Periodic Trends in Highly Dispersed Groups IV and V Supported Metal Oxide Catalysts for Alkene Epoxidation with H2O2

Article abstract of DOI:10.1021/acscatal.5b01105

Supported metal oxides are important catalysts for selective oxidation processes like alkene epoxidation with H₂O₂. The reactivity of these catalysts is dependent on both identity and oxide structure. The dependence of the latter on the synthesis method can confound attempts at comparative studies across the periodic table. Here, SiO₂-supported metal oxide catalysts of Ti(IV), Zr(IV), Hf(IV), V(V), Nb(V), and Ta(V) (all of groups IV and V) were synthesized by grafting a series of related calixarene coordination complexes at surface densities less than ～0.25 nm^-2. Select catalysts were investigated by solid state NMR, UV-visible, and X-ray absorption near-edge spectroscopies. As-synthesized and calcined materials were examined for the epoxidation of cyclohexene and styrene (1.0 M) with H₂O₂ (0.10 M) at 45 and 65 °C. Nb catalysts emerge as high-performing materials, with calcined Nb-SiO₂ proceeding at a cyclohexene turnover frequency of 2.4 min^-1 (>2 times faster than Ti-SiO₂) and with ～85% selectivity toward direct (nonradical) epoxidation pathways. As-synthesized Zr, Hf, and Ta catalysts have improved direct pathway selectivities compared with their calcined versions, particularly evident for Ta-SiO₂. Finally, when the materials are synthesized from these precursors but not simple metal chlorides, the direct pathway reaction rate correlates with Pauling electronegativities of the metals, demonstrating clear periodic trends in intrinsic Lewis acid catalytic behavior.

Full text of DOI:10.1021/acscatal.5b01105

Page 1 of 35
ACS Catalysis
1
2
3
4
5
6
7
8
9
Periodic Trends in Highly Dispersed Group IV and
V Supported Metal Oxide Catalysts for Alkene
Epoxidation with H2O2
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Nicholas E. Thornburg, Anthony B. Thompson, Justin M. Notestein*
Department of Chemical and Biological Engineering, Northwestern University, 2145 Sheridan
Road, Technological Institute E136, Evanston, IL 60208, USA
KEYWORDS. supported catalyst, supported oxide, niobium, calixarene, olefin epoxidation,
hydrogen peroxide, green chemistry.
ABSTRACT. Supported metal oxides are important catalysts for selective oxidation processes,
such as alkene epoxidation with H2O2. The reactivity of these catalysts is dependent on both
identity and oxide structure. The dependence of the latter on synthesis method can confound
attempts at comparative studies across the periodic table. Here, SiO2ꢀsupported metal oxide
catalysts of Ti(IV), Zr(IV), Hf(IV), V(V), Nb(V) and Ta(V) (all of Groups IV and V) were
synthesized by grafting a series of related calixarene coordination complexes at surface densities
less than ~0.25 nmꢀ2. Select catalysts were investigated by solid state NMR, UVꢀvisible, and Xꢀ
ray absorption near edge spectroscopies. Asꢀsynthesized and calcined materials were examined
for the epoxidation of cyclohexene and styrene (1.0 M) with H2O2 (0.10 M) at 45°C and 65°C.
Nb catalysts emerge as highꢀperforming materials, with calcined NbꢀSiO2 proceeding at a
ACS Paragon Plus Environment
1

ACS Catalysis
Page 2 of 35
1
2
cyclohexene turnover frequency of 2.4 minꢀ1 (>2x faster than TiꢀSiO2) and with ~85% selectivity
towards direct (nonꢀradical) epoxidation pathways. Asꢀsynthesized Zr, Hf and Ta catalysts have
improved direct pathway selectivities compared to their calcined versions, particularly evident
for TaꢀSiO2. Finally, when the materials are synthesized from these precursors, but not simple
metal chlorides, the direct pathway reaction rate correlates with Pauling electronegativities of the
metals, demonstrating clear periodic trends in intrinsic Lewis acid catalytic behavior.
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
INTRODUCTION
Supported metal oxide catalysts are ubiquitous in both commodity and fine chemical
manufacturing, particularly for selective oxidations of alkanes, alkenes, alcohols, and aromatics,
as well as oxidative dehydrogenation, hydroxylation, and ammoxidation.1ꢀ3 Separately, alkene
epoxidation is widely studied for organic synthesis, and there is a desire to use supported metal
oxide catalysts with H2O2 instead of alkyl hydroperoxides. H2O2 is environmentally
advantageous, and its process economics are not dependent on selling or disposing of a coꢀ
product.4ꢀ5
Ti(IV) oxide on SiO2 (TiꢀSiO2) is a benchmark Lewis acid catalyst for epoxidation with
organic hydroperoxides, while Ti embedded in a silicate framework (e.g. TSꢀ1) is preferred for
H2O2.2, 6ꢀ11 It is accepted that tetrahedrallyꢀcoordinated Ti sites are the most reactive and
selective for epoxidation with hydroperoxides,6ꢀ7, 12ꢀ14 so synthesis methods are designed to give
the high dispersions on SiO2 that result in a preponderance of such sites.14ꢀ17 For example,
grafting bulky molecular precursors leads to more active and selective catalysts on a perꢀTi
basis.9ꢀ10, 14ꢀ15, 17
Aside from TiꢀSiO2, Ta(V)18ꢀ21 and Nb(V)22ꢀ27 oxides on SiO2 (TaꢀSiO2 and NbꢀSiO2) have
demonstrated high selectivity, activity and stability for epoxidation with aqueous H2O2 oxidant,
ACS Paragon Plus Environment
2

Page 3 of 35
ACS Catalysis
1
2
3
4
5
6
7
8
9
but in fewer reports. The extra anionic coordination site of Ta oxides has been suggested to be
responsible for their higher water tolerance and faster turnover in the presence of water relative
to TiꢀSiO2.19ꢀ21 Vanadium(V) oxides on SiO2 (VꢀSiO2) are wellꢀstudied for oxidation,28ꢀ29 but are
poorly selective epoxidation catalysts,23 believed to operate via oxovanadium intermediates
rather than through peroxymetal species like TiꢀSiO2, TaꢀSiO2 and NbꢀSiO2.4, 30ꢀ31 Grafted
Zr(IV) catalysts (ZrꢀSiO2) have not been reported for epoxidation, to our knowledge, but inꢀ
framework Zrꢀbeta materials32ꢀ33 and Zrꢀcontaining polyoxometalates34 show poor activity and
selectivity for epoxidation compared to analogous Ti materials. Lastly, to our knowledge,
supported Hf(IV) catalysts (HfꢀSiO2) have not been reported for epoxidation, though
homogeneous studies show similar selectivities to Ti analogues.35ꢀ37
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Beyond individual catalyst studies, there are few headꢀtoꢀhead comparisons of different
supported metal oxides for epoxidation. In the 70s, Sheldon benchmarked several homogeneous
and SiO2ꢀsupported catalysts for olefin epoxidation with alkyl hydroperoxides and found high
activity and selectivity with Ti(IV) and Mo(VI), the latter of which poorly performed when
supported due to significant leaching.4, 30 Because supported metal oxide structure (e.g. extent of
MꢀOꢀM vs. MꢀOꢀSi bonding) plays a strong role in its epoxidation reactivity,8, 10ꢀ11, 17, 26, 38ꢀ39
crossꢀcomparisons among related metal oxides may become confounded if differences in
synthesis method or precursor lead to a variation in surface oxide species. Therefore, here we
graft wellꢀdefined calixarene coordination complexes of Groups IV and V onto SiO2 to create a
structurallyꢀuniform set of catalysts for comparison in epoxidation with H2O2. The calixarene
ligand is bulky and strongly chelated to the metal, which some of us have shown to give
40ꢀ42
21
undercoordinated, highly active TiꢀSiO2
and TaꢀSiO2 epoxidation catalysts. Here, we
ACS Paragon Plus Environment
3

