Reaction pathways and deactivation mechanisms of isostructural Cr and Fe MIL-101 during liquid-phase styrene oxidation by hydrogen peroxide

Catalysis

Science &

Technology

View Article Online

View Journal | View Issue

PAPER

Reaction pathways and deactivation mechanisms

of isostructural Cr and Fe MIL-101 during liquid-

phase styrene oxidation by hydrogen peroxide†

Cite this: Catal. Sci. Technol., 2021,

11, 5282

*

Rachel A. Yang

and Michele L. Sarazen

Isostructural Cr and Fe nanoporous MIL-101, synthesized without mineralizing agents, are investigated for

styrene oxidation utilizing aqueous hydrogen peroxide to yield valuable oxygenates for chemical synthesis

applications. Styrene conversion rates and oxygenate product distributions both depend on metal identity,

as MIL-101(Fe) is more reactive for total styrene oxidation and is more pathway selective, preferring

aldehyde (benzaldehyde) formation at the α-carbon to the aromatic ring, where MIL-101(Cr) sustains

epoxide (styrene oxide) production at the same α-carbon. These pathways often involve hydrogen

peroxide derived radical intermediates (O, –HOO˙, –HO⁻˙) and metallocycle transition states. We postulate

that the higher reactivity of one of these surface intermediates, Fe(^IV)O relative to Cr(IV)O, leads to

higher styrene oxidation rates for MIL-101(Fe), while higher electrophilicity of Cr(^III)–OOH intermediates

translates to the higher styrene oxide selectivity observed for MIL-101(Cr). Secondary styrene oxide and

benzaldehyde conversions are observed over both analogs, but the former is more prevalent over MIL-

101(Fe) due to higher Lewis/Brønsted acid site density and strength compared to MIL-101(Cr). Recyclability

experiments combined with characterization via XRD, SEM/EDXS, and FT-IR and UV-vis spectroscopies

show that the nature of MIL-101(Fe) sites does not change significantly with each cycle, whereas MIL-

101(Cr) suffers from metal leaching, which impacts styrene conversion rates and product distribution. Both

catalysts require active site regeneration, though MIL-101(Fe) sites are more susceptible to reactivation,

even under mild conditions. Finally, examination of styrene conversion for three unique synthesized phases

of MIL-101(Cr) rationalizes that nodal defects are largely responsible for observed reactivity and selectivity

but predispose the framework to metal leaching as a predominant deactivation mechanism.

Received 31st March 2021,

Accepted 30th June 2021

DOI: 10.1039/d1cy00567g

rsc.li/catalysis

valorization applications. Many of these processes are

performed in liquid-phase, batch systems and utilize

Introduction

Metal–organic frameworks (MOFs) are a class of crystalline,

porous solids that have garnered attention by merit of their

modular and attractive physiochemical properties. They have

been investigated for diverse applications including gas

separation and storage,¹water purification,²drug delivery,³

sensing,^4,5nonlinear optics,⁶and catalysis.^7–10In the context

of catalysis, MOFs have exhibited activity for small-molecule

transformations including the carbonation of light

alkenes,^11–13Friedel–Crafts reactions,¹⁴cross-coupling

reactions,¹⁵and the oxidation of aromatic and linear

hydrocarbons.^16–18

transition metal molecular catalysts, which have some

unfavorable economical and sustainability attributes, such as

additional separation costs and challenges in regeneration

for reuse in subsequent reaction cycles. Though there are

some notable examples of recyclable homogeneous catalysts

for aryl and alkyl oxidations,^19–21generally, industrial

application of homogeneous catalysts is still limited by costly

separations needed for catalyst recovery. As such, we

investigate MOFs as solid analogs for transition metal

organometallics in heterogeneous liquid-phase, small-

molecule organic oxidations. A subset of these nanoporous

(<100 nm) networks have demonstrated notable reactivity

and selectivity for the oxidations of some alkyl and aryl

alkenes by benign oxidants like hydrogen peroxide (H₂O₂)

and air.^7,9,17,22However, widespread MOF use is marred by

thermochemical instability relative to more conventional

porous materials like bulk metal oxides and zeolites. One

exception is the MIL-101 family, which includes some

transition metal variants that are both active for oxidative

Here, we focus on MOF-catalyzed oxidation reactions,

motivated by the prevalence of this class of transformations

in pharmaceutical, fine chemical, and other hydrocarbon

Department of Chemical and Biological Engineering, Princeton University,

Princeton, New Jersey 08544, USA. E-mail: msarazen@princeton.edu

† Electronic supplementary information (ESI) available. See DOI: 10.1039/

d1cy00567g

5282 | Catal. Sci. Technol., 2021, 11, 5282–5296

This journal is © The Royal Society of Chemistry 2021

View Article Online

Catalysis Science & Technology

Paper

chemistry and are stable in air (up to 548 K), water, and

many common polar and nonpolar solvents.^7,22,23Further, it

is a higher surface area material unlike most bulk metal

oxides and does not suffer from limited pore size ranges and

fast deactivation in liquid-phase reactions due to poisoning,

pore occlusion, and severe metal leaching unlike zeolites.^17,22

The oxidation of styrene by hydrogen peroxide is selected

as a probe reaction for the value of its oxygenate products in

chemical synthesis applications and for its green appeal,

since hydrogen peroxide is a cleaner oxidant compared to

organic peroxides, which often have toxic byproducts.¹⁷

Further, this system affords an opportunity to probe metal-

oxidant interactions, reactivity, and material stability in an

oxidizing environment that can accelerate catalytic/metal

cycles. The complex nature of the involved reaction pathways

can probe different sites and defects within the model MIL-

101 family that can be extended to other small-molecule

organic oxidations on isolated-site transition metal catalysts.

Structurally, MIL-101 is a nanoporous solid comprised of

metal oxide trimer node species ([M₃O(OH)₃]ⁿ⁺) coordinated

to 1,4-benzene dicarboxylic acid (BDC) linkers that assemble

to form an MTN zeotype crystal structure (Fig. 1) with 3.4 nm

cages accessible by 1.6 nm hexagonal windows and smaller

2.9 nm cages accessible by 1.2 nm pentagonal windows.²³

This large (mesoporous) and accessible pore network tends

to mitigate diffusion-limited reactions for a wide range of

substrates, as larger pores allow more facile access to internal

active sites. Further, its ordered network with well-defined

active site location and nature facilitates a more controlled

study of site evolution during catalytic cycles. This family is

synthetically diverse, including chromium, iron, vanadium,

and titanium analogs, some of which have established

activity for various catalytic alkyl and aryl oxidation reactions

in the liquid-phase at mild reaction conditions.^{16,17,24–27}

However, precise reaction pathways and deactivation

mechanisms are still elusive for many MOF-catalyzed

oxidation systems. As such, this work investigates a probe

liquid-phase aryl oxidation over transition metal MIL-101 to

elucidate catalyst reactivity and pathways, physicochemical

property evolution to reveal deactivation mechanisms, and

active site recyclability, which is broadly important for

efficient material use.

Here, we focus on MIL-101(Fe) and MIL-101(Cr) because

they contain earth-abundant transition metal sites and have

demonstrated activity for styrene oxidation,¹⁷as well as other

small-molecule organic oxidations,^16,24,28by a variety of

oxidants like H₂O₂, air, and t-butyl hydrogen peroxide

(TBHP). Moreover, MIL-101(Fe) and MIL-101(Cr) reportedly

do not suffer from the same extensive metal leaching¹⁷that

is observed with V-based MOFs or from inactivity observed

with Ti-based MOFs during cyclic alkene oxidation.²⁷Metal

leaching has been reported for MIL-101(Cr) and MIL-101(Fe)

in the presence of H₂O₂,²⁹but the impacts on catalytic

activity have not been thoroughly quantified. It has been

observed that styrene oxidation at mild reaction conditions

(353 K) on both MIL-101(Cr) and MIL-101(Fe) have the

highest epoxide selectivity when TBHP is used, followed by

air and H₂O₂.¹⁷Selectivity to other oxygenates is also

impacted by oxidant identity, as are styrene conversion

trends. MIL-101(Fe) shows higher styrene conversion for air

> TBHP > H₂O₂in contrast to H₂O₂> air > TBHP for MIL-

101(Cr).¹⁷This suggests that there are underlying textural

and electronic differences between these two analogs that

potentially manifest in different mechanisms. Yet, these

underlying properties and their implications for MOF-

catalyzed aromatic oxidation reaction pathways are neither

well understood nor quantified, even though existing

literature has found differences in catalytic activity between

Fig. 1 MIL-101(M) nanoporous MTN-zeotype framework with trimeric metal node structure and organic linker secondary building units (SBUs)

with nominal chemical formulas.

This journal is © The Royal Society of Chemistry 2021

Catal. Sci. Technol., 2021, 11, 5282–5296 | 5283

View Article Online

Paper

Catalysis Science & Technology

these two isostructural frameworks for a handful of other

liquid-phase small molecule oxidations.^25,28Further, there is

a dearth of studies that go beyond assessment of activity in

terms of conversion or quantitatively investigate deactivation

mechanisms for MOF-catalyzed oxidation reactions. These

mechanisms are imperative to understanding material

limitations that arise under realistic reaction conditions, to

proposing deactivation mitigation strategies, and to

overnight without stirring. The emerald-green solid was

separated by centrifugation, washed with methanol and

acetone sequentially, dried at 348 K overnight and then in

vacuum (125 torr) at ambient temperature for 2 days. Typical

yield of the emerald-green solid was 14% by mass on a Cr basis

with a nominal chemical formula of [Cr₃(O)(OH)(BDC)₃(H₂O)₂].

