions. However, site
cooperativity in mixed-site samples of CIT-6 is another possi-
bility that warrants consideration in future studies. Because
determination of cooperativity would require quantitation of
each type of site, determination of their proximity, and
measurement of intrinsic site kinetics, such efforts are outside
of the scope of this work.
Scheme 2 Generalized description of Diels–Alder cycloaddition–
dehydration reactions of substituted furans (R1, R2 ¼ CH3, CH2OR,
CHO, or CHOOR, where R ¼ H or alkyl group). This is a multistep
process, whose rate limiting steps are greatly influenced by the identity
of the R1 and R2 groups.
previously observed to catalyze the DA reactions of such
substrates with appreciable selectivity, with Zr-beta resulting in
considerably higher selectivities, for reasons currently not
understood.13 The apparent absence of strong Brønsted acid
sites in CIT-6 makes it a candidate catalyst for these kinds of DA
reactions.
While other porous zincosilicate materials (Zn-MCM-41, Zn-
MFI, VPI-8, and deboronated SSZ-33 that has been post-
synthetically zincated) have varying distributions of Zn Lewis
acid sites (i.e., different proportions of Z0, Z1, and Z2 type of sites
as characterized by CD3CN IR experiments (Fig. S2†)), all of
these materials produce negligible levels of MMBC and result in
minor conversion of MMFC and appear brown in color aer
reaction (Table S1†). The reason for the lack of activity of Zn
sites in these materials remains unknown. However, the 10MR
pores in Zn-MFI may be too small. VPI-8 (VET structure) has a 1-
D linear 12MR pore system and SSZ-33 (CON structure) has
12MR pores intersected by 10 MR pores, so both materials
should be able to accommodate molecules that can enter the
*BEA framework, but the additional limitation of 1-D diffusion
may render these materials effectively inactive. The lack of
activity of the un-constrained sites in Zn-MCM-41 or Zn on
amorphous silica are hard to reconcile, but these results are
In heptane solvent, CIT-6 catalyzes the formation of methyl
4-(methoxymethyl) benzenecarboxylate (MMBC) in the reaction
of methyl 5-(methoxymethyl)furan-2-carboxylate (MMFC) with
ethylene at 190 ꢀC, with a yield of 16% at 51% selectivity (Fig. 5).
No signicant quantities of soluble byproducts were detected by
1H NMR (Fig. S8†) or observed in the GC chromatograms.
Interestingly, no conversion is observed in dioxane, the solvent
found to give the best selectivity for this reaction when Zr-beta
or Sn-beta catalysts are used.13 This result is consistent with the
hypothesis that, in CIT-6, there is Lewis acid site passivation
through competitive binding of oxygenated species, even at
these relatively high temperatures. Furthermore, Brønsted acid
sites are expected to not be passivated in dioxane, and the lack
of MMFC conversion in this solvent suggests that no catalyti-
cally relevant Brønsted acid sites are accessible at the reaction
temperatures. Additionally, bulk ZnO produces no detectable
conversion of MMFC in heptane, demonstrating that it is the
Fig. 5 Diels–Alder cycloaddition–dehydration reactions of MMFC with ethylene catalyzed by CIT-6 and its various modified forms with different
Zn contents and site distributions.
2270 | Chem. Sci., 2016, 7, 2264–2274
This journal is © The Royal Society of Chemistry 2016

View Article Online
Edge Article
Chemical Science
consistent with the reported lack of activity of Sn- and Zr-MCM- here, is still lower in heptane than that reported for Zr-beta in
41,13 and suggest that the *BEA framework may play a role in the dioxane (ꢂ70–80%), it is higher than that of Sn-beta in dioxane
stabilization of intermediates or transition states for such DA (ꢂ50%).13,43 Furthermore, a number of reactions reported here
reactions. We also note that Na–Al-beta that has been ion exceed the selectivity of Zr-beta in heptane under identical
exchanged with Zn2+ leads to negligible selectivity towards reaction conditions (measured to be 42% here), and result in
MMBC, and apparent coking of catalyst.
comparable net yields of MMBC. The MMBC selectivity of CIT-6
Assuming homogeneous dispersion of Zn in CIT-6, the can be further incꢀreased to 62% by lowering the reaction
relatively low Si/Zn ratio of the material (12.2 ꢃ 0.9) implies that temperature to 170 C (Table S1†).