ACS Catalysis
Page 4 of 35
1
2
3
4
5
6
broaden this strategy to the rest of Groups IV and V to draw correlations for Lewis acidꢀ
catalyzed epoxidation over a large family of supported oxides.
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
4

Page 5 of 35
ACS Catalysis
1
2
3
4
5
6
EXPERIMENTAL METHODS
Catalyst Synthesis
7
8
9
All catalyst syntheses were performed using standard Schlenk line techniques under N2 or in a
controlled Ar atmosphere glove box. Solvents (toluene, tetrahydrofuran, benzene, nꢀhexane)
were distilled and degasified by standard methods and stored inside the glovebox. TiCl4(THF)2,
ZrCl4(THF)2, HfCl4, VCl3(THF)3, NbCl5, and TaCl5 were used as received from Strem
Chemicals and stored inside the glove box. pꢀtertꢀButylcalix[4]arene (95%) (“Calix”) was used
as received from Sigma Aldrich. 1,3ꢀDimethoxyꢀtertꢀbutylcalix[4]arene (“dmCalix”) was
prepared according to a published procedure.43 Calixarene coordination complexes were
prepared by reacting stoichiometric metal chloride and calixarene ligand, generally following
known procedures described below. Synthesis was followed by isolation or direct grafting onto
SiO2 (Selecto Scientific, 32ꢀ62 µm particle size, 570 m2 gꢀ1, 5.4 nm avg. pore dia.) that had been
partially dehydroxylated at 300°C under dynamic vacuum (<50 mtorr) for 10 h (Scheme 1).
Following grafting, all dry solids were seived to ≤88 µm particle size. See Supporting
Information for full syntheses, NMR resonances, and elemental analyses.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For complexes of Ti, Nb, and Ta, dmCalix (Ti) or Calix (Nb, Ta) and stoichiometric metal
chloride were refluxed in toluene with a N2 sparge until HCl evolution ceased, giving the
monomethoxyꢀcalixarene complex (mmCalixꢀTiCl)44 or the associated dimers of the Calixꢀ
NbCl45 and CalixꢀTaCl46 complexes. Complexes were isolated by solvent removal via rotary
evaporation and trituration with nꢀhexane. The calixarene complexes were grafted to SiO2 from
refluxing toluene. After reaction, the solids were washed in toluene, Soxhlet extracted in toluene,
and dried under dynamic vacuum. Some of us have previously reported SiO2 grafting of the Ti40
and Ta21 complexes to give bright orange solids.
ACS Paragon Plus Environment
5

ACS Catalysis
Page 6 of 35
1
2
3
4
5
6
7
8
9
, O2/N2
550oC
toluene
reflux
SiO2
OH
O
O
O
O
(a)
OH OH
O
O
O
O
O
O
Ti
Cl
reflux
+
Ti
Ti
O
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
TiCl4(THF)2
OH
OH
O
O
O
OH
OH
SiO2
SiO2
(1) n-BuLi, THF
SiO2
, O2/N2
550oC
OH
O
O
O
O
O
O
O
O
(b)
(2) MIVCl4,
toluene,
reflux
O
O
OH OH
MIV
O
MIV
MIV
benzene, reflux
Cl
Cl
O
O
OH
OH
O
O
OH
OH
(MIV = Zr,Hf)
SiO2
SiO2
(1) n-BuLi, THF
SiO2
, O2/N2
550oC
O
O
O
O
O
O
O
O
O
(c)
(2) VCl3(THF)3,
toluene,
RT
O
O
OH OH
RT
V
V
O
V
O
O
OH
OH
O
O
OH
OH
Cl
SiO2
SiO2
O
O
O
O
O
Cl
O
MV
O
O
SiO2
, O2/N2
550oC
toluene
MV
O
HO
OH
OH OH
+
HO
OH
O
(d)
Cl
reflux
reflux
O
MV
O
MV
MVCl5
O
O
OH
O
O
O
OH
OH
(MV = Nb,Ta)
SiO2
SiO2
Scheme 1. Synthesis of Group IV and V calixarene complexes, their grafting onto SiO2, and
subsequent calcination to yield highly dispersed metal oxides: (a) for Ti, (b) for Zr and Hf, (c)
for V, and (d) for Nb and Ta. Full details of synthetic procedures are included in Supporting
Information.
For complexes of Zr47, Hf, and V48ꢀ49, dmCalix was suspended in THF, reacted with 2 eq. nꢀ
butyllithium (1.6 M in hexanes, Sigma), and added dropwise to metal chloride solutions in
benzene (Zr, Hf) or THF (V). The solvent was removed by rotary evaporation, and residues were
ACS Paragon Plus Environment
6

Page 7 of 35
ACS Catalysis
1
2
3
4
5
6
7
8
9
extracted into 5:1 anhydrous nꢀhexane:benzene. The isolated calixarene complexes were
redissolved in toluene inside a glovebox and subsequently grafted to SiO2. Following grafting,
solids were washed with toluene, THF (Zr, Hf) or nꢀhexane (V), and water before drying under
dynamic vacuum.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Asꢀsynthesized CalixꢀMꢀSiO2 (M = Ti, Zr, Hf, V, Nb, Ta) catalysts were heated in air at a
ramp of 10°C minꢀ1 and held at 550°C for 6 h to remove organic content and give MꢀSiO2. A
control Nbꢀcontaining catalyst was synthesized by refluxing NbCl5 and SiO2 in toluene for 24 h
with continuous sparging of N2. The final suspension was vacuum filtered, washed with toluene,
dried under dynamic vacuum and calcined as above to give NbCl5ꢀSiO2.
Characterization
Metal content was quantified using inductivelyꢀcoupled plasma atomic emission spectroscopy
(ICPꢀAES; Varian Vista MPX) calibrated against serial dilution of commercial standard
solutions (Fluka Analytical) in 0.9 wt% HNO3. Catalysts were digested in 48 wt% HF and
diluted with 0.9 wt% HNO3. [CAUTION: handle and store HF with extreme care.] Multiple
trials were averaged. C and H contents of calixarene complexes were determined by combustion
analysis (Galbraith Laboratories, Knoxville, TN).
Thermogravimetric analysis (TGA; TA Instruments Q500) of asꢀsynthesized catalysts was
carried out under dry synthetic air (90% O2, 10% N2, 100 mL minꢀ1) at a ramp of 10°C minꢀ1
from room temperature to 800°C and compared to mass loss by bare SiO2 to correct for support
hydroxyl condenstation. Calixarene content was estimated by assuming combustion of ligands
with molecular weight 629 g molꢀ1 (mmCalixꢀTi), 643 g molꢀ1 (dmCalixꢀZr, ꢀHf, and ꢀV) or 613
g molꢀ1 (CalixꢀNb and ꢀTa). Surface densities are reported relative to the N2 physisorption
surface area (Micromeritics ASAP 2010 CE system) of the unmodified, dehydroxylated support.
ACS Paragon Plus Environment
7