Typical synthesis of MIL-101(Fe)

providing

a thorough basis for comparison for future

synthetic efforts to improve material stability. Overall, this

work examines isostructural MIL-101(Cr) and MIL-101(Fe)

frameworks to rationalize the effects of metal identity and

other intrinsic material properties on MOF-catalyzed

oxidation pathways and elucidates deactivation mechanisms

with rigorous quantification of turnover rates and product

distributions to define the efficacy and limitations of these

novel transition metal catalysts through their life cycles for

styrene oxidation by hydrogen peroxide, a representative aryl

oxidation.

In a typical synthesis of MIL-101(Fe), 2.45 mmol of iron(^III)

chloride hexahydrate and 1.24 mmol of BDC were mixed in

15 mL of DMF until clear with a goldenrod-yellow color. The

mixture was sonicated for 15 minutes and then charged into

a 23 mL Teflon-lined stainless-steel Parr autoclave. The

autoclave was heated at 383 K for 20 hours and after cooling

to ambient temperature naturally, the resulting red-orange

solid was collected via centrifugation (1000 × g, 5 min). The

solid was washed two times, 3 hours each, with ethanol at

333 K under refluxing conditions. The purified orange

product was separated via vacuum filtration and dried at 343

K overnight. Typical yield of the orange solid was 32% by

mass on an Fe basis with a nominal chemical formula of

[Fe₃(O)(Cl)(BDC)₃(H₂O)₂].

Experimental

Materials and methods

1,4-Benzene dicarboxylic acid (BDC, Acrōs Organics, 99+%),

iron(III) chloride hexahydrate (FeCl₃·6H₂O, Acrōs Organics,

99%), chromium(^III) nitrate nonahydrate (Cr(NO₃)₃·6H₂O,

Acrōs Organics, 99%), N,N-dimethylformamide (DMF, Fisher

Chemical, 99.9%), ethanol (EtOH, Fisher Chemical, 97%),

methanol (MeOH, J.T. Baker, 99.9%), and acetone (VWR

Chemicals, 99.5%) were used for synthesis of MIL-101(Fe)

and MIL-101(Cr). Styrene (stabilized, Acrōs Organics, 99.5%),

acetonitrile (CH₃CN, Honeywell Riedel-de Haën, 99.7+%),

and hydrogen peroxide (H₂O₂, Fisher Chemical, 30% v/v in

H₂O) were used for catalytic experiments. Deuterated

Catalyst characterization

Powder X-ray diffraction (XRD) patterns were obtained on a

Bruker D8 Discover diffractometer with Cu-Kα radiation (λ =

1.5418 Å) operating at 40 kV and 40 mA. Images were

procured using an XL30 field emission gun scanning electron

microscope (SEM) and an Evex energy dispersive X-ray

spectroscopy (EDXS) accessory was used to determine

approximate elemental composition and for elemental

mapping. A Nicolet iS50 infrared spectrophotometer was

used to gather spectra for all materials; all solid samples (2

wt%) were mixed with KBr and pressed into pellets utilizing

a hydraulic press at 11 000 psi for 5 minutes each. UV-vis

spectroscopy was used to analyze both liquid samples and

solid suspensions and measurements were performed on an

dimethyl

sulfoxide

(DMSO-d₆,

Cambridge

Isotope

Laboratories, 99.9%) was used as the solvent for ¹H NMR

analysis. All starting materials and organic solvents were

used as received without further purification. MIL-101(Cr)

and MIL-101(Fe) were synthesized without the use of toxic

mineralizing agents like hydrofluoric acid (HF), following

published literature procedures with some modifications.^16,30

Agilent

Technologies

Cary

5000

UV-vis-NIR

spectrophotometer. Multipoint Brunauer–Emmett–Teller

(BET) surface area and non-local density functional theory

(NLDFT) pore size measurements were calculated from N₂

adsorption and desorption isotherms (77 K) collected on a

Micromeritics 3Flex physisorption unit. The pressure (P/P₀)

ranges used for BET surface area calculations were chosen to

satisfy the maximum number of consistency criterion³¹and

typical ranges were 0.05 to 0.18 for MIL-101(Cr) and 0.01 to

0.15 for MIL-101(Fe). The N₂-Tarazona NLDFT model with

cylindrical geometry was used for pore size distributions. All

samples were degassed at 383 K under mild vacuum (125

torr) overnight prior to N₂physisorption unless stated

otherwise. A Micromeritics Autochem II 2920 was used to

perform temperature programmed desorption (TPD)

measurements utilizing NH₃and CO₂, as detailed in the

supporting information.

Typical synthesis of MIL-101(Cr)

MIL-101(Cr) was synthesized by first mixing equimolar (5

mmol) chromium(^III) nitrate nonahydrate and BDC in 24 mL

deionized water at ambient temperature and pressure. The

acidic dark indigo suspension was sonicated for 10 minutes

before it was charged into a 45 mL Teflon-lined stainless-

steel Parr autoclave and put into a convection isothermal

oven at 491 K for 18 hours at autogenous pressure without

stirring. The mixture was cooled naturally to ambient

temperature. Afterward, the solids were separated by

centrifugation (5000 × g, 10 min) and washed with water,

methanol, and acetone in succession. The remaining solid

was separated from acetone via centrifugation, introduced to

20 mL DMF, sonicated for 10 minutes, and kept at 343 K

5284 | Catal. Sci. Technol., 2021, 11, 5282–5296

This journal is © The Royal Society of Chemistry 2021

View Article Online

Catalysis Science & Technology

Paper

Catalytic styrene oxidation by hydrogen peroxide

reported in literature and with the simulated pattern for a

defect-free crystal with long-range translational order

acquired from RIETAN-FP in VESTA.^16,30,32FT-IR spectra

(Fig. 2b) for both materials exhibit characteristic symmetric

and asymmetric stretching modes of O–C–O functionalities

in the 1400 to 1600 cm⁻¹range, aromatic CC stretching

vibrations near 1500 cm⁻¹, and C–H in-plane bending

vibrations at approximately 1160 and 1020 cm⁻¹and out-of-

plane bending at around 885 and 750 cm⁻¹.^32–34The Fe

analog has an additional band at 550 cm⁻¹, which is typical

for a MIL-101 Fe–O stretch.³⁵Since Cr–O has a lower reduced

mass than Fe–O (12.2 g versus 12.4 g), the band at 587 cm⁻¹

is likely the analog Cr–O vibration, as this band is also

observed in Cr(^III) oxide;³⁶however, this band has not been

assigned formally in MIL-101 literature to our knowledge

despite being present. That assignment aside, all

aforementioned bands are consistent with the MIL-101

framework and provide further evidence of successful

syntheses of both analogs with functionalities comparable to

other reported materials.¹⁷From SEM images (Fig. 2c and d),

both analogs have self-consistent, well-defined crystal

aggregate geometry and size distributions (Fig. S3†) centered

at 0.31 0.09 μm and 1.2 0.3 μm for MIL-101(Cr) and MIL-

101(Fe), respectively; this size disparity is a result of different

synthetic conditions from the available literature procedures

used for each material.^30,32For example, hydrothermal

synthesis for MIL-101(Cr) starts with a metal : linker : solvent

molar ratio of 1 : 1 : 222 whereas solvothermal synthesis in

DMF for MIL-101(Fe) starts with a ratio of 2 : 1 : 156. These

differences, in addition to higher temperatures and shorter

times for MIL-101(Cr) synthesis (491 K, 18 h) compared to

MIL-101(Fe) synthesis (383 K, 20 h), likely impact crystal

nucleation and growth rates. EDXS elemental mapping (Fig.

S5a and S6†) indicates that the Cr (12.8 at%) and Fe (6.6

at%) are distributed throughout MIL-101(Cr) and MIL-

101(Fe), respectively. This, combined with the lack of Cr

and Fe oxide peaks in XRD patterns (Fig. S2†), suggests that

Reaction–reactivation cycling experiments were conducted at 323

K in 100 mL round-bottom flasks under refluxing conditions with

a starting broth composition of 10 mL acetonitrile, 10 mmol

styrene, and 30 mmol hydrogen peroxide. 50 mg of catalyst was

subsequently added to initiate the reaction. After 3600 s (1 h), the

solid was separated using centrifugation (1000 × g, 5 min) and a

20 μL aliquot of the liquid supernatant was mixed thoroughly in

DMSO-d₆for analysis via ¹H NMR on a Bruker 500 MHz

spectrometer. Resulting spectra were integrated and the areas

were utilized to calculate styrene conversion and product

selectivities on a fractional basis. All spectra were repeatedly

integrated, and associated error was calculated from standard

deviations of these integrated areas; error was propagated

appropriately with each functional transformation. The recovered

solid was reactivated by washing with acetonitrile and ethanol in

succession and then drying under mild vacuum conditions

overnight (125 torr). The resulting solid was then retested for

catalytic activity in fresh reaction medium. Temporal experiments

were run similarly, with a 50 mg initial catalyst charge, but

liquids were scaled up by a factor of 5 (50 mL acetonitrile, 50

mmol styrene, 150 mmol H₂O₂) with respect to the cycling

experiment broths so that aliquots (20 μL) taken at various time

points did not significantly impact the total solution volume.