4.8 Zn sites are found per unit cell. Considering that each
Another interesting and more stable furan, 2,5-fur-
MMFC molecule possesses 4 oxygen atoms, and spans a large andicarboxylic acid (FDCA), is formed upon oxidation of both
fraction of the void space of a *BEA unit cell, coordination to side-groups of 5-HMF to carboxylic acid groups, and can be
multiple Zn sites is possible at high Zn loadings. Multiple obtained in high yield and selectivity from 5-HMF, or can be
coordination points with the framework are expected to further made from fructose, with 5-HMF as an intermediate in a “one-
increase the molar enthalpy of desorption for reactants, inter- pot” scheme.44,45 The dimethyl ester of FDCA, dimethyl 2,5-
mediates, and products. The close spacing of Zn sites also furandicarboxylate (DMFDC), is used in its purication by
increases the likelihood of coupling reactions, as adsorbates vacuum distillation, and can be formed either through a sepa-
may be positioned in intimate contact for extended periods of rate esterication of FDCA, or directly and quantitatively during
time. We investigated the possibility of increasing catalyst the oxidation of 5-HMF.46 DA cycloaddition–dehydration reac-
selectivity by adapting a procedure used to ion-exchange Zn tions of ethylene with FDCA and DMFDC (R1 ¼ R2 ¼ COOH or
amines onto amorphous silica supports in order to generate COOMe in Scheme 2, respectively), are attractive direct routes to
dispersed Zn sites primarily in Z0 conguration, as character- terephthalic acid (TPA) and dimethyl terephthalate (DMT),
ized by CD3CN IR experiments (Fig. S2†).24 Briey, this proce- respectively, that do not require subsequent oxidation steps.
dure involves contacting a material possessing silanol nests Prior investigations of such reactions catalyzed by Sn-beta
with an aqueous solution containing 0.1 M ZnCl2, and 2.0 M showed that both of these substrates were highly resistant to DA
NH4Cl, adjusted to desired pH by NH4OH or HCl, followed by reactions. Only under forcing conditions (300 ꢀC) did DMT form
recovery and washing of solids, drying, and calcination. For from DMFDC in dioxane solvent, with a molar yield of 0.4% and
amorphous silica, this procedure is reported to generate mate- >1% selectivity.13,47 A BP patent reports no formation DMT, but
rials with Si/Zn ratios as low as 9 when the pH of the starting a molar yield of 0.023% of TPA, at 0.038% selectivity, when
solution is 7.4, while decreasing pH of the starting solution DMFDC is reacted with ethylene without a catalyst in toluene
leads to lower Zn contents.24 CIT-6 exposed to 1.0 M H2SO4 for solvent at 190–195 C.48 Similarly a Furanix – Coca-Cola patent
ꢀ
12 h, at ambient temperatures, (denoted CIT-6-de-Zn) loses application reports a 7.2–11.6% yield of TPA, at an undisclosed
most of its Zn (Si/Zn ¼ 188 ꢃ 17 is near the detection limit of conversion or selectivity, without any DMT observed, when
EDS), leaving behind silanol nests that can host re-introduced DMFDC is reacted with ethylene in homogeneous acid/acetic
Zn. Indeed, the application of the Zn–amine-based procedure anhydride mixtures at 240 C.49
ꢀ
leads to incorporation of Zn in CIT-6-de-Zn, with higher pH
CIT-6 is able to catalyze the formation of DMT from DMFDC
corresponding to higher Zn incorporation and higher MMBC at 190 ꢀC, with a yield of 3.4% and 5.1% selectivity (Fig. 6). GC-
yields (up to pH ¼ 6.9), at nearly identical conversions (see MS and 1H NMR spectra (Fig. S9 and S10†) of the reaction
entries in Fig. 5 denoted as CIT-6-re-Zn-pH ¼ X, where X solutions indicate signicant quantities of three byproducts:
corresponds to the initial pH in the re-zincation solution). At Si/ methyl 2-furoate (MF), methyl benzoate (MB), and 2-cyclo-
Zn ¼ 5.8 ꢃ 0.2, CIT-6-re-Zn-pH ¼ 7.4 has a nominal Zn loading hexenone (CHO), at 50.5%, 5.3%, and 2.8% yield (measured by
that exceeds that of the original material by a factor of 2. Such GC-FID), respectively. MF is hypothesized to form through
high Zn loading either suggests that the original CIT-6 material, decarboxylation reactions of DMFDC, while MB may form either
prior to de-zincation, already possesses a number of unoccu- through a DA step with ethylene from MF, or as a decarboxyl-
pied silanol nests, or that immobilization of extra-framework ation product from DMT. No signicant formation of MB was
Zn species may also occur at higher pH.