ACS Catalysis
Page 8 of 35
1
2
3
4
5
6
7
8
9
UVꢀvisible spectra of solids were measured from 700ꢀ200 nm at ambient conditions using a
Shimadzu UVꢀ3600 and a Harrick Praying Mantis diffuse reflectance accessory (DRUVꢀvis).
Polytetrafluoroethylene powder was used to collect a ‘perfect’ reflectance baseline for Kubelkaꢀ
Munk pseudoabsorbances and as a 20:1 diluent for catalyst samples. Optical edge energies were
computed from the xꢀintercept of the indirect Tauc plot of [F(R)•hν]1/2 vs. hν (eV), where F(R) is
the KubelkaꢀMunk pseudoabsorbance. UVꢀvisible spectra of some calixarene complexes were
collected in acetonitrile solution and a 1ꢀcm pathlength quartz cell over the same wavelength
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1
range (Figure S1). Solutionꢀphase HꢀNMR and 13CꢀNMR spectra of these complexes were
obtained on a Bruker AVANCE III 500 MHz spectrometer with direct cryoprobe. Solid state 13C
crossꢀpolarization magic angle spinning (CPꢀMAS) NMR spectra were collected on a 400 MHz
Varian spectrometer at a spin rate of 5 kHz.
Xꢀray absorption nearꢀedge spectroscopy (XANES) was performed at the Ti Kꢀedge (4,966.4
eV), the V Kꢀedge (5,465.1 eV), the Nb Kꢀedge (18,985.6 eV), and the Ta L3ꢀedge (9,881.1 eV)
using bending magnet station D at the Advanced Photon Source beamline at Sector 5 (DuPontꢀ
NorthwesternꢀDow Collaborative Access Team (DNDꢀCAT)) of Argonne National Laboratory.
The beam energy was controlled using a Si(111) monochromator with 10ꢀ4 eV resolution. Beam
energies and detectors were calibrated to the above edges using pure metal foil in transmission
mode. Standards were brushed onto Kapton tape, and spectra were collected at ambient
conditions in transmission mode. V2O5, Nb2O5, and Ta2O5 standards were used as received from
Strem Chemicals. Anatase TiO2 was uses as received from Sigma Aldrich. KNbO3 was used as
received from Alfa Aesar. Ba2TiO4 was prepared according to a literature procedure.50 YNbO4
and CaNb2O6 and were prepared according to slightly modified literature procedures (see
Supporting Information).51ꢀ52 Powder Xꢀray diffraction patterns of YNbO4 and CaNb2O6 were
ACS Paragon Plus Environment
8

Page 9 of 35
ACS Catalysis
1
2
3
4
5
6
7
8
9
obtained using a Rigaku Ultima IV diffractometer with Cu Kα radiation, a step width of 0.1° and
a receiving slit size of 0.6 nm, and verified using the patternꢀmatching features of the MDI Jade
2010 analysis software (Figure S2). Fresnoite (Ba2(TiO)Si2O7) was provided by the E. Bruce
Watson Research Group at the Rensselaer Polytechnic Institute. Catalyst samples were analyzed
in fluorescence mode by using a fourꢀchannel Canberra SII Vortex ME4 detector, with a
controlled atmosphere sample cell oriented at 45°±5° to the detector and to the incident xꢀray
beam. Catalysts were loaded into a sealed controlled atmosphere cell, evacuated to ꢀ20 inHg
gauge, heated under dynamic vacuum at 150°C for 2 h, and sealed under purified Ar prior to
spectroscopy.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Catalysis
Cyclohexene (>99%, Sigma) and styrene (>99%, Sigma) were passed over a column of
activated Alumina N (Selecto Scientific, 63ꢀ150 µm) immediately before use. A 4.0 M solution
of H2O2 in acetonitrile was prepared by diluting 10 mL aqueous H2O2 (30% w/w) in 20 mL
acetonitrile, followed by drying with 7.0 g anhydrous MgSO4 (>99%, Sigma) and decanting.
H2O2 concentration was quantified using permanganimetric titration.
Epoxidation was performed in 20 mL septumꢀcap glass vials vented with a needle. In a typical
experiment, 10.9 mmol alkene, 0.01 mmol metal (60ꢀ90 mg of solids) catalyst and 1.30 mmol
1,2ꢀdichlorobenzene (98%, Alfa Aesar) internal standard were added to 9 mL acetonitrile (>99%,
spectrophotometric grade, Alfa Aesar), heated to 35ꢀ65°C and shaken at 700 r.p.m. on a Glasꢀ
Col heated vortexer. For reactions at 75°C, a 25 mL roundꢀbottom flask equipped with a reflux
condenser and a stirbar was used. The reaction was initiated with the instantaneous addition of
1.08 mmol H2O2. This gives initial conditions of [cyclohexene or styrene]0 = 1.02 M, [H2O2]0 =
0.103 M, alkene:H2O2:M = 1000:100:1, and a total reaction volume of 10.4ꢀ10.6 mL. Reaction
ACS Paragon Plus Environment
9