Hot filtration tests were performed with the same starting

broth composition as the cycling experiments, but after 3600

s of reaction, the solid was removed via vacuum filtration

and the warm filtrate was quickly reintroduced to a fresh 100

mL reaction flask at temperature. The filtrate was allowed to

react for another 3600 s and then a 20 μL aliquot was taken

for ¹H NMR analysis.

Results and discussion

Textural characterization of catalytic materials

XRD patterns (Fig. 2a) for MIL-101(Cr) and MIL-101(Fe) show

characteristic framework peaks that are consistent with those

Fig. 2 (a) MIL-101(Cr, Fe) XRD patterns with simulated MIL-101 pattern, (b) FT-IR spectra of MIL-101(Cr, Fe), and SEM images of (c) MIL-101(Fe)

and (d) MIL-101(Cr).

This journal is © The Royal Society of Chemistry 2021

Catal. Sci. Technol., 2021, 11, 5282–5296 | 5285

View Article Online

Paper

Catalysis Science & Technology

Fig. 3 (a) N₂adsorption (Cr (black circles), Fe (black diamonds)) and desorption (Cr (grey circles), Fe (grey diamonds)) isotherms at 77 K and

multipoint BET surface areas. NLDFT PSD for (b) MIL-101(Cr) and (c) MIL-101(Fe).

neither material forms large metallic or metal oxide

clusters.

in the 2.1–2.9 nm range. Lower surface area MIL-101(Fe)

samples (<300 m²g⁻¹) show no microporosity and very broad

mesopore distribution whereas intermediate samples (300

m²g⁻¹< x < 500 m²g⁻¹) retain some microporosity but start

to show less defined mesopore distributions with increased

broadening (Fig. S7†). However, we note that experimental

conditions for N₂physisorption are not representative of the

pretreatments or liquid-phase environments employed in the

following catalytic studies and instead serve as a means to

compare synthesized materials to literature. In fact, two

samples of MIL-101(Fe) with nearly identical textural

properties that showed BET surface areas of 590 10 m²g⁻¹

and 860 10 m²g⁻¹showed identical catalytic performance,

as measured by turnover rates and product distribution (Fig.

S19 and S20†). Overall, the MIL-101(Fe) and MIL-101(Cr)

syntheses were successful in producing high surface area,

crystalline materials to be utilized in the following catalytic

experiments.

N₂physisorption measurements at 77 K (Fig. 3a) result in

high multipoint BET surface area, 3300 50 m²g⁻¹for MIL-

101(Cr) and 590

10 m²g⁻¹for MIL-101(Fe). MIL-101(Fe),

however, was found to exhibit lower BET surface areas (150–

500 m²g⁻¹) when degassed at 383 K and 125 torr overnight,

seemingly because the harsher procedure leads to partial

collapse of pores due to strong capillary forces associated

with solvent evacuation^37,38(Fig. S7†). As such, milder

degassing conditions (313 K, 125 torr for 2 days) were

adopted and yielded higher surface area measurements

(Fig. 3a and S8a†). Differences in literature purification and

activation procedures^16,17,32may also be implicated for lower

surface area MIL-101(Fe); insufficient washing solvent volume

or washing with solvents that exhibit low impurity solubility

allows pervasive occluding species, including unreacted

starting materials and amorphous Fe-BDC phases, to remain

in the pores. Meanwhile, subsequent thermal or reduced

pressure activation to remove washing solvent molecules

from material pores may lead to occluded residual solvent, if

the treatment is too mild, or collapse of pores, if the

treatment is too harsh.^37–39NLDFT pore size distributions

(PSD), based on N₂physisorption, demonstrate that MIL-

101(Cr) (Fig. 3b) has micropore diameters in the 0.6–1.1 nm

range and mesopore diameters in the 1.9–2.9 nm range,

which are consistent with literature PSDs for this analog.¹⁷

The former range is attributed to the microporous super

tetrahedra (Fig. 1), whilst the latter range represents

framework cages.⁴⁰Similarly, high surface area (>500 m²g⁻¹)

MIL-101(Fe) (Fig. 3c) has dimensions in the microporous

range of 0.6–1.0 nm and sharply defined mesopore diameters

Temporal experiments and the examination of reaction

pathways

Temporal fractional styrene conversion (Fig. 4a) and

oxygenate concentration profiles (Fig. 4b and c) are first

obtained to define the reactivity of each analog. Styrene

conversion is higher for MIL-101(Fe) compared to MIL-

101(Cr) at all time points for the same initial charge of

catalyst on a total mass basis. Reactivity for styrene oxidation

for MIL-101(Fe) is also higher relative to MIL-101(Cr), as

determined from pseudo first-order reaction rates normalized

by initial catalyst mass loadings and by moles of metal

determined from EDXS. Further, oxygenate concentration

5286 | Catal. Sci. Technol., 2021, 11, 5282–5296

This journal is © The Royal Society of Chemistry 2021

View Article Online

Catalysis Science & Technology

Paper

Fig. 4 Temporal (a) fractional styrene conversion for MIL-101(Cr) (grey) and MIL-101(Fe) (black) and oxygenate (styrene oxide (red), benzaldehyde

(blue), styrene glycol (grey), benzoic acid (green), phenylacetaldehyde (purple)) concentrations for (b) MIL-101(Cr) and (c) MIL-101(Fe) at 323 K (50

mg catalyst, 50 mL acetonitrile, 50 mmol styrene, 150 mmol H₂O₂). Dashed lines are to guide the eye.

profiles differ between MIL-101(Fe) and MIL-101(Cr), though

they both eventually generate benzaldehyde as the majority

product by 3600 s and neither catalyst produces detectable

amounts of phenylacetaldehyde. Most notably, sustained

styrene oxide concentrations are observed over MIL-101(Cr),

while any produced over MIL-101(Fe) is quickly consumed

after 240 s. After 480 s, product ordering from highest to

lowest concentration species stabilizes over MIL-101(Fe)

(benzaldehyde > benzoic acid > styrene glycol > styrene

oxide). In contrast, the product profile for MIL-101(Cr)

switches at 1800 s, which coincides with a change from a

clear to a purple tinted reaction broth. Before 1800 s, MIL-

101(Cr) only produces styrene oxide and benzaldehyde

whereas after 1800 s, styrene glycol and benzoic acid are

observed with product ordering settling to: benzaldehyde >

styrene glycol > styrene oxide > benzoic acid at 3600 s. We

postulate that the change in product distribution on MIL-

101(Cr) and the concurrent color changes suggest leached Cr

complexes that are catalytically active, which is further

discussed in the following section.

We note that other products are observed, with the

majority being styrene oligomers. Polymerization of styrene

occurs readily via self-initiation where two styrene molecules,

in the absence of stabilizing agents, first form a Diels–Alder

dimer followed by hydrogen transfer to another styrene

molecule in molecular-assisted homolysis; the resulting two

monoradicals then undergo polymerization to form

polystyrene.^41,42However, stabilized styrene conversion in the

absence of catalyst is negligible (<1%) at 3600 s and 323 K,

suggesting polymerization must be catalyzed by solid acids,

or by MIL-101(Fe)/MIL-101(Cr) in this case. Other trace

products in the presence of MIL-101(Cr) and MIL-101(Fe)

Scheme 1 Proposed reaction pathways, metallocycle transition states, and radical surface intermediates for styrene oxidation by hydrogen

peroxide over MIL-101(M).

This journal is © The Royal Society of Chemistry 2021

Catal. Sci. Technol., 2021, 11, 5282–5296 | 5287

View Article Online

Paper

Catalysis Science & Technology

include various hydrated species from carbonyl-containing

compounds and condensation products. However, we focus

on the main products that arise from oxidative catalytic

turnovers here and in all following discussion, where

oxygenate concentration profiles demarcate distinct product

distributions by 3600 s for each analog.

We construct a reaction network (Scheme 1) consistent

with reported observations to understand the origin of

measured oxygenates.^43,44Direct oxidation of styrene yields

styrene oxide and benzaldehyde as primary products,

although the nucleophilic attack of hydrogen peroxide on

styrene oxide followed by CC bond cleavage has been

reported as an alternate pathway to form benzaldehyde.^17,45

The latter pathway, however, is not prevalent at the reaction

conditions studied here. This direct styrene oxidation to form

styrene oxide or benzaldehyde is transition metal catalyzed

Fe sites, while hydroxyl radical formation occurs similarly in

the presence of both. It is the reactivity of these

intermediates, however, that confers the observed differences

in reactivity for styrene oxidation and in product selectivities

between MIL-101(Cr) and MIL-101(Fe). Specifically, oxophilic

character suggests that oxygen abstraction or superoxide/

hydroperoxo hydrolysis is more facile for Cr than for Fe,

which is supported by lower enthalpies of formation for the

former calculated via DFT.⁵³However, high valent Fe-oxo

(Fe(^IV)O) is well known to be proficient at abstracting

hydrogen atoms from strong C–H bonds.^53,54For example,

Fe(^IV)O has a lower calculated apparent energy barrier than

Cr(^IV)O for apolar C–H activation in methane.⁵³In this

work, Fe(^IV)O serves to activate C–H bonds in styrene,

leading to styrene oxide or benzaldehyde. Pseudo first-order

reaction rates normalized by initial catalyst mass loadings

and by moles of metal over MIL-101(Fe) are indeed higher

than those over MIL-101(Cr), which suggests that oxophilic

Cr leads to more high valent Cr(^IV)O species that are less

active overall for oxidation compared to formed Fe(^IV)O.