observed when DMT was used as a reactant, suggesting that the
Interestingly, CIT-6-de-Zn converts a signicant amount of latter scenario is not likely. CHO is potentially formed from the
MMFC (27%), with low selectivity towards MMBC (15%). These DA product of ethylene and furan, as shown in Scheme S2.†
data suggest that the side reactions do not correlate with Aer the DA cycloaddition, the resulting 7-oxabicyclo[2.2.1]
decreased Zn spacing, but reveal that underlying non-selective hept-2-ene adduct can rearrange to an epoxide, as is proposed to
sites may exist in CIT-6. Standard CIT-6 is synthesized in occur in the Lewis acid catalyzed dehydration of the DA adduct
hydroxide media with Ludox AS-40 as the silica source. Thus, of dimethyl furan.42 Subsequently, the epoxide can be iso-
the byproduct-forming sites may be either highly active impu- merized into the enone by a Lewis-acid-promoted hydride
rities from Ludox (Al or Fe) that are not removed by the acid shi.50 The presence of this product suggests that furan and
treatment, or the numerous silanol defects that are generated benzene should also form, but these species are only observed
upon de-zincation.
in GC-MS at very small relative amounts, and are poorly resolved
While the MMBC selectivity of CIT-6, or the various post- from solvent peaks. Furthermore, because these species are
synthetically generated Zn-containing beta zeotypes presented signicantly more volatile than any other component of the
This journal is © The Royal Society of Chemistry 2016
Chem. Sci., 2016, 7, 2264–2274 | 2271

View Article Online
Chemical Science
Edge Article
does not possess Li-bearing sites, and results in nearly double
DMT yield (5.6%) and signicantly higher selectivity (30.2%)
than the parent CIT-6 material. Furthermore, elimination of the
Li-bearing sites signicantly reduces the production of MF, MB,
and CHO to 2.1%, 0.8%, 1.7%. Analogous decarboxylation
products, 2-(methoxymethyl)furan and (methoxymethyl)
benzene, were not observed in product solutions of MMFC
reactions with Li-containing catalysts (Fig. S8†), but such
species may be intermediates in coking that is observed for the
MMFC reactions on CIT-6. While undesirable here, sites
responsible for decarboxylation reactions of FDCA may be of
interest in the context of Henkel reactions that have been shown
to be catalyzed by ZnCl2.52 We note that at 190 C, Zn/Al-beta
ꢀ
also catalyzes the formation of DMT, with a 9.0% DMT yield and
10.8% selectivity. Signicant amounts of decarboxylation
products are observed with this material, with 36.4%, 1.4%, and
0.7% yields of MF, MB, and CHO, respectively.