ACS Catalysis
Page 10 of 35
1
2
3
4
5
6
7
8
9
aliquots were passed through a Whatman 0.7 µm glass microfiber filter, and contacted with Ag
powder (>99%, <250 µm, Sigma) to quench unreacted H2O2. Products from cyclohexene
oxidation were identified by GCꢀMS (Shimadzu QP2010 GCꢀMS, Zebron ZBꢀ624 capillary
column) and quantified using GCꢀFID (Shimadzu 2010 GCꢀFID, TRꢀ1 capillary column) against
calibrated standards. Products from styrene oxidation were identified and quantified by GCꢀMS.
In all kinetic experiments, mass balances were ≥85%.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Catalyst leaching was tested by running cyclohexene oxidation for 10ꢀ15 min at 65°C, after
which the shaking was stopped, the solids allowed to settle, and the liquid contents were quickly
transferred using a glass pasteur pipette (no filter) to a clean, heated 20 mL septumꢀcap vial,
where the reaction continued as usual. Some samples were transferred to vials also containing
100 mg dehydroxylated SiO2 and an additional 0.54 mmol H2O2 (half the amount added at t=0).
Some catalyst used in cyclohexene epoxidation at 65°C were vacuumꢀfiltered, washed with 50
mL acetonitrile, dried under dynamic vacuum, and reused. After washing, some used catalysts
were also recalcined at 550°C in air for 6 h before reuse.
RESULTS
Materials Synthesis and Characterization
Calix[4]arene complexes of Ti(IV), Zr(IV), Hf(IV), V(III), Nb(V) and Ta(V) were synthesized in
accordance with literature, or with analogy to known complexes in the case of Hf. Elemental
analyses (see Supporting Information) indicate that the Ti, Nb, and Ta complexes strongly retain
approximately 1 molecule of toluene solvent, with no evidence of uncomplexed metal chloride.
The complexes of Zr, Hf, and V contain ~10% uncomplexed metal chloride and retain the LiCl
coꢀproduct. However, the final water washing of the grafted catalyst removes residual LiCl to
<0.006 mmol gꢀ1 (Li:M <3%) as determined by ICPꢀAES. Metal content in the calcined catalysts
ACS Paragon Plus Environment
10

Page 11 of 35
ACS Catalysis
1
2
3
4
5
6
and ligand content in the asꢀsynthesized catalysts were determined using ICPꢀAES and TGA,
respectively, and are reported in Table 1.
7
8
9
Table 1. Summary of grafted catalysts.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Metal Contentb
Catalyst
Precursor
mmCalixꢀTi
dmCalixꢀZr
dmCalixꢀHf
dmCalixꢀV
CalixꢀNb
[Calix]a
(mmol g-1)
0.18
(mmol g-1) (M nm-2)c wt % M:Calix
0.18
0.19
0.19
0.25
0.18
0.19
0.19
0.19
0.21
0.20
0.26
0.19
0.19
0.21
0.9
1.8
3.5
1.3
1.7
1.8
3.6
1.0
1.8
1.8
2.3
1.7
ꢀꢀ
0.11
0.11
0.11
0.11
ꢀꢀ
0.18
NbCl5
CalixꢀTa
1.1
a
b
c
From TGA combustion analysis of asꢀsynthesized catalysts.
From ICPꢀAES of calcined catalysts.
Using the SiO2 support BET specific surface area of 570 m2 gꢀ1.
Catalyst syntheses used 0.3ꢀ0.5 complexes per nm2 of added SiO2, but the final metal surface
densities are between 0.19 and 0.26 M nmꢀ2, consistent with a previouslyꢀobserved geometric
maximum for grafting of these bulky complexes,40 and suggesting that the metalꢀcalixarene
complex is intact during synthesis. The control NbꢀSiO2 catalyst from NbCl5 was synthesized to
match these surface densities; it grafts quantitatively under these conditions. For Ti and Ta,
metal:calixarene ratios in the asꢀsynthesized catalysts are ~1, indicating that the complex is
stable on the surface. Ratios are higher for other metals, indicating that the ligand can be lost
during postꢀsynthesis washing steps. In particular, the Zr, Hf, and Vꢀcontaining materials were
extensively washed with water to remove residual LiCl, which may contribute to ligand loss. Xꢀ
ray absorption nearꢀedge spectroscopy (XANES) of the Ti9, 14, 53, V54ꢀ55, Nb39, 54ꢀ58, and Ta59ꢀ60
materials gave edge energies within 1 eV of standards for each metal’s highest oxidation state, as
expected (Table S1). This does confirm that the V3+ precursor oxidized to V5+ after grafting, as
ACS Paragon Plus Environment
11

ACS Catalysis
Page 12 of 35
1
2
3
4
5
6
intended. Additional discussion of the surface speciation of Ti and Nb catalysts is found in the
Oxide Structure and Stability section.
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1
Figure 1. HꢀNMR spectra of calixarene complexes of Groups IV and V and of their starting
calixarenes. * = NMR solvent († = CDCl3, § = C6D6, ‡ = CD2Cl2); TMS = tetramethylsilane
(internal standard); + = residual nꢀhexane (from synthesis); # = vacuum grease.
1
Solution HꢀNMR spectra of the calixareneꢀmetal complexes (Figure 1) do not show the OH
resonances associated with the free ligands, and the diagnostic bridging methylene resonances
are all shifted from that of the free ligand and are appropriate for the expected symmetry of the
complexes.61ꢀ63 Comparing the solidꢀstate 13C CPꢀMAS NMR spectra of the asꢀsynthesized
catalysts with the corresponding solution spectra of the precursors (Figure S3) shows diagnostic
ACS Paragon Plus Environment
12

Page 13 of 35
ACS Catalysis
1
2
downfield shifts in the ipso carbons,21 indicating calixareneꢀmetal connectivity even when
supported.
3
4
5
6
7
8
9
Peaks and optical edge energies from DRUVꢀvis spectroscopy provide insight into the metal
oxide surface speciation.8, 14, 64ꢀ67 Figure 2 shows the calixarene phenolate ligandꢀtoꢀmetal charge
transfer band (LMCT) from 560 to 320 nm, aromatic πꢀπ* transitions near 285 nm, and LMCT
from support oxygens, typically near 250 nm. Normalized to the πꢀπ* transition, the calixarene
phenolate LMCT intensity varies dramatically from metal to metal, in the general order Nb > Ti
~ V > Ta > Zr > Hf. The Ta LMCT also appears to be shifted to higher energies than for the
other materials.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 2. Diffuse reflectance UVꢀvisible spectra for (a) asꢀsynthesized materials, normalized to
the πꢀπ* transitions, and (b) calcined materials, normalized to the sole intense feature.
After calcination (Figure 2b), the materials absorb only in the UV, evidence of highly
dispersed metal oxide groups. Optical edge energies from the indirect Tauc plot14, 64ꢀ65, 68 are
ACS Paragon Plus Environment
13