Superoxide/hydroperoxo species contribute to higher styrene

oxide selectivities while hydroxyl radicals yield carbonyl

species like benzaldehyde in acetonitrile.⁴⁴This suggests that

Cr–OOH, stabilized in comparison to Fe–OOH due to the

lower electronegativity of Cr, is largely responsible for the

higher styrene oxide selectivity observed for MIL-101(Cr)

compared to that for MIL-101(Fe) since this species is more

adept at C–H and O–H activation compared to Cr(^IV)O.⁵⁵

Further, more electrophilic M–OOH intermediates reportedly

confer greater epoxide selectivity,⁴⁹indicating that the

electrophilic nature of Cr–OOH is greater than that of Fe–

OOH, leading to enhanced epoxide formation.

and represents

a

branch point involving open-site

metallocycle transition states (Scheme 1).^46,47It is not clear

from literature whether styrene oxide and benzaldehyde

branch from the same 4-member metallocycle transition state

(‡₁),^17,44or if they originate from distinct 4- and 5-member

ring transition states (‡₁and ‡₂, respectively).⁴⁷We note that

there is another conformer of ‡₂wherein styrene interacts

through its β carbon to the aromatic ring instead of the α

carbon as drawn in Scheme 1, but this structure is not well-

reported.

To rationalize the disparate primary product selectivity

between MIL-101(Cr) and MIL-101(Fe) (Fig. 4b and c), we first

explore H₂O₂-derived radical intermediates leading to

metallocycle formation. Metal-bound species that reportedly

form from transition metal interactions with H₂O₂include

−

superoxide (O₂˙)/hydroperoxo (HOO˙) formed from hydrogen

abstraction, oxo (O) formed from oxygen abstraction or the

cleavage of superoxide/hydroperoxo species, and hydroxyl

(HO⁻˙) formed from homolytic O–O bond cleavage, but are

often experimentally difficult to quantify because they are

short-lived.^44,47–51Rates of superoxide/hydroperoxo formation

should be higher on MIL-101(Cr) than MIL-101(Fe) since Cr

has a lower Pauling electronegativity (1.66) compared to Fe

(1.83), which can result in a lower barrier for radical

formation. Transition metal oxo formation is related to

oxophilicity, or the tendency of a metal species to form an

oxide by oxygen abstraction from or hydrolysis of another

species. Oxophilicities (a scale of 0–1) calculated utilizing

density functional theory (DFT) on diatomic metal–oxygen

molecules utilizing sulfur as a reference ligand indicate that

Fe (0.4) is less oxophilic than Cr (0.6),⁵²suggesting that there

is a larger energy barrier associated with forming metal-oxo

intermediates on isolated Fe sites compared to Cr. Finally,

hydroxyl radicals are known to form rapidly through

homolytic O–O bond cleavage of H₂O₂on Fe(II), which acts as

a reducing agent in a Fenton-like reaction scheme,⁴⁸and on

Cr(^III),⁴⁴suggesting similar energy barriers on each metal.

These differences in metal electronegativity and

oxophilicity suggest lower energy barriers for formation of

superoxide/hydroperoxo and oxo on Cr sites compared to on

Beyond primary oxygenate production, Scheme 1 also

denotes that styrene oxide can undergo ring-opening

isomerization or hydration to yield phenylacetaldehyde or

styrene glycol, respectively, while benzaldehyde can react

further via deep oxidation by hydrogen peroxide to form

benzoic acid. Phenylacetaldehyde, formed from the

isomerization of styrene oxide, is not observed due to the

large presence of water in the reaction broth that facilitates

hydration reactions at these reaction conditions. This water

is predominantly from the initial charge of 30% aqueous

hydrogen peroxide in addition to that generated as an

oxidation byproduct, but can also be formed via hydrogen

peroxide decomposition if higher reaction temperatures are

used. Since there is a branch point at primary oxygenate

production after which secondary conversions occur in their

respective branches, we also consider composite pathways for

MIL-101(Cr) and MIL-101(Fe) to understand overall flux

toward each branch. Both analogs have a higher selectivity

for pathways stemming from benzaldehyde than for those

from styrene oxide, but MIL-101(Fe) is at least 10% more

selective than MIL-101(Cr) for benzaldehyde pathways. This

is overwhelmingly apparent by 3600 s, as MIL-101(Fe)

selectively favors benzaldehyde pathways (0.797 0.087) over

styrene oxide pathways (0.203 0.024), whereas MIL-101(Cr)

5288 | Catal. Sci. Technol., 2021, 11, 5282–5296

This journal is © The Royal Society of Chemistry 2021

View Article Online

Catalysis Science & Technology

Paper

does not discern between benzaldehyde pathway products

(0.558 0.128) and styrene oxide pathway products (0.442

0.105) to the same degree. This difference in branch

preference further supports that there are underlying

electronic differences between each metal.

MIL-101(Fe) (12.62 4.35 mmol g_cat⁻¹) is more than double

that for MIL-101(Cr) (5.57 1.66 mmol g_cat⁻¹), consistent with

the tendency of MIL-101(Fe) to facilitate styrene oxide

hydration to styrene glycol, as this pathway is catalyzed by

both Lewis and Brønsted acids. Lewis and Brønsted bases

can also catalyze the hydration of styrene oxide to styrene

glycol. Total basic site density, comprised of postulated

hydroxyl (ρ₄), oxo (ρ₅), and coordinatively unsaturated metal

(ρ₆) species, is comparable between MIL-101(Fe) (1.57 0.59

mmol g_cat⁻¹) and MIL-101(Cr) (1.63 0.52 mmol g_cat⁻¹). Basic

site strengths are also similar between the two analogs. While

these sites further aid in the conversion of styrene oxide to

styrene glycol, the disparity in acid site strength and

abundance appears more important in rationalizing the

higher proclivity of MIL-101(Fe) to hydrate styrene oxide.

Examining the proclivity of each catalyst for further

oxidation of primary products reveals that neither favors deep

oxidation of benzaldehyde to benzoic acid, with

benzaldehyde:benzoic acid fractions of more than 1.4 for

MIL-101(Cr) and MIL-101(Fe) at all relevant time points,

which is consistent with literature that report minor benzoic

acid formation from benzaldehyde.¹⁷Both, however,

eventually prefer hydration of styrene oxide to form styrene

glycol with styrene oxide:styrene glycol fractions of 0.58 for

MIL-101(Cr) and 0 for MIL-101(Fe) at 3600 s. This complete

conversion of styrene oxide to styrene glycol on MIL-101(Fe)

while MIL-101(Cr) is able to stabilize and turnover styrene

oxide is inconsistent with quantitative metal oxophilicity (Cr

= 0.6 and Fe = 0.4),⁵²which suggests MIL-101(Cr) should be

more proficient at activating the epoxide.¹¹

Cycling experiments highlight changes in styrene conversion

and oxygenate selectivity profiles

Stability and reusability are key considerations in the design

of heterogeneous catalysts to not only promote green

chemistry, but also to determine industrial applicability;

thus, we probe catalyst deactivation and reactivation

potential. Here, we describe catalyst deactivation

quantitatively by focusing on single-time point conversion,

oxygenate production rates, and product selectivities to

observe how catalytic activity changes in each progressing

cycle. Fractional styrene conversion (Fig. 5a) decreases with

each cycle for both MIL-101(Fe) and MIL-101(Cr), primarily

due to observed material losses incurred from solid recovery

via centrifugation after each cycle and, to a lesser extent,

from acetonitrile and ethanol reactivation washings. We note

that cycle 1 conversions over fresh material are 60% higher

than temporal conversion points at 3600 s because catalyst

concentrations in the scaled-up (temporal) reaction broths

are lower than those in the smaller batch studies; however,

mass-normalized styrene consumption rates are the same

order of magnitude from both experimental setups. That

being said, calculated instantaneous oxygenate formation

rate trends (Fig. 5b) differ between the two materials. For

MIL-101(Fe), these rates are relatively consistent between

cycles when normalized by recovered catalyst mass, with a

13.2% decrease from cycle 1 (1.97 0.10 mM oxygenates per

Since oxophilicity alone cannot explain differences

between MIL-101(Cr) and MIL-101(Fe) tendencies to hydrate

styrene oxide, the existence of additional sites is explored. In

particular, acid and base densities of as-synthesized samples

are quantified via temperature programmed desorption

(TPD) of NH₃and CO₂, respectively. We note that NH₃and

CO₂do not selectively bind to only Lewis or Brønsted sites,

primarily because both molecules are small and are therefore

able to interact with both sites. However, desorption

temperatures can inform relative site strengths, so these

temperature-resolved site densities are reported along with

total acid and base site densities in Table 1. Acid sites that

we attribute to coordinatively unsaturated Fe(^II) (ρ₂) and Fe(III)

(ρ₃) sites desorb NH₃at higher temperatures and are more

abundant than coordinatively unsaturated Cr(^III) sites (ρ₃).

Distributions of Fe(^II) and Fe(III) sites have been observed in

similar Fe-carboxylate MOFs via Mössbauer spectrometry and

X-ray photoelectron spectroscopy;^56,57these two Fe oxidation

states have reportedly different reactivities for oxidant

activation⁵⁶and hydrocarbon oxidation.^57,58Thus, despite

similar NH₃desorption temperatures for ρ₂and ρ₃(Fig.