ꢀ
TGA mass losses above 250 C for the acetone-washed cata-
Fig. 6 Diels–Alder cycloaddition–dehydration reactions of DMFDC
catalyzed by CIT-6 and CIT-6-re-Zn-pH ¼ 6.9 at various tempera-
tures. Resulting yields (%) of DMT, MF, MB, CHO, and DMFDC are
calculated as ratio of moles formed to initial moles of DMFDC. Mass on
catalyst (%) is expressed as ratio of combustible mass on catalyst
(measured by TGA) to initial mass of DMFDC. Reaction conditions: 12
h, effective concentration of 0.033 M DMFDC in 10 mL heptane, 100
mg of catalyst, and 35 bar C2H2 at 25 ꢀC.
lyst from DMFDC reactions catalyzed by CIT-6 and CIT-6-re-Zn-
pH ¼ 6.9 correspond to 3.4% and 7.4% of the initial DMFDC
mass, respectively (aer adjusting for any solvent retention,
determined by TGA of acetone-washed, unreacted catalyst).
Thus, despite the large difference in conversion and side-
product formation, the difference in substrate adsorption or
coking is minor. Additionally, unlike the yellow-brown color of
catalysts aer reactions of MMFC, both catalysts appear white
aer DMFDC reactions. In contrast, the Zn/Al-beta catalyst
turned red-brown aer DMFDC reactions, suggesting a greater
extent of coking occurred on this catalysts, as corroborated by
the increase in TGA mass loss to 22.5% of initial DMFDC mass.
Mole balance data (based on moles DMFDC) for CIT-6 and CIT-
reaction solution they may be removed in part during vessel
depressurization. For these reasons, we did not attempt to
quantify the yields of furan and benzene in this work. Scheme 3
summarizes the hypothesized reaction network.
With Sn-beta catalysts, the disodium salt of FDCA was re-
ported to primarily decarboxylate rather than produce DA
products.47 Furthermore, decarboxylation reactions of esters are
generally hypothesized to proceed through free acid or carbox-
ylate intermediates, which may be stabilized by alkali ions.51
Thus, the Li-bearing sites of CIT-6 were suspected as a potential
source for such decarboxylation reactions. CIT-6-re-Zn-pH ¼ 6.9
ꢀ
6-re-Zn-pH ¼ 6.9, for temperatures ranging from 170 C to 230
ꢀC, are shown in Fig. 6. The yield of DMT progressively increases
with temperature, and despite the accompanying growth of MF,
MB, and CHO, the apparent selectivity for DMT also increases
up to 36.2%, at 210 ꢀC, as the combustible mass deposited onto
the catalyst does not appear to change signicantly, and
constitutes a progressively smaller fraction of the conversion.
The large decrease in mole balance closure at 230 ꢀC is not
accompanied by increase in coking or detectable formation of
new product species (as characterized by GC and NMR);
however, volatile species like benzene and furan escape quan-
tication under the current experimental protocols, and may
constitute a larger fraction of the conversion.
Preliminary investigation of catalyst stability and recycla-
bility were performed with CIT-6-reZn-pH ¼ 6.9 for the DA
ꢀ
reaction of DMFDC and ethylene at 210 C. These experiments
suggest that, while the catalyst retains activity without inter-
mediate calcination, the product distribution aer each run
changes, with decarboxylation reactions becoming more
prominent, and contributing to decreased DMT selectivity
(Fig. S11†). Furthermore, calcination of the catalyst aer the
second reuse does not restore the initial selectivity. Neither the
XRD pattern of (Fig. S12†) nor the qualitative IR spectrum of
CD3CN adsorbed on (Fig. S13†) the catalyst recovered aer the
third run are appreciably changed. Within certainty of EDS
measurements, the Si/Zn ratio of the catalyst also remained
Scheme
3 Full Diels–Alder cycloaddition–dehydration reaction
diagram for DMFDC as a substrate. In the case of CIT-6, decarboxyl-
ation reactions are primarily catalyzed by Li-bearing sites.