ACS Catalysis
Page 14 of 35
1
2
3
4
5
6
7
8
9
compared to literature values in Table 2. These high edge energies are consistent with prior
values reported in literature for grafted calixarenes and other precursors. Though edge energies
have not been previously reported for analogous ZrꢀSiO2 and HfꢀSiO2, optical edges and peak
maxima estimated from published DRUVꢀvis of Zrꢀbeta and Hfꢀbeta69 are consistent with those
of our materials. Though the 3.1 eV edge of VꢀSiO2 is lower than the other supported oxides, the
value is consistent with energies reported by others.49, 70 The TaꢀSiO2 spectrum features a
shoulder with an edge of 3.4 eV, suggesting some aggregation for this material. Crucially, the
control NbꢀSiO2 from NbCl5 has an edge energy 0.4 eV lower than NbꢀSiO2 from grafted
calixareneꢀNb, indicating that the grafted calixareneꢀNb leads to more highly dispersed NbOx
structures than grafted NbCl5 at equivalent metal loadings.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Table 2. Optical edge energies of calcined materials.
Edge
(eV)
3.8
3.7
4.2
3.1
4.0
3.6
Literature Max
edge (eV) (nm)
Catalyst
TiꢀSiO2
ZrꢀSiO2
HfꢀSiO2
VꢀSiO2
NbꢀSiO2
NbCl5ꢀSiO2
TaꢀSiO2
4.171
3.969
4.169
3.270
4.172
4.172
4.821
244
239
230
298
240
252
222
3.4*; 4.5
* shoulder feature.
Cyclohexene Epoxidation
Following reaction networks proposed for TiOxꢀSiO2 systems,73ꢀ74 cyclohexene epoxidation
may occur via nonꢀradical heterolytic activation of H2O2 (“direct”) and along radical pathways
via cyclohexenyl hydroperoxides in parallel (Scheme 2). Other reactions include epoxide
hydrolysis to the transꢀdiol, catalyzed by both Lewis and Brønsted acids, and H2O2
decomposition that accounts for the incomplete cyclohexene conversion over most materials.
ACS Paragon Plus Environment
14

Page 15 of 35
ACS Catalysis
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Scheme 2. Proposed reaction network for cyclohexene oxidation.21, 73ꢀ74
Accordingly, we use reaction pathway branching selectivity, R2/(R1+R2) in Scheme 2, as a key
metric for assessing catalyst performance. The directlyꢀproduced epoxide is the sum of epoxide +
diol – (cyclohexenol – cyclohexenone), where the latter difference accounts specifically for the
cyclohexenol produced during epoxidation by cyclohexenyl hydroperoxide.21, 73ꢀ74 The direct
pathway selectivity (concisely referred to as pathway selectivity) is then the fraction of direct
epoxide relative to all C6 products. Grafted calixareneꢀTa catalysts were used to find optimal
temperatures for further study (Table 3), following prior experience with this system.21 Both
pathway and product selectivities are integrated over the entire product yield to that point, and
for most systems studied here, these selectivities change significantly as the reaction proceeds.
ACS Paragon Plus Environment
15

ACS Catalysis
Page 16 of 35
1
2
Table 3. Selected data for cyclohexene epoxidation by CalixꢀTaꢀSiO2 and TaꢀSiO2.a
3
4
5
6
7
8
9
Initial TOF
6-h
yield
(%)
58
63
62
63
53
64
69
Pathway
Sel. at ~25%
yield (%)
6-h selectivity (%)
Epox. Diol Otherb
Temp.
(°C)
(mol molTa-1 min-1)
Catalyst
Total
0.79
0.99
1.5
Direct
0.77
0.96
1.5
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
35
45
55
65
75
35
45
55
65
75
73
73
74
70
72
47
49
48
56
54
26
26
24
27
26
26
25
27
24
17
1
1
2
3
2
27
26
25
21
29
97
98
97
97
97
40
27
21
35
27
Calixꢀ
TaꢀSiO2
2.0
2.0
2.0
2.0
0.60
0.68
2.2
1.8
1.9
0.32
0.32
0.43
0.67
0.42
TaꢀSiO2
87
80
65
a
See standard conditions in text. Yields based on all C6 products detected, relative to limiting
H2O2. Turnover frequencies (TOF) relative to total metal content. Initial rates through 15 min.
b
Cyclohexenol and cyclohexenone. No cyclohexanone or cyclohexanol observed.
Initial rates of epoxidation increase with temperature while product and pathway selectivities
at any level of conversion remain largely unchanged with temperature. Generally, CalixꢀTaꢀSiO2
is a much more selective and active catalyst than calcined TaꢀSiO2, consistent with previous
reports by some of us.21, 75 For bare TaꢀSiO2, the highest pathway selectivity was observed at
35°C, while at 75°C, yields began to suffer, likely due to catalytic decomposition of H2O2.
Reaction at 45°C and 65°C offers a balance of high selectivities and fast initial rates, and these
temperatures were chosen for kinetic experiments across the periodic series of catalysts. Because
initial TOFs across this series span an order of magnitude, here we also report selectivities at
~25% yield of C6 products as well as the 6ꢀh yield of all C6 products. Figure 3 gives results for
the asꢀsynthesized and calcined materials. For summary of full results, see Table S2. Selected
timeꢀcourse data are given in Figure 4. For calcined catalysts, the initial reactivity trends Nb > Ta
>> Ti > V > Hf ~ Zr; the pathway selectivity at 25% yield trends Nb >> Ta ~ Ti ~ Hf ~ Zr >> V;
and the total 6ꢀh yield trends Nb ~ Ti > Ta >> V > Zr ~ Hf.
ACS Paragon Plus Environment
16

Page 17 of 35
ACS Catalysis
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 3. Selected kinetic data for the epoxidation of cyclohexene at 65°C by Group IV and V
calixareneꢀMꢀSiO2 and calcined MꢀSiO2. See standard reaction conditions in text. Yields based
on all C6 products detected, relative to limiting H2O2. Turnover frequencies (TOF) relative to
total metal content. Initial rates through 15 min.
Figure 4. Product yields vs. time for cyclohexene oxidation at 65°C over calcined (a) TiꢀSiO2,
(b) NbꢀSiO2, and (c) the control NbCl5ꢀSiO2. Note that product yields have the same values when
expressed as concentration (mM), % yield, or turnover numbers.
ACS Paragon Plus Environment
17

ACS Catalysis
Page 18 of 35
1
2
3
4
5
6
7
8
9
The most striking result is that NbꢀSiO2 derived from the calixarene precursor significantly
outperforms all other catalysts, including the common TiꢀSiO2, with respect to rates and pathway
selectivity. However, this selectivity and rate advantage disappear for the control Nb catalyst
prepared from NbCl5. ZrꢀSiO2 and HfꢀSiO2 are signficantly less productive than the other
materials and are not significantly different from one another in this and all other metrics. Taꢀ
SiO2 gives a high total initial rate, but it initially produces significant amounts of radical
products, contributing to a low direct rate. Taꢀ, Tiꢀ, Zrꢀ and HfꢀSiO2 have similar pathway
selectivities by the time they reach 25% yield of C6 products. Finally, VꢀSiO2 is a poor
epoxidation catalyst that proceeds almost exclusively via radical oxidation pathways to give
cyclohexenol.76 For the asꢀsynthesized, calixareneꢀcontaining catalysts, the initial reactivity trend
is Nb ~ Ta >> V > Ti > Hf ~ Zr; the trend in pathway selectivity at ~25% yield is Nb ~ Ta > Zr ~
Hf > Ti > V; and the total yield at 6 h trends Nb > Ta > V > Ti > Zr ~ Hf. Retaining the
calixarene ligand significantly improves the pathway selectivity of the 2nd and 3rd row cations in
both groups compared to their calcined versions.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
All of the highꢀperforming catalysts were tested for potential leaching of active species into
solution.77 For Nbꢀ, Taꢀ, and TiꢀSiO2, oxidation rates by any leached species were more than an
order of magnitude slower than those of the solid (full results in Figures S4ꢀS5 and Table S3),
consistent with the observation of negligible leaching by others.78 Although not active in
solution, leached metals cannot be ruled out completely, as adding fresh SiO2 and extra H2O2 to
the decanted liquid does improve rates relative to the clear solution, suggesting reꢀ
immobilization to synthesize new, active species. VꢀSiO2 leaches significantly, and
homogeneous catalysis appears to contribute substantially to the overall observed reactivity
(Figure S5, Table S3), as previously observed by others.79ꢀ81
ACS Paragon Plus Environment
18