S12a†), they are reported separately here, though we note that

multivalency in MIL-101 is neither well-understood nor

irrefutably evidenced.⁵⁹Additional sites that desorb NH₃at

lower temperatures (ρ₁) and are likely associated with

coordinated oxygen species for both catalysts are more

abundant in MIL-101(Fe). Overall, total acid site density for

g_catalysts; 50.1 mg_catalyst

)

to cycle

2

(1.71

0.13 mM

oxygenates per g_catalysts; 37.7 mg_catalyst) and a 10.5% decrease

from cycle 2 to cycle 3 (1.53 0.08 mM oxygenates per g_catalyst

s; 17.0 mg_catalyst). This suggests that there is some site

Table 1 Mass normalized acidic site (ρ₁–ρ₃) and basic site (ρ₄–ρ₆) densities for fresh MIL-101(M) as determined by NH₃-TPD and CO₂-TPD, respectively.

Value in parenthesis is the desorption temperature, determined from the center of curve-fitted peaks

M

ρ₁

ρ₂

ρ₃

ρ_acid,total

ρ₄

ρ₅

ρ₆

ρ_base,total

mmol NH₃

g_cat

mmol CO₂

g_cat

−1

mmol NH₃g_cat⁻¹(K)

Cr 0.68 0.29 (440)

mmol CO₂g_cat(K)

—

4.89 1.63 (544)

5.57 1.66 0.13 0.05 (368) 0.18 0.08 (468) 1.32 0.51 (546) 1.63 0.52

Fe 1.47 0.63 (438) 1.21 0.53 (566) 9.94 4.26 (580) 12.62 4.35 0.07 0.03 (365) 0.13 0.06 (442) 1.37 0.58 (549) 1.57 0.59

This journal is © The Royal Society of Chemistry 2021

Catal. Sci. Technol., 2021, 11, 5282–5296 | 5289

View Article Online

Paper

Catalysis Science & Technology

Fig. 5 (a) Fractional styrene conversion for each cycle (grey: Fe, white: Cr), (b) instantaneous rate of total oxygenate production (grey: Fe, white:

Cr), and fractional oxygenate selectivity (styrene oxide (red), benzaldehyde (blue), styrene glycol (grey), benzoic acid (green)) at 323 K, 3600 s over

(c) MIL-101(Cr) and (d) MIL-101(Fe).

deactivation, but that the nature of sites in MIL-101(Fe) does

not change drastically between catalytic cycles and observed

losses in conversion are largely due to solid transfer losses.

In contrast, MIL-101(Cr) rates increase between cycles with

an 82.7% increase from cycle 1 (0.75 0.05 mM oxygenates

101(Fe) and large increases in rates are observed between

all three cycles for MIL-101(Cr).

Product selectivities for the main products (styrene oxide,

benzaldehyde, styrene glycol, and benzoic acid) at 3600 s are

also recorded for each cycle for both analogs at 323 K

(Fig. 5c and d). It is apparent that fresh (cycle 1) MIL-101(Cr)

and MIL-101(Fe) show different product distributions at 3600

s, consistent with temporal concentration profile trends.

Considering primary products from the initial oxidation

event of styrene, MIL-101(Cr) has a higher selectivity to

per g_catalysts; 49.9 mg_catalyst) to cycle 2 (1.37

0.07 mM

oxygenates per g_catalysts; 25.6 mg_catalyst) and a 32.8% increase

from cycle 2 to cycle 3 (1.82 0.10 mM oxygenates per g_catalyst

s; 13.9 mg_catalyst), despite lower styrene conversions. This

indicates that, in addition to solid losses, there are other

deactivation phenomena present, such as Cr leaching, which

is consistent with the light purple color of the reaction broth

after solid removal.

benzaldehyde (0.368

0.131) over styrene oxide (0.102

0.035) at 3600 s. At the same time point, MIL-101(Fe) also

prefers benzaldehyde as its majority primary product (0.502

0.037), but no styrene oxide is observed. The observed

proclivity of both analogs to generate benzaldehyde as the

majority oxygenate by 3600 s is consistent with other studies

at mild temperatures below 373 K that observe benzaldehyde

as the major oxygenate product from styrene oxidation by

hydrogen peroxide over transition metal catalysts, including

Cr- and Fe-based materials.^17,61,62

For overall process sustainability, it is paramount to

understand reactivity and selectivity not only as functions of

chemical and electronic properties, but also as functions of

their evolution in several reaction cycles. After the first cycle

at 323 K, reactivated MIL-101(Cr) and MIL-101(Fe) show

different product distributions compared to one another,

though these differences are subtler in cycles 2 and 3. The

assertion that MIL-101(Fe) site nature does not change

drastically from cycle to cycle is upheld, as the product

profile (benzaldehyde > benzoic acid > styrene glycol) is

maintained within error. In contrast, MIL-101(Cr) shows a

clear change in its product profile, notably in the apparent

loss of its ability to form styrene oxide after cycle 1.

Additionally, the deep oxidation of benzaldehyde to form

benzoic acid becomes favored in cycles 2 and 3, which is not

Further, both analog broths exhibit small concentrations

(10⁻⁴M) of the BDC linker, as ¹H NMR spectroscopy is

sufficiently sensitive for detection at those concentrations.

MIL-101(Fe) only shows evidence of linker leaching in cycle

1, whereas MIL-101(Cr) exhibits leaching in all three cycles

(Fig. S16b†). Linker leaching for MIL-101(Fe) may also

persist for cycles 2 and 3 but exist in concentrations below

¹H NMR detection limits (10⁻⁹M) and are thus

unambiguously lower than linker leaching in MIL-101(Cr).

The mole fraction of leached BDC for the Cr analog is

nearly 1.5 times that of its Fe counterpart in cycle 1. This

BDC is unlikely to be residual starting material left from

synthetic methods, as a singlet at 8.04 ppm in ¹H NMR for

the aromatic hydrogens on BDC is not observed for either

catalyst after 3600 s at 323 K in neat acetonitrile (Fig.

S16a†). Additionally, a peak at 17.7° 2θ (ref. 60) in XRD

diffractograms for crystalline BDC with long-range order is

not observed for fresh MIL-101(Cr) or MIL-101(Fe). Thus,

the detected BDC is reasonably attributed to linker leaching

during the catalytic experiments, which is consistent with

oxygenate production rate trends (Fig. 5b), as the largest

decrease in rates is between cycle 1 and cycle 2 for MIL-

5290 | Catal. Sci. Technol., 2021, 11, 5282–5296

This journal is © The Royal Society of Chemistry 2021

View Article Online

Catalysis Science & Technology

Paper

observed for MIL-101(Fe). These changes coincide with

increased rates, despite decreases in conversion, and Cr

leaching with each cycle. Altogether, these observations

support that the leached Cr species are indeed catalytically

active and have consequences for product selectivity. This is

further confirmed by hot filtration tests (Fig. S17 and S18†)

where solvated Cr complexes remaining in the reaction broth

after MIL-101(Cr) solid removal are allowed to react at 323 K

for another 3600 s; these complexes continue to catalyze

styrene oxidation and affect product distributions by

increasing selectivity to benzoic acid over benzaldehyde and

showing no selectivity for styrene oxide. In contrast, hot

filtration tests over MIL-101(Fe) show negligible changes in

styrene conversion, turnover rates, and product distribution

after solid removal, supporting the conclusion that MIL-

101(Fe) operates as a heterogeneous catalyst (Fig. S19 and

S20†).

that reaction conditions change the nature of the MOF

surfaces. Specifically, the loss of bands assigned to CC and

C–H vibrations suggests that the nature of the BDC backbone

has changed, most likely through involvement in catalytic

turnovers or through interactions with hydrogen peroxide

and not due to the thermal effects, as these bands remained

visible after exposure to neat solvent at reaction temperature

(Fig. S10†). In contrast, bands indicative of O–C–O

functionalities, also belonging to the BDC framework

backbone, remain in both MIL-101(Fe) and MIL-101(Cr),

although the shape of these bands changes. Band-

broadening, like in the case of MIL-101(Cr), or changes in

relative intensity between O–C–O bands, as is the case with

both analogs, suggest interactions between the reactants or

products with the organic backbone that are mitigated to

differing degrees by reactivation procedures. Specifically,

spectra before and after reactivation washings appear similar

to one another but distinctly different from the fresh spectra

for MIL-101(Cr), whereas there is a marked difference in O–

C–O bands for MIL-101(Fe) before and after reactivation. This

suggests that MIL-101(Fe) is more receptive to the reported

reactivation procedures and different strategies need to be

utilized for MIL-101(Cr).

XRD patterns of spent materials after just one cycle (Fig.

S9†) reveal the loss of low angle characteristic peaks for both

MIL-101(Cr) and MIL-101(Fe), possibly implicating severe

pore occlusion, partial pore collapses, and subtle structural

disordering during catalytic turnovers as possible

deactivation mechanisms. A reduction in BET surface areas

after reaction combined with PSD broadening in the

mesoporous range commiserate with a loss of microporosity

(Fig. S8†) implicate the first two phenomena for both MIL-

101(Cr) and MIL-101(Fe). The last phenomena could be

caused by metal or linker leaching or a combination of the

two. Consistent with literature,²⁹we observe that Cr leaching

Characterizing key changes in physicochemical properties

after recycling experiments

To complement catalytic experiments, SEM, XRD, and FT-IR

and UV-vis spectroscopies probe changes in textural and

electronic properties after reaction. Specifically, SEM images

are taken after cycle 3, but before reactivation procedures, to

observe any changes in aggregate morphology after reaction

(Fig. 6a and b). Both materials show no evidence of

catastrophic collapse of aggregate structure and retain some

qualitative semblance to fresh samples (Fig. 2c and d).