2272 | Chem. Sci., 2016, 7, 2264–2274
This journal is © The Royal Society of Chemistry 2016

View Article Online
Edge Article
Chemical Science
constant throughout these experiments (Fig. S14†). Thus, more SC0001004. M. O. acknowledges funding from the National
extensive recyclability tests and new probes for the changes to Science Foundation Graduate Research Fellowship Program
active site structure are needed to understand the evolution of under Grant DGE-1144469. Any opinions, ndings, and
the catalyst activity and selectivity.
conclusions or recommendations expressed in this material are
The net processes for biomass-based production of DMT those of the author(s) and do not necessarily reect the views of
involve the synthesis and purication steps of the furan reac- the National Science Foundation. We thank Dr Stacey I. Zones
tants, as well as any subsequent product processing. DMFDC is (Chevron Energy Technology Company) for supplying the
a more attractive substrate from the standpoint of its stability structure directing agent used in the synthesis of SSZ-33 and for
and the elimination of oxidation steps aer the DA reactions. helpful discussions, Dr Mona Shahgholi (Caltech) for the use of
The data presented here show that isolated Zn sites in the *BEA GC-MS, and Dr Joshua Pacheco and Dr Mark Deimund for
framework offer signicant improvement in both yield and helpful discussions regarding the technical aspects of Diels–
selectivity of DA cycloaddition–dehydration reactions of Alder reactions and CIT-6 synthesis, respectively.
ethylene and DMFDC over previously investigated heteroge-
neous catalysts. While both metrics remain lower for the DA
reactions of DMFDC than for MMFC, these results indicate that
such reactions are feasible, and warrant further consideration.
Notes and references
Additionally, the identied side products of DMFDC reactions,
MF, MB, CHO, furan, and benzene, have industrial applica-
tions, and do not correspond to complete loss of carbon.53,54
1 A. Corma, L. T. Nemeth, M. Renz and S. Valencia, Nature,
2001, 412, 423–425.
2 A. Corma, M. E. Domine and S. Valencia, J. Catal., 2003, 215,
294–304.
3 M. Boronat, A. Corma and M. Renz, J. Phys. Chem. B, 2006,
110, 21168–21174.
4. Conclusions
Framework Zn sites in microporous zincosilicates behave as
Lewis acid centers in probe-molecule IR spectroscopy, with
unusually high adsorption energies of Lewis bases on such
materials. The strong interactions with adsorbates severely
limit the activity of zincosilicates at low temperatures and in
solvents bearing Lewis basic groups for reactions catalyzed by
Lewis acids (e.g. inter- and intra-molecular MVPO reactions).
However, at higher temperatures, in heptane solvent, CIT-6 (Zn-
beta) is able to catalyze Diels–Alder cycloaddition–dehydration
reactions of ethylene with methyl 5-(methoxymethyl)furan-2-
carboxylate, a promising route to biomass-based terephthalic
acid. Furthermore, a CIT-6-based catalyst enables the use of the
dimethyl ester of furan-2,5-dicarboxilic acid, a furan resistant to
Diels–Alder cycloaddition–dehydration reactions catalyzed by
4 T. Blasco, M. A. Camblor, A. Corma, P. Esteve, J. M. Guil,
´
´
´
A. Martınez, J. A. Perdigon-Melon and S. Valencia, J. Phys.
Chem. B, 1998, 102, 75–88.
5 J. D. Lewis, S. van de Vyver, A. J. Crisci, W. R. Gunther,
´
V. K. Michaelis, R. G. Griffin and Y. Roman-Leshkov,
ChemSusChem, 2014, 7, 2255–2265.
6 B. Notari, in Chemistry of Microporous CrystalsProceedings of
the International Symposium on Chemistry of Microporous
Crystals, ed. S. N. Tomoyuki lnui and T. Tatsumi, Elsevier,
1991, vol. 60, pp. 343–352.
7 C. B. Khouw, C. B. Dartt, J. A. Labinger and M. E. Davis, J.
Catal., 1994, 149, 195–205.