Page 19 of 35
ACS Catalysis
1
2
3
4
Styrene Epoxidation
5
6
7
8
Styrene’s epoxidation reaction network includes overoxidation and isomerization products
phenylacetaldehyde, styrene diol, and benzaldehyde (Scheme 3), and the latter is believed to
form via oxidative decarboxylation of a transient αꢀhydroxycarboxylic acid.82 Epoxidation of the
electronꢀpoor styrene is a challenging reaction for d0 metal oxide catalysts, and yields and
epoxide selectivities are correspondingly poorer than for cyclohexene.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
O
O
[O]
OH
O
OH
H
Scheme 3. Reaction products in the catalytic oxidation of styrene.
Figure 5. Selected kinetic data for the epoxidation of styrene at 65°C by Group IV and V
calixareneꢀMꢀSiO2 and calcined MꢀSiO2. See standard reaction conditions in text. Yields based
on detected epoxide, diol, phenylacetaldehyde and benzaldehyde relative to limiting H2O2.
Turnover frequencies (TOF) relative to total metal content. Initial rates through 15 min.
ACS Paragon Plus Environment
19

ACS Catalysis
Page 20 of 35
1
2
3
4
5
6
7
8
9
For reaction at 65°C, Figure 5 gives initial rates, epoxide selectivites at ~10% yield, and 6ꢀh
selectivites and total yields. (Full data in Table S4.) After 6 h, all catalysts give <60% epoxide
selectivity due to ring opening, isomerization, and other oxidations. For calcined materials, Nbꢀ
SiO2 is again initially the most reactive and selective to epoxide, while HfꢀSiO2 and ZrꢀSiO2
perform equally poorly. VꢀSiO2 is nearly as reactive for styrene as it is for cyclohexene, giving it
the highest total yields over 6 h, but again displaying low selectivity to the epoxide. Selectivites
change little with reaction progress, indicating that the other products may be predominantly
formed in parallel, rather than in sequential pathways. Retaining the calixarene ligand in the asꢀ
synthesized catalysts markedly improves the rate for VꢀSiO2 and the initial selectivties of Nbꢀ
SiO2 and TaꢀSiO2. The selectivities of these latter two catalysts decrease significantly with
reaction progress, consistent with the formation of the other oxygenates as secondary products.
This indicates that the ligand specifically enhances the rate to the epoxide.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Oxide Structure and Stability
XANES was used to further understand surface speciation and local coordination geometry for
Ti9, 14, 53 and Nb39 catalysts before and after use. These metal oxides display distinct preꢀedge
features from 1s to 3d transitions that allow for fingerprinting of metal–oxygen coordination
environment by comparison with oxide standards of known coordination number.83 The intensity
and position of preꢀedge peaks from Tiꢀcontaining materials were correlated to average Ti–O
coordination numbers using Ba2TiO4 (4ꢀcoordinate), Ba2(TiO)Si2O7 Fresnoite (5ꢀcoordinate),
and TiO2 anatase (6ꢀcoordinate) standards (Figure S6).42, 84ꢀ86 Similarly, the preꢀedge intensity in
Nbꢀcontaining materials was correlated to average NbꢀO coordination numbers using YNbO4 (4ꢀ
coordinate)51, CaNb2O6 (5ꢀcoordinate)52, 87 and KNbO3 (6ꢀcoordinate) standards (Figures 6, S7).
Nb2O5 is also octahedral, but exhibits firstꢀcoordination disorder, giving it a lower preꢀedge
ACS Paragon Plus Environment
20

Page 21 of 35
ACS Catalysis
1
2
3
4
5
6
7
8
9
feature height. For discussion purposes, the experimental catalysts are assigned a coordination
number based on correlations to these materials as follows (Table 4). For the Nbꢀcontaining
catalysts, preꢀedge peak heights of 0.19ꢀ0.15 indicate coordination numbers between 4 and 5, or
disordered 4ꢀcoordinate structures. Values of 0.15ꢀ0.12 indicate coordination numbers between 5
and 6, or disordered 5ꢀcoordinate structures. Values below 0.12 indicate disordered 6ꢀcoordinate
structures.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 6. Nb Kꢀedge XANES and fitting of preꢀedge feature for (a) 4ꢀcoordinate YNbO4, (b) 5ꢀ
coordinate CaNb2O6, (c) 6ꢀcoordinate KNbO3, (d) NbꢀSiO2, (e) NbCl5ꢀSiO2. Fitting details in
Experimental Methods and SI. (f) Linear correlation between preꢀedge feature intensity and
coordination number of standards, then used to interpolate values for the experimental materials
NbꢀSiO2, CalixꢀNbꢀSiO2, and NbCl5ꢀSiO2.
ACS Paragon Plus Environment
21

ACS Catalysis
Page 22 of 35
1
2
3
4
5
6
7
Table 4. XANES summary for Ti and Nb catalysts.
8
9
Pre-edgea
Condition Energy (eV) Sharpnessb C.N.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Catalyst
fresh
spentc
fresh
4,970.6
4,970.9
4,970.0
4,970.1
18,993
18,994
18,993
18,994
18,994
18,993
0.77
0.66
0.88
0.89
0.09
0.12
0.17
0.12
0.17
0.13
4.8
5.4
4.4
4.4
6.0e
5.9
4.5
5.9
4.6
5.6
mmCalixꢀTiꢀSiO2
TiꢀSiO2
spentc
fresh
CalixꢀNbꢀSiO2
spentc
fresh
NbꢀSiO2
spentc
recalcinedd
fresh
NbCl5ꢀSiO2
a
See Figures S6 (Ti) and 6 and S7 (Nb) for preꢀedge fitting and correlations between fitted
features and coordination number (C.N.)
b For Ti: sharpness parameter (A2+A3)/ATOT – see Figure S6; for Nb: peak intensity.
c
Following 6 h cyclohexene epoxidation at 65°C, standard conditions. Solids were vacuum
filtered, washed with excess acetonitrile and dried at 150°C under dynamic vacuum for 2 h.
d Treated as in (c) with subsequent calcination in air at 550°C for 6 h.
e Substantially disordered.
Three conclusions are made by the fits to the fresh catalysts. First, the calixarene ligand is
clearly attached to the metal sites, as it increases coordination number (C.N.) and/or disorder for
both Nb and Ti, relative to the calcined catalysts. Second, the NbꢀSiO2 derived from calixarene is
much more undercoordinated than the analogous control catalyst derived from NbCl5, consistent
with its lower optical edge and catalyst proficency. Finally, the calcined TiꢀSiO2 and NbꢀSiO2
catalysts have nearly identical effective C.N. Overall, these results demonstrate that the ligand
enforces a similar – and similarly undercoordinated – environment at the active metal.
Following use in epoxidation, the C.N. of NbꢀSiO2 increases from 4.5 to 5.9, while TiꢀSiO2
maintains its mostly tetrahedral coordination environment. However, recalcination of spent Nbꢀ
SiO2 recovers the preꢀedge feature of the fresh material, indicating that the catalyst strongly
ACS Paragon Plus Environment
22