FT-IR spectroscopy probes both materials further after

reaction and reactivation and the resulting spectra

(Fig. 6c and d) assert that there are changes in surface

functionality in both materials. Characteristic band

intensities in the fingerprint region at lower wavenumbers

notably decrease after reaction for both analogs, indicating

Fig. 6 SEM images of spent (cycle 3, no reactivation) (a) MIL-101(Cr) and (b) MIL-101(Fe). FT-IR spectra of fresh, spent (cycle 3), and reactivated

(cycle 3) (c) MIL-101(Cr) and (d) MIL-101(Fe).

This journal is © The Royal Society of Chemistry 2021

Catal. Sci. Technol., 2021, 11, 5282–5296 | 5291

View Article Online

Paper

Catalysis Science & Technology

is also exacerbated by interactions with hydrogen peroxide as

evidenced by UV-vis spectra (Fig. S11b†) of the visibly purple

broth after exposure to 10 mL acetonitrile and 30 mmol

aqueous hydrogen peroxide at 323 K for 3600 s in the

absence of styrene. Characteristic diffraction peaks do not

reappear after reactivation procedures for both analogs, but

this may also be due to the small amount of sample (on the

order of 30 mg) remaining after such procedures. We note

that the lack of characteristic peaks in XRD patterns after

reactivation procedures can coexist with FT-IR observations if

oxidative events during reaction induce mild degrees of pore

collapse or restructuring that are not significant enough to

eliminate local structure and detectable functionality in FT-

IR spectra, but are large enough that the distortion of long-

range structure results in an “amorphous” XRD pattern. In

all, neither analog suffered complete structural collapse.

Instead, both show changes in surface functionality of

[Cr(H₂O)₆]³⁺given the large presence of water in the reaction

broth, that have leached from the framework into the broth

during reaction at 323 K (Fig. 7a). The amount of solvated Cr

in the reaction broth increases with each subsequent reaction

cycle, suggesting that defects in the crystal structure caused

by metal leaching in each preceding cycle catalyzes further

degradation. The MIL-101(Fe) broth after reaction appears

clear with a yellow hue, suggesting Fe leaching, which is

consistent with literature.²⁹However, leached Fe species do

not have significant implications for catalysis, as hot

filtration tests (Fig. S19 and S20†) show no further styrene

conversion or changes in product distribution after the solid

is removed. Further, relatively constant mass-normalized

rates from cycling experiments (Fig. 5b) suggest that leached

Fe species from MIL-101(Fe) are not catalytically active.

Overall, material characterization findings are consistent with

catalytic observations and together, show that although both

analogs are active for styrene conversion to form various

oxygenate products, the reactivity of Cr and Fe nodes are

different, as well as the mechanisms by which the two

deactivate and their propensity for reactivation.

differing

reactivation

potential,

where

MIL-101(Fe)

demonstrates a higher susceptibility to this set of reactivation

procedures compared to MIL-101(Cr).

Motivated by the observed changes in surface chemistry,

UV-vis is used to further probe changes in electronics after

reaction. Fresh MIL-101(Fe) (Fig. 7b) has two distinguishable

features: a sharper peak at 290 nm that can be attributed to a

ligand-centered transition of the BDC linker and a broader

feature centered around 400 nm attributed to the ligand to

metal charge transfer (LMCT) band. After reaction at 323 K,

the LMCT band disappears, confirming that the species

associated with this electronic transition, notably the

coordinated Fe center, is involved in catalytic turnovers. This

species shows a degree of reactivation, as washing and drying

procedures described in this work result in the re-emergence

of the LMCT band. This observation complements FT-IR

spectra (Fig. 6d), showing that Fe species can also be

regenerated after reactivation procedures in addition to some

O–C–O functionalities in the linker backbone. A parallel UV-

vis analysis for MIL-101(Cr) could not be performed because

the solid aggregates are too large and do not form a

suspension in acetonitrile or other solvents on the timescale

needed for these measurements. However, liquid-phase UV-

vis confirms the presence of solvated Cr(^III) species,⁶³likely

Styrene oxidation on both ordered and disordered phases of

MIL-101(Cr)

Thus far, we have demonstrated that HF-free hydrothermal

MIL-101(Cr) synthesis yielded catalytically active, albeit

relatively unstable, MIL-101(Cr) that was comparatively

studied with isostructural MIL-101(Fe). However, unlike MIL-

101(Fe), three chromatically distinguishable phases (Fig.

S1a†) were observed upon synthesis: a light green (hex:

#8BA194) phase with the lowest density,

a grey (hex:

#9B928C) intermediate density phase, and an emerald green

(hex: #1F4932) phase with the highest density. Typical yield

of these three phases combined is 48% on a Cr basis. There

are scant reports on the existence of these three phases from

hydrothermal syntheses, let alone their comparative catalytic

efficacies, but acid mineralizing agents such as hydrofluoric

acid and nitric acid, have been shown to produce single MIL-

101 phases.^23,64The highest density phase was the subject

for comparison with MIL-101(Fe) in previous sections, but

Fig. 7 Liquid-phase UV-vis spectra at ambient temperature of (a) reaction broth after MIL-101(Cr) solid removal for each cycle and of (b)

suspended fresh, spent (cycle 3), and reactivated MIL-101(Fe) in acetonitrile.

5292 | Catal. Sci. Technol., 2021, 11, 5282–5296

View Article Online

Catalysis Science & Technology

Paper

Fig. 8 (a) XRD patterns and (b) FTIR spectra of MIL-101(Cr)-ρ_low, Cr-BDC-ρ_med, MIL-101(Cr)-ρ_high. SEM images of (c) Cr-BDC-ρ_medand (d) MIL-

101(Cr)-ρ_low. The SEM image for MIL-101(Cr)-ρ_highat the same magnification is in Fig. 2d.

despite its activity for styrene oxidation, this phase suffers

from Cr leaching. As such, we investigate these three Cr

phases to elucidate textural differences and impacts on

catalytic activity.

MIL-101(Cr)-ρ_high. Cr-BDC-ρ_medparticles (Fig. 8c) deviate

significantly from the structure and size of MIL-101(Cr) and

instead are much larger (1.6 0.2 μm), spherical, and have a

needle-like substructure, reminiscent of yarn balls. From

these textural characterizations, it is apparent that MIL-

101(Cr)-ρ_lowand MIL-101(Cr)-ρ_highshow similar bulk and

surface properties with subtle differences, but both differ

significantly from Cr-BDC-ρ_medand these differences have an

impact on catalytic activity.

XRD patterns (Fig. 8a) show that the lowest density phase,

termed MIL-101(Cr)-ρ_low, exhibits all characteristic MIL-101

peaks compared to the highest density phase, termed MIL-

101(Cr)-ρ_high. The intermediate density phase, however, lacks

most characteristic MIL-101 peaks except for very weak twin

peaks in the 8–9° 2θ range and is appropriately termed Cr-

BDC-ρ_med. Further, there is a peak at 19.8° 2θ, which is not

consistent with MIL-101, Cr oxides, or Cr carbides, that

provides support for the hypothesis that this is a unique

phase.⁶⁴This unique phase (Cr-BDC-ρ_med) is postulated to be

a transitionary phase between MIL-101 and MIL-53 (a higher

density microporous Cr carboxylate) that forms from the

degradation of MIL-101 at longer synthesis times.⁶⁵Though

rearrangement of these decomposed species forms MIL-53,⁶⁵

Cr-BDC-ρ_meddoes not exhibit any characteristic MIL-53 XRD

peaks (Fig. S14b†), further supporting that this phase is

transitionary.

Single-time point (3600 s, 323 K) experiments followed by

hot filtration leaching tests demonstrate that styrene

conversion by hydrogen peroxide differs among MIL-101(Cr)-

ρ_low

,

Cr-BDC-ρ_med

,

and MIL-101(Cr)-ρ_high

.

We first

acknowledge that the styrene conversion here (Fig. 9) for

MIL-101(Cr)-ρ_highis 41.8% of that in Fig. 5a. This is due to a

lower Cr mole percent in the batch used in this experiment

FT-IR spectra (Fig. 8b) complement XRD patterns, where

all bands for MIL-101(Cr)-ρ_lowand MIL-101(Cr)-ρ_highsuggest

similar surface functionalities characteristic of MIL-101(Cr).

Cr-BDC-ρ_medretains some MIL-101(Cr) functionality,

especially strong O–C–O bands between 1400–1600 cm⁻¹, but

exhibits lower overall transmittance with peak broadening

compared to MIL-101(Cr)ρ_lowand MIL-101(Cr)-ρ_high

,

supporting its transitionary, disordered structure. From SEM

images, MIL-101(Cr)-ρ_low(Fig. 8d) is comprised of octahedral

crystals with an average diameter of 0.32 0.05 μm, which is

comparable in size to those observed in MIL-101(Cr)-ρ_high

(0.31 0.09 μm), though the higher density MIL-101 crystals

are qualitatively more irregular in shape, quantitatively

reflected in a wider particle size distribution (Fig. S4†) for

Fig. 9 Styrene conversion for MIL-101(Cr)-ρ_high(white), Cr-BDC-ρ_med

(light grey), and MIL-101(Cr)-ρ_low(dark grey) at 3600 s (1 h), and 323 K.