8 C. Khouw and M. E. Davis, J. Catal., 1995, 151, 77–86.
9 C. B. Dartt and M. E. Davis, Appl. Catal., A, 1996, 143, 53–73.
known Lewis acid zeotypes (e.g., Sn-beta), resulting in direct 10 A. Corma, M. E. Domine, L. Nemeth and S. Valencia, J. Am.
formation of dimethyl terephthalate without the need for Chem. Soc., 2002, 124, 3194–3195.
further oxidation reactions. Elimination of alkali-bearing sites 11 M. Moliner, Y. Roman-Leshkov and M. E. Davis, Proc. Natl.
is shown to signicantly improve the selectivity of such reac- Acad. Sci. U. S. A., 2010, 107, 6164–6168.
tions towards dimethyl terephthalate by lowering the extent of 12 M. S. Holm, S. Saravanamurugan and E. Taarning, Science,
decarboxylation side-reactions that result in the formation of 2010, 328, 602–605.
the notable byproducts: methyl 2-furoate, methyl benzoate, and 13 J. J. Pacheco and M. E. Davis, Proc. Natl. Acad. Sci. U. S. A.,
2-cyclohexenone. Here, only zincosilicates with *BEA topology
2014, 111, 8363–8367.
have been demonstrated to be of catalytic interest, but probe- 14 M. Orazov and M. E. Davis, Proc. Natl. Acad. Sci. U. S. A., 2015,
molecule IR characterization suggests that the Lewis-acidic 112, 11777–11782.
character of isolated Zn sites in pure-silica frameworks is 15 H. J. Cho, C.-C. Chang and W. Fan, Green Chem., 2014, 16,
general, and that such materials warrant broader consideration 3428–3433.
for high temperature catalytic applications, especially those 16 R. Fernandez-Lopez, J. Kofoed, M. Machuqueiro and
´
sensitive to the presence of strong Brønsted acid sites.
T. Darbre, Eur. J. Org. Chem., 2005, 5268–5276.
17 C. Palomo, M. Oiarbide and A. Laso, Angew. Chem., Int. Ed.,
2005, 44, 3881–3884.
18 K. A. McCall, C. Huang and C. A. Fierke, J. Nutr., 2000, 130,
1437S–1446S.
Acknowledgements
This work was nancially supported as part of the Catalysis
Center for Energy Innovation, an Energy Frontier Research 19 J. Penzien, A. Abraham, J. A. van Bokhoven, A. Jentys,
¨
Center funded by the US Department of Energy, Office of
Science, Office of Basic Energy Sciences under Award DE-
T. E. Muller, C. Sievers and J. A. Lercher, J. Phys. Chem. B,
2004, 108, 4116–4126.
This journal is © The Royal Society of Chemistry 2016
Chem. Sci., 2016, 7, 2264–2274 | 2273

View Article Online
Chemical Science
Edge Article
20 Y. G. Kolyagin, I. I. Ivanova, V. V. Ordomsky, A. Gedeon and 39 R. Bermejo-Deval, R. Gounder and M. E. Davis, ACS Catal.,
Y. A. Pirogov, J. Phys. Chem. C, 2008, 112, 20065–20069. 2012, 2, 2705–2713.
21 R. Bermejo-Deval, M. Orazov, R. Gounder, S.-J. Hwang and 40 H. Y. Luo, D. F. Consoli, W. R. Gunther and Y. Roman-
M. E. Davis, ACS Catal., 2014, 4, 2288–2297. Leshkov, J. Catal., 2014, 320, 198–207.
22 N. M. Schweitzer, B. Hu, U. Das, H. Kim, L. A. Curtiss, 41 C.-C. Chang, S. K. Green, C. L. Williams, P. J. Dauenhauer
´
P. C. Stair, J. T. Miller and A. S. Hock, ACS Catal., 2014, 4,
1091–1098.
23 G. Connell and J. A. Dumesic, J. Catal., 1987, 298, 285–298.
24 A. Kozawa, J. Inorg. Nucl. Chem., 1961, 21, 315–324.
and W. Fan, Green Chem., 2014, 16, 585.
42 N. Nikbin, P. T. Do, S. Caratzoulas, R. F. Lobo,
P. J. Dauenhauer and D. G. Vlachos, J. Catal., 2013, 297,
35–43.