Page 23 of 35
ACS Catalysis
1
2
3
4
5
6
7
8
9
adsorbs species from the reaction mixture, rather than being irreversibly deactivated through
oxide aggregation. Others have recently noted negligible restructuring in NbOxꢀSiO2 catalysts for
limonene epoxidation.78 DRUVꢀvis spectra mirror these changes (Figure S8), with the optical
edge of NbꢀSiO2 decreasing from 4.0 eV to 3.0 eV after use and washing, but recovering to 3.7
eV after recalcination. In accordance with these spectral changes, the recovered NbꢀSiO2 catalyst
is poorly active after simple washing, but recovers ~90% of its activity after a recalcination. In
contrast, DRUVꢀvis spectra of recovered TaꢀSiO2 indicate severe restructuring upon
recalcination (Figure S9).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Finally, it must be noted that the asꢀsynthesized CalixꢀNbꢀSiO2 is not particuarily robust: the
DRUVꢀvis (Figure S8) suggests the ligand is largely lost during reaction and/or washing, also
consistent with an increase in the XANES preꢀedge intensity after use and washing, and the
known tendency for ligand loss during synthesis. While some of us have seen mmCalixꢀTiꢀSiO2
and CalixꢀTaꢀSiO2 appear to be quite stable in prior studies,21, 40 it is recommended to calcine the
freshly prepared CalixꢀNbꢀSiO2 in the interest of stability.
DISCUSSION
The reactivity data set is examined in several ways to draw larger conclusions. First, Figure 7
plots pathway selectivity vs. yield of oxygenates in cyclohexene oxidation. Tiꢀ, Taꢀ, and NbꢀSiO2
are all able to reach >80% yield of oxygenates, whereas Zrꢀ and Hfꢀcontaining catalysts give
poor yields even over long times. VꢀSiO2, which is selective to the radical pathway, gives the
worst yields since it utilizes oxidant less efficiently.21, 75 It should also be emphasized that Figure
7 plots the integrated pathway selectivity, not the net selectivity to epoxide, which is given in
Table S2. The most active catalysts for epoxidation do also have a high rate of epoxide
hydrolysis.
ACS Paragon Plus Environment
23

ACS Catalysis
Page 24 of 35
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 7. Integrated selectivity towards direct oxidation pathways vs. total yield of C6 product
for cyclohexene oxidation at 65°C by (a) calcined catalysts (filled circles) and (b) calixareneꢀ
containing catalysts (open circles). Calculated selectivities at less than ~10% yield are unreliable
and have been removed from comparison.
Figure 7a illustrates three main modes of reactivity within this catalyst family: (1) calixareneꢀ
derived NbꢀSiO2, which uniquely maintains high pathway selectivity (~85%) across all yields;
(2) VꢀSiO2, which is always poorly selective (<20%); and (3) all other catalysts – including
NbCl5ꢀSiO2 – which suffer from earlyꢀstage (<10% yield) low pathway selectivities that improve
at higher conversions. We note again that Figure 7 plots selectivities integrated over the entire
product yield to that point; thus, the upward trend in pathway selectivity (dotted curve) is a
consequence of the system being poorly selective at low conversion (i.e. large initial production
of radical products), followed by a pronounced switch to high pathway selectivity as the reaction
proceeds.
Figure 7b shows that retaining the calixarene ligand enhances the pathway selectivity of the 2nd
and 3rd row cations relative to their calcined analogues at all levels of conversion. In particular,
ACS Paragon Plus Environment
24

Page 25 of 35
ACS Catalysis
1
2
3
4
5
6
CalixꢀTaꢀSiO2 is remarkably selective compared to its calcined version, consistent with earlier
observations by some of us.21 The ligand has a weaker effect on Ti and V.
7
8
9
A unifying hypothesis is that aggregates of MOx, as distinguished from isolated M cations, are
responsible for the observed initial radical reactivity. DRUVꢀvis (Figure 2b) and XANES (Figure
6) show the NbꢀSiO2 to be undercoordinated relative to NbCl5ꢀderived catalysts, and DRUVꢀvis
indicates the presence of some larger aggregates in TaꢀSiO2 (Figure 2b). All these latter catalysts
initially show the radical pathway. Further, the presence of coordinated calixarene will help
minimize MOx formation under reaction conditions, boosting pathway selectivity in nearly all
cases. Application of chemical titration techniques recently developed by our group88 is currently
underway to address these unique behaviors and others.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For the calcined catalysts, we further compare reactivity patterns with the ionic character of the
metal–oxygen bond at the SiO2 surface using the Pauling scale of electronegativity (Figure 8).
Here, we approximate ionic character by subtracting each metal’s Pauling electronegativity89
from that of oxygen (i.e. χO–χM) and correlate this metric with our calculated initial turnover
frequencies. Alternately, the following conclusions hold if the stoichiometry of an assumed MO4
monomer is geometrically averaged for the electronegativity metric (i.e. by calculating
[(χM)•(χO)4]1/5 – see Figure S10).
ACS Paragon Plus Environment
25

ACS Catalysis
Page 26 of 35
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 8. Correlations between oxidation TOFs (minꢀ1) with ionic character of the M–O bond for
calcined catalysts. Ionic character is estimated by subtracting the Pauling electronegativity of
each metal from that of oxygen.89 Error bars represent alternative calculations of TOF from 10
min (high) and 30 min (low) of reaction. Redoxꢀactive VꢀSiO2 has been excluded from
comparison.
Initial, direct TOF for cyclohexene is both strongly sensitive to, and tightly correlated with,
decreasing ionic character over a nearly 100ꢀfold change in rate, suggesting that this reaction is a
readout of intrinsic Lewis acidity. Others have strongly correlated atomic electronegativity with
Lewis acid strength for mainꢀgroup elements in their highest oxidation state,90 as well as for
lanthanide(III) complexes.89, 91 The total TOF for styrene also correlates strongly with ionic
character, but is less sensitive. Interestingly, the relative intensity of the calixarene phenolate
LMCT in the DRUVꢀvis spectra of Figure 2a varies dramatically across different metals, and in
the same order as epoxidation reactivity and ionic character, again with the exception of V. This
trend suggests that the relative intensity of this feature is sensitive to the strength of ligand
binding, and itself is a proxy for reactivity in Lewis acid catalysis.
Finally, the total TOF for cyclohexene (i.e. including radical products) is less strongly
predictive of Lewis activity, confounded by the kinetically coupled pathways. VꢀSiO2 has been
ACS Paragon Plus Environment
26