Styrene conversion before reaction is 0.0039 0.0001.

Catal. Sci. Technol., 2021, 11, 5282–5296 | 5293

View Article Online

Paper

Catalysis Science & Technology

(3.6 at%) compared to the one in Fig. 5a (12.8 at%), though

normalized oxygenate production rates representing a lower

bound where all Cr species are considered active are on the

0.054) but no styrene oxide. This suggests that the

coordination environment around Cr in crystalline MIL-101

affords electronic properties that enable desorption of

styrene oxide rather than complete hydration to styrene

glycol. Overall, we observed three distinct materials

produced from HF-free hydrothermal MIL-101(Cr) synthesis

that demonstrated differing efficacies for styrene oxidation

due in part to Cr trimer defects and different coordination

environments.

same order of magnitude (this batch: 6.30

oxygenates per s mmol Cr, Fig. 5a: 4.23 0.25 mM oxygenates

per mmol Cr). Here, we emphasize that properly

0.36 mM

s

normalized rates can overcome differences in metal content

between syntheses and instead provides insight into the

intrinsic activity of Cr sites in each sample. Notably, the

oxygenate production rate normalized by Cr content for MIL-

101(Cr)-ρ_low(0.19 0.02 mM oxygenates per s mmol Cr) is

lower than those of Cr-BDC-ρ_med(0.96 0.09 mM oxygenates

Conclusion

per s mmol Cr) and MIL-101(Cr)-ρ_high(6.30

0.36 mM

For styrene oxidation by H₂O₂, temporal and single-time

point recyclability experiments assert that compared to MIL-

101(Cr), MIL-101(Fe) is more reactive, recyclable, and pathway

selective, with a higher preference for primary aldehyde

(benzaldehyde) formation over epoxide (styrene oxide)

formation. Reactivity and selectivity differences between MIL-

101(Fe) and MIL-101(Cr) are rooted in relative ability to form

reactive radical intermediates from H₂O₂and open-site

metallocycle transition states. Oxidation by H₂O₂often

implicates reactive surface-bound radical intermediates,

M(^III)–OOH, M(III)–OH, and high valent M(IV)O. Oxophilicity

suggests that oxygen abstraction to form MO is more facile

for Cr, but the higher reactivity of Fe(^IV)O to activate C–H

bonds translates to the higher observed styrene oxidation for

MIL-101(Fe). Meanwhile, a lower electronegativity allows Cr

sites to better stabilize reactive M(^III)–OOH radicals after

initial hydrogen abstraction; the higher electrophilicity of

formed Cr(^III)–OOH confers increased styrene oxide selectivity

compared to MIL-101(Fe). Secondary conversions occur over

both analogs, forming styrene glycol through ring-opening

styrene oxide hydration and benzoic acid through deep

oxidation of benzaldehyde by H₂O₂. The former pathway is

more prevalent for MIL-101(Fe) than for MIL-101(Cr),

consistent with a higher acid site density on MIL-101(Fe).

Catalyst recyclability experiments complemented by various

characterization techniques determine that MIL-101(Fe) sites

are more regenerable utilizing the mild reactivation

procedures reported in this work whereas MIL-101(Cr) suffers

from metal leaching that impacts oxygenate production rates

and selectivities and is not regenerable to a significant

degree. Lastly, styrene turnover rates for three chromatically

unique phases of MIL-101(Cr) obtained for a single synthesis

suggest that Cr-trimer defects resulting in open active sites

contained in MIL-101 crystalline pores are largely responsible

for observed reactivity and selectivity, but increase framework

susceptibility to deactivation via metal or linker leaching.

Overall, this work rigorously quantifies differing material

stabilities, reactivities, and oxygenate selectivities between

isostructural MIL-101(Cr) and MIL-101(Fe) as a function of

their structural and electronic property differences and the

fundamental understanding of catalyst behavior and

limitations defined here can be applied to future catalyst

design and to other hydrocarbon oxidation reactions that are

ubiquitous in modern chemistry industries.

oxygenates per s mmol Cr) with styrene conversions over

MIL-101(Cr)-ρ_lowbeing an order of magnitude lower than

those of Cr-BDC-ρ_medand MIL-101(Cr)-ρ_high. This is not an

intuitive result in the case of MIL-101(Cr)-ρ_lowand MIL-

101(Cr)-ρ_high, as textural characterization suggests that they

have similar properties. However, if we consider elemental

composition, MIL-101(Cr)-ρ_lowhas

a

higher Cr mole

percentage (6.0 at%) compared to MIL-101(Cr)-ρ_high(3.6 at%),

the former being within error compared to the theoretical 5

at% Cr (7 at% for EDXS that does not detect H) for the

nominal chemical formula of [Cr₃(O)(OH)(BDC)₃(H₂O)₂] for

defect-free MIL-101(Cr). This, together with lower styrene

conversion and oxygenate production rates over MIL-101(Cr)-

ρ_low, suggests that MIL-101(Cr)-ρ_highcontains a higher density

of active Cr species, potentially from defects from incomplete

Cr trimers. Enhanced adsorption and catalytic activity on

open active sites have been observed for other transition

metal carboxylate MOFs with defects caused by missing

linkers, metal sites, or both.^66–68Defects are further

implicated by the higher styrene conversion exhibited by the

disordered Cr-BDC-ρ_medcompared to MIL-101(Cr)-ρ_low. We

note that the defects in fresh MIL-101(Cr)-ρ_highare randomly

distributed and are scarce enough not to impact overall long-

range crystal structure or functionality assessed by XRD and

FT-IR, respectively. Further, defect density is challenging to

control and quantify, and this is reflected by the varying Cr

mole fractions of MIL-101(Cr)-ρ_highbatches mentioned

earlier. Cr-trimer defects are further supported by the

observation of a purple reaction broth (Fig. S11a†), indicating

leached Cr species, after 3600 s of reaction for MIL-101(Cr)-

ρ_highand Cr-BDC-ρ_med, but not for MIL-101(Cr)-ρ_low. These

leached species are again confirmed to be active for styrene

oxidation with a 114.5% increase in styrene conversion for

MIL-101(Cr)-ρ_highand a 66.1% increase for Cr-BDC-ρ_medafter

hot filtration.

Though defects are postulated to increase styrene

conversion, Cr-BDC-ρ_medand MIL-101(Cr)-ρ_highshow

different product distributions, indicating that the MIL-101

crystalline structure has important implications for

oxygenate selectivity. Styrene oxide production is observed

over MIL-101(Cr)-ρ_highbut not over disordered Cr-BDC-ρ_med

which produces benzaldehyde (0.023 0.005), styrene glycol

(major product, 0.747 0.175), and benzoic acid (0.230

,

5294 | Catal. Sci. Technol., 2021, 11, 5282–5296

View Article Online

Catalysis Science & Technology

Paper

Conflicts of interest

19 B. Ma, W. Zhao, F. Zhang, Y. Zhang, S. Wu and Y. Ding, RSC

Adv., 2014, 4, 32054–32062.

20 Y. Ding, B. Ma, D. Tong, H. Hua and W. Zhao, Aust. J. Chem.,

2009, 62, 739–746.

21 Y. Ding, W. Zhao, H. Hua and B. Ma, Green Chem., 2008, 10,

910–913.

22 A. Dhakshinamoorthy, M. Alvaro and H. Garcia, Catal. Sci.

Technol., 2011, 1, 856–867.

23 C. Férey, C. Mellot-Draznieks, C. Serre, F. Millange, J.

Dutour, S. Surblé and I. Margiolaki, Science, 2005, 309,

2040–2042.

24 O. A. Kholdeeva, Catal. Today, 2016, 278, 22–29.

25 I. D. Ivanchikova, I. Y. Skobelev and O. A. Kholdeeva,

J. Organomet. Chem., 2015, 793, 175–181.

There are no conflicts to declare.

Acknowledgements

This work was supported by start-up funds from Princeton

University. The authors acknowledge the use of Princeton's

Imaging and Analysis Center, which is partially supported

through the Princeton Center for Complex Materials (PCCM),

a National Science Foundation (NSF)-MRSEC program (DMR-

2011750). The authors also appreciate access to the

Department of Chemistry's NMR facility at Princeton that has

been instrumental to this work. Finally, the authors also

thank Matthew Ko for his assistance in the synthetic efforts

to yield MIL-101(Cr).

26 F. Farzaneh and Y. Sadeghi, J. Mol. Catal. A: Chem.,

2015, 398, 275–281.

27 D. E. Kravchenko, I. A. Tyablikov, P. A. Kots, B. A.

Kolozhvari, D. A. Fedosov and I. I. Ivanova, Pet. Chem.,

2018, 58, 1255–1262.

28 O. A. Kholdeeva, I. Y. Skobelev, I. D. Ivanchikova, K. A.

Kovalenko, V. P. Fedin and A. B. Sorokin, Catal. Today,

2014, 238, 54–61.

29 O. Kholdeeva and N. Maksimchuk, Catalysts, 2021, 11, 283.

30 L. Bromberg, Y. Diao, H. Wu, S. A. Speakman and T. A.

Hatton, Chem. Mater., 2012, 24, 1664–1675.

31 D. A. Gómez-Gualdrón, P. Z. Moghadam, J. T. Hupp, O. K.

Farha and R. Q. Snurr, J. Am. Chem. Soc., 2016, 138,

215–224.