25 T. Takewaki, L. W. Beck and M. E. Davis, J. Phys. Chem. B, 43 J. J. Pacheco, J. A. Labinger, A. L. Sessions and M. E. Davis,
1999, 103, 2674–2679. ACS Catal., 2015, 5904–5913.
26 T. Takewaki, L. W. Beck and M. E. Davis, Top. Catal., 1999, 9, 44 O. Casanova, S. Iborra and A. Corma, ChemSusChem, 2009, 2,
35–42. 1138–1144.
27 M. A. Deimund, J. A. Labinger and M. E. Davis, ACS Catal., 45 M. L. Ribeiro and U. Schuchardt, Catal. Commun., 2003, 4,
2014, 4, 4189–4195. 83–86.
28 P. Andy and M. E. Davis, Ind. Eng. Chem. Res., 2004, 43, 2922– 46 O. Casanova, S. Iborra and A. Corma, J. Catal., 2009, 265,
2928. 109–116.
29 The British Petroleum Company P.L.C., US5208201 A, 1993, 47 J. J. Pacheco, PhD thesis: New Catalysts for the Renewable
pp. 1–8.
Production of Monomers for Bioplastics, California Institute
of Technology, 2015.
48 BP Corporation North America Inc., US 7,385,081 B1, 2008, vol.
1, pp. 1–6.
49 Furanix Technologies B.V.; the Coca-Cola Company, WO
Patent, Appl. PCT/NL2013/050740, 2013.
50 B. C. Ranu and U. Jana, J. Org. Chem., 1998, 63, 8212–8216.
30 T. Takewaki, S.-J. Hwang, H. Yamashita and M. E. Davis,
Microporous Mesoporous Mater., 1999, 32, 265–278.
31 C.-C. Chang, Z. Wang, P. Dornath, H. Je Cho and W. Fan,
RSC Adv., 2012, 2, 10475.
32 M. F. Faggin and M. A. Hines, Rev. Sci. Instrum., 2004, 75,
4547.
33 A. G. Pelmenschikov, R. A. van Santen, J. Janchen and 51 J. A. Tunge and E. C. Burger, Eur. J. Org. Chem., 2005, 2005,
E. Meijer, J. Phys. Chem., 1993, 97, 11071–11074.
34 J. C. Lavalley, Catal. Today, 1996, 27, 377–401.
1715–1726.
ˆ
52 G. J. S. Dawes, E. L. Scott, J. Le Notre, J. P. M. Sanders and
35 L. Ren, Q. Guo, P. Kumar, M. Orazov, D. Xu, S. M. Alhassan,
J. H. Bitter, Green Chem., 2015, 17, 3231–3250.
K. A. Mkhoyan, M. E. Davis and M. Tsapatsis, Angew. Chem., 53 L. Kesavan, R. Tiruvalam, M. H. Ab Rahim, M. I. bin Saiman,
Int. Ed., 2015, 54, 10848–10851.
36 C. M. Osmundsen, M. S. Holm, S. Dahl and E. Taarning,
Proc. R. Soc. London, Ser. A, 2012, 468, 2000–2016.
D. I. Enache, R. L. Jenkins, N. Dimitratos, J. a. Lopez-
Sanchez, S. H. Taylor, D. W. Knight, C. J. Kiely and
G. J. Hutchings, Science, 2011, 331, 195–199.
37 M. Boronat, P. Concepcion, A. Corma, M. T. Navarro, 54 J. E. Holladay, J. F. White, J. J. Bozell and D. Johnson, Top
M. Renz and S. Valencia, Phys. Chem. Chem. Phys., 2009,
11, 2876–2884.
38 D. P. Serrano, R. van Grieken, M. E. Davis, J. A. Melero,
a. Garcia and G. Morales, Chem.–Eur. J., 2002, 8, 5153–5160.
Value-Added Chemicals from Biomass – Volume II-Results of
Screening for Potential Candidates from Biorenery Lignin,
Richland, WA, 2007, vol. II.
2274 | Chem. Sci., 2016, 7, 2264–2274
This journal is © The Royal Society of Chemistry 2016