Page 27 of 35
ACS Catalysis
1
2
3
4
5
6
7
8
9
excluded from these comparisons since it appears to proceed by a distinct redox mechanism and
due to observations of leaching.4, 30 According to this calculation, Nb is an intrinsically good
catalyst for electrophilic activation of H2O2 because it can be synthesized in a stable, highly
undercoordinated state, is strongly covalent, and is practically redoxꢀinactive under these
conditions. Beyond high activity for electrophilic activation of H2O2, the NbꢀSiO2 synthesized
from the calixarene precursor is also uniquely pathway selective within this series, giving <15%
radical pathway selectivity at all conversions. Crucially, the rates and pathway selectivities of the
conventionallyꢀsynthesized NbCl5ꢀSiO2 do not fit along these trends, further demonstrating the
importance of the calixarene platform for controlling the oxide structure and enabling this
systematic study.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
CONCLUSIONS
Catalysts for styrene and cyclohexene epoxidation with H2O2 were synthesized by grafting
calixarene complexes of Ti(IV), Zr(IV), Hf(IV), V(V), Nb(V) and Ta(V) (all of Groups IV and
V) to SiO2 at surface densities 0.22 ± 0.03 M nmꢀ2. Typically, different supported metal oxides
are synthesized in separate investigations and by a range of techniques, resulting in a range of
supported oxide structures. This structural diversity can make it difficult to compare reactivities
and selectivities across related metal oxides. Here, the systematic use of a single type of
precursor reveals a 100ꢀfold change in Lewis acidꢀcatalyzed direct cyclohexene epoxidation
rates, which correlate well with ionic bond character of the supported oxide. VꢀSiO2 is an
exception that appears to follow a distinct mechanism. This trend is only observed when the
calixarene ligand is used in the precursor synthesis; NbꢀSiO2 from NbCl5 is much less reactive
and selective. DRUVꢀvis and XANES indicate that systematic use of the calixarene complex in
synthesis helps keep the final, calcined catalyst highly dispersed, improving reactivity and
ACS Paragon Plus Environment
27

ACS Catalysis
Page 28 of 35
1
2
3
4
5
6
7
8
9
allowing comparisons amongst structurallyꢀsimilar sites. Even for TaꢀSiO2, where some
aggregated TaOx is evidenced from the DRUVꢀvis, its agreement with the correlation of Figure
8 indicates that the vast majority of the Ta sites are active, isolated sites. Finally, the asꢀ
synthesized catalysts (with the ligand retained) have markedly higher direct pathway selectivities
for the 2nd and 3rd row transition metal oxides, again presumably because the ligand helps keep
the individual cations isolated from one another.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Throughout this study, the calcined NbꢀSiO2 derived from the calixarene precursor is highꢀ
performing, with an initial TOF in cyclohexene epoxidation (2.4 minꢀ1) >2x faster than a more
standard TiꢀSiO2. Further, all materials except this NbꢀSiO2 exhibit an initial period of radical
oxidation before switching to direct pathways as the reaction proceeds. This gives the NbꢀSiO2
an exceptional pathway selectivity of ~85% over the entire course of the reaction. Finally,
calcined NbꢀSiO2 is stable to leaching and maintains activity and pathway selectivity upon
regeneration and reuse. Overall, NbOxꢀSiO2 materials have received much less attention as
selective oxidation catalysts78, 92ꢀ93 in contrast with extensive work on TiꢀSiO2. This includes
supported22, 25ꢀ26 and coꢀcondensed23, 94ꢀ95 NbOxꢀSiO2 catalysts that have been demonstrated by
others as highly active and selective for alkene epoxidation. These results highlight supported
NbꢀSiO2 as a highly active and pathwayꢀselective catalyst for these transformations – a
conclusion that is only reached when it is synthesized utilizing bulky precursors such as the
calixarene complex that enable and enforce high dispersion of the oxide. Additional studies on
these highꢀperforming NbꢀSiO2 catalysts are underway.
ASSOCIATED CONTENT
Material included in Supporting Information: detailed synthetic procedures, NMR resonances
and elemental analyses for calixarene complexes and grafted catalysts; synthesis procedures and
ACS Paragon Plus Environment
28

Page 29 of 35
ACS Catalysis
1
2
3
4
5
6
7
8
9
powder xꢀray diffraction spectra of xꢀray absorption standards YNbO4 and CaNb2O6; full kinetic
data for cyclohexene and styrene epoxidation at 45°C and 65°C; cyclohexene epoxidation (65°C)
leaching (Ti, V, Nb, Ta) and recycle (Nb) experiments; additional characterization (UVꢀvis, 13Cꢀ
NMR) for calixarene complexes of Ti, Zr, Hf, V, Nb, and Ta; xꢀray absorption spectroscopy E0
(eV) for Ti, V, Nb and Ta; Gaussian peak fitting for Ti and Nb preꢀedge XANES features;
DRUVꢀvis spectroscopy of recovered NbꢀSiO2, CalixꢀNbꢀSiO2 and TaꢀSiO2; and an alternate
correlation between oxidation TOF and ionic character of assumed MO4 oxide monomers. This
material is available free of charge via the Internet at http://pubs.acs.org.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
AUTHOR INFORMATION
Corresponding Author
* Email: jꢀnotestein@northwestern.edu
Author Contributions
The manuscript was written through contributions of all authors. All authors have given approval
to the final version of the manuscript.
ACKNOWLEDGMENT
N.E.T. and J.M.N. acknowledge financial support from the Dow Chemical Company.
A.B.T. acknowledges financial support from NSF grant CBETꢀ0933667. Characterization was
performed at IMSERC at Northwestern University with the support of NSF grant DMRꢀ
0521267, and at the DuPontꢀNorthwesternꢀDow Collaborative Access Team (DNDꢀCAT) at the
Advanced Photon Source at Argonne National Laboratory (DOE Contract No. DEꢀAC02ꢀ
06CH11357). N.E.T. thanks Dr. Qing Ma for his assistance with Xꢀray absorption spectroscopy
ACS Paragon Plus Environment
29

ACS Catalysis
Page 30 of 35
1
2
3
4
5
6
7
8
9
experiments. Powder xꢀray diffraction experiments were performed at the J. B. Cohen XꢀRay
Diffraction Facility supported by the MRSEC program of the National Science Foundation
(DMRꢀ1121262) at the Materials Research Center of Northwestern University.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
30