References

1 Y. Lin, C. Kong and L. Chen, RSC Adv., 2012, 2, 6417–6419.

2 J. H. Wang, M. Li and D. Li, Chem. – Eur. J., 2014, 20,

12004–12008.

3 P. Horcajada, C. Serre, G. Maurin, N. A. Ramsahye, F. Balas,

M. Vallet-Regí, M. Sebban, F. Taulelle and G. Férey, J. Am.

Chem. Soc., 2008, 130, 6774–6780.

4 Z. Hu, B. J. Deibert and J. Li, Chem. Soc. Rev., 2014, 43,

5815–5840.

5 D. Sun, Q. Deng and J. Long, J. Solid State Electrochem.,

2018, 22, 487–493.

6 Y. Cui, J. Zhang, H. He and G. Qian, Chem. Soc. Rev.,

2018, 47, 5740–5785.

32 H. Hu, H. Zhang, Y. Chen and H. Ou, Environ. Sci. Pollut.

Res., 2019, 101, 24720–24732.

7 A. Corma, H. García and F. X. Llabrés I Xamena, Chem. Rev.,

2010, 110, 4606–4655.

33 A. D. Yeganeh, M. M. Amini and N. Safari, J. Porphyrins

Phthalocyanines, 2019, 23, 1118–1131.

8 K. Leng, Y. Sun, X. Li, S. Sun and W. Xu, Cryst. Growth Des.,

2016, 16, 1168–1171.

34 Q. Liu, L. Ning, S. Zheng, M. Tao, Y. Shi and Y. He, Sci. Rep.,

2013, 3, 1–6.

9 A. Dhakshinamoorthy, A. M. Asiri and H. Garcia, Chem. –

Eur. J., 2016, 22, 8012–8024.

35 E. Rahmani and M. Rahmani, Front. Chem. Sci. Eng.,

2020, 14, 1100–1111.

10 L. Ma, C. Abney and W. Lin, Chem. Soc. Rev., 2009, 38,

1248–1256.

36 C. S. Inc., NIST Chemistry WebBook, NIST Standard Reference

Database, National Institute of Standards and Technology,

Gaithersburg, MD, 20899, Number 69, 2018.

37 J. Ma, A. P. Kalenak, A. G. Wong-foy and A. J. Matzger,

Angew. Chem., Int. Ed., 2017, 56, 14618–14621.

38 J. E. Mondloch, O. Karagiaridi, O. K. Farha and J. T. Hupp,

CrystEngComm, 2013, 15, 9258–9264.

11 B. R. James, J. A. Boissonnault, A. G. Wong-Foy, A. J. Matzger

and M. S. Sanford, RSC Adv., 2018, 8, 2132–2137.

12 O. V. Zalomaeva, A. M. Chibiryaev, K. A. Kovalenko, O. A.

Kholdeeva, B. S. Balzhinimaev and V. P. Fedin, J. Catal.,

2013, 298, 179–185.

13 J. Kim, S. N. Kim, H. G. Jang, G. Seo and W. S. Ahn, Appl.

Catal., A, 2013, 453, 175–180.

14 P. Horcajada, S. Surblé, C. Serre, D. Y. Hong, Y. K. Seo, J. S.

Chang, J. M. Grenèche, I. Margiolaki and G. Férey, Chem.

Commun., 2007, 2820–2822.

39 A. J. Howarth, A. W. Peters, N. A. Vermeulen, T. C.

Wang, J. T. Hupp and O. K. Farha, Chem. Mater.,

2017, 29, 26–39.

40 G. Han, Q. Qian, K. Mizrahi Rodriguez and Z. P. Smith, Ind.

Eng. Chem. Res., 2020, 59, 7888–7900.

15 J. Tang, M. Yang, M. Yang, J. Wang, W. Dong and G. Wang,

New J. Chem., 2015, 39, 4919–4923.

16 I. Y. Skobelev, A. B. Sorokin, K. A. Kovalenko, V. P. Fedin and

O. A. Kholdeeva, J. Catal., 2013, 298, 61–69.

17 J. Sun, G. Yu, Q. Huo, Q. Kan and J. Guan, RSC Adv., 2014, 4,

38048–38054.

41 F. R. Mayo, J. Am. Chem. Soc., 1953, 75, 6133–6141.

42 K. S. Khuong, W. H. Jones, W. A. Pryor and K. N. Houk,

J. Am. Chem. Soc., 2005, 127, 1265–1277.

43 T. A. G. Duarte, A. C. Estrada, M. M. Q. Simões, I. C. M. S.

Santos, A. M. V. Cavaleiro, M. G. P. M. S. Neves and J. A. S.

Cavaleiro, Catal. Sci. Technol., 2015, 5, 351–363.

44 H. Chen, W. Wang, Y. Yang, P. Jiang, W. Gao, R. Cong and T.

Yang, Appl. Catal., A, 2019, 588, 117283.

18 N. V. Maksimchuk, K. A. Kovalenko, V. P. Fedin and O. A.

Kholdeeva, Adv. Synth. Catal., 2010, 352, 2943–2948.

Catal. Sci. Technol., 2021, 11, 5282–5296 | 5295

View Article Online

Paper

Catalysis Science & Technology

45 V. Hulea and E. Dumitriu, Appl. Catal., A, 2004, 277, 99–106.

46 A. K. Rappé and W. A. Goddard, Nature, 1980, 285, 311–312.

47 Q. Wang, X. Liang, R. Bi, Y. Liu, Y. He, J. Feng and D. Li,

Dalton Trans., 2019, 48, 16402–16411.

48 J. J. Pignatello, E. Oliveros and A. MacKay, Crit. Rev. Environ.

Sci. Technol., 2006, 36, 1–84.

57 A. Dhakshinamoorthy, M. Alvaro, P. Horcajada, E. Gibson,

M. Vishnuvarthan, A. Vimont, J. M. Grenèche, C. Serre, M.

Daturi and H. Garcia, ACS Catal., 2012, 2, 2060–2065.

58 J. N. Hall and P. Bollini, Chem.

16639–16643.

– Eur. J., 2020, 26,

59 J. N. Hall and P. Bollini, React. Chem. Eng., 2019, 4, 207–222.

60 E. Haque, N. A. Khan, J. E. Lee and S. H. Jhung, Chem. – Eur.

J., 2009, 15, 11730–11736.

49 D. T. Bregante, N. E. Thornburg, J. M. Notestein and D. W.

Flaherty, ACS Catal., 2018, 8, 2995–3010.

50 D. T. Bregante, J. Z. Tan, R. L. Schultz, E. Z. Ayla, D. S. Potts,

C. Torres and D. W. Flaherty, ACS Catal., 2020, 10,

10169–10184.

51 A. S. Petit, R. C. R. Pennifold and J. N. Harvey, Inorg. Chem.,

2014, 53, 6473–6481.

52 K. P. Kepp, Inorg. Chem., 2016, 55, 9461–9470.

53 A. S. Rosen, J. M. Notestein and R. Q. Snurr, Angew. Chem.,

2020, 132, 19662–19670.

54 M. C. Simons, J. G. Vitillo, M. Babucci, A. S. Hoffman, A.

Boubnov, M. L. Beauvais, Z. Chen, C. J. Cramer, K. W.

Chapman, S. R. Bare, B. C. Gates, C. C. Lu, L. Gagliardi and

A. Bhan, J. Am. Chem. Soc., 2019, 141, 18142–18151.

55 J. Cho, J. Woo, J. Eun Han, M. Kubo, T. Ogura and W. Nam,

Chem. Sci., 2011, 2, 2057–2062.

61 W. Tanglumlert, T. Imae, T. J. White and S. Wongkasemjit,

Catal. Commun., 2009, 10, 1070–1073.

62 D. R. Das, P. Kalita and A. K. Talukdar, J. Porous Mater.,

2020, 27, 893–903.

63 A. Mohammadi, M. M. Doroodmand and M. M. R. Sadegh,

J. Text. Sci. Eng., 2016, 6, 1–6.

64 T. Zhao, F. Jeremias, I. Boldog, B. Nguyen, S. K. Henninger

and C. Janiak, Dalton Trans., 2015, 44, 16791–16801.

65 N. A. Khan and S. H. Jhung, Cryst. Growth Des., 2010, 10,

1860–1865.

66 R. S. Forgan, Chem. Sci., 2020, 11, 4546–4562.

67 H. Guo, S. Su, Y. Liu, X. Ren and W. Guo, Environ. Sci. Pollut.

Res., 2020, 27, 17194–17204.

68 J. Wang, L. Liu, C. Chen, X. Dong, Q. Wang, L. Alfilfil, M. R.

Alalouni, K. Yao, J. Huang, D. Zhang and Y. Han, J. Mater.

Chem. A, 2020, 8, 4464–4472.

56 M. Pu, Y. Ma, J. Wan, Y. Wang, J. Wang and M. L. Brusseau,

Catal. Sci. Technol., 2017, 7, 1129–1140.

5296 | Catal. Sci. Technol., 2021, 11, 5282–5296

Article Doi

Reaction pathways and deactivation mechanisms of isostructural Cr and Fe MIL-101 during liquid-phase styrene oxidation by hydrogen peroxide

DOI: 10.1039/d1cy00567g

Source and publish data:

Authors:

Article abstract of DOI:10.1039/d1cy00567g

Full text of DOI:10.1039/d1cy00567g

Products guided by the article

Relevant to this article

Hot Product