Engineering Second Sphere Interactions in a Host-Guest Multicomponent Catalyst System for the Hydrogenation of Carbon Dioxide to Methanol

Thomas M. Rayder, Adam T. Bensalah, Banruo Li, Je ﬀ ery A. Byers,* and Chia-Kuang Tsung *

Article

Engineering Second Sphere Interactions in a Host−Guest

Multicomponent Catalyst System for the Hydrogenation of Carbon

Dioxide to Methanol

Cite This: J. Am. Chem. Soc. 2021, 143, 1630 −1640

Read Online

ACCESS

Metrics & More

Article Recommendations

sı Supporting Information

ABSTRACT: Many enzymes utilize interactions extending

beyond the primary coordination sphere to enhance catalyst

activity and/or selectivity. Such interactions could improve the

eﬃcacy of synthetic catalyst systems, but the supramolecular

assemblies employed by biology to incorporate second sphere

interactions are challenging to replicate in synthetic catalysts.

Herein, a strategy is reported for eﬃciently manipulating outer-

sphere inﬂuence on catalyst reactivity by modulating host−guest

interactions between a noncovalently encapsulated transition-

metal-based catalyst guest and a metal−organic framework

(

MOF) host. This composite consists of a ruthenium PNP pincer

complex encapsulated in the MOF UiO-66 that is used in tandem

with the zirconium oxide nodes of UiO-66 and a ruthenium PNN

pincer complex to hydrogenate carbon dioxide to methanol. Due to the method used to incorporate the complexes in UiO-66,

structure−activity relationships could be eﬃciently determined using a variety of functionalized UiO-66-X hosts. These

investigations uncovered the beneﬁcial eﬀects of the ammonium functional group (i.e., UiO-66-NH ). Mechanistic experiments

revealed that the ammonium functionality improved eﬃciency in the hydrogenation of carbon dioxide to formic acid, the ﬁrst step in

the cascade. Isotope eﬀects and structure−activity relationships suggested that the primary role of the ammonium functionality is to

serve as a general Brønsted acid. Importantly, the cooperative inﬂuence from the host was eﬀective only with the functional group in

close proximity to the encapsulated catalyst. Reactions carried out in the presence of molecular sieves to remove water highlighted

the beneﬁcial eﬀects of the ammonium functional group in UiO-66-NH₃⁺and resulted in a 4-fold increase in activity. As a result of

the modular nature of the catalyst system, the highest reported turnover number (TON) (19 000) and turnover frequency (TOF)

−

(

9100 h ) for the hydrogenation of carbon dioxide to methanol are obtained. Moreover, the reaction was readily recyclable, leading

to a cumulative TON of 100 000 after 10 reaction cycles.

INTRODUCTION

assisted processes have led to the conversion of CO to

4,5 6,7

■

methanol at low pressures, low temperatures, and with high

The high activity and selectivity of catalysis in natural systems

frequently inspire the development of synthetic catalysts. For

activity. Despite this tremendous achievement, high activity in

these reactions requires super-stoichiometric amounts of amine

additives. In contrast, Leitner and co-workers recently reported a

catalytic system that is active in the absence of an additive, albeit

with catalysis that proceeds with lower turnover than those that

instance, biological systems often evolve complex networks that

employ supramolecular assemblies to traﬃc substrates between

isolated active sites. This strategy allows biological systems to

overcome inherent limitations to selectivity, reactivity, and

compatibility. Host−guest catalyst systems are an attractive way

for synthetic chemists to mimic the beneﬁcial eﬀects that

require amine additives. Most closely related to the catalytic

system reported herein are those that use a cascade of chemical

reactions to hydrogenate carbon dioxide to methanol through

enzymatic reactions enjoy.

11,12

formic acid and formate ester intermediates.

Sanford and co-

For example, we previously reported a bioinspired catalytic

system that used a metal−organic framework (MOF) to isolate

multiple catalytic components for the hydrogenation of carbon

dioxide to methanol that proceeds by a cascade of chemical

Received: August 19, 2020

Published: January 19, 2021

reactions. Numerous recent reports have focused on this

transformation, largely because of its promise in converting a

gaseous byproduct of fossil fuel combustion to a potentially

carbon-neutral liquid fuel. For example, advances in amine-

2021 American Chemical Society

Journal of the American Chemical Society

J. Am. Chem. Soc. 2021, 143, 1630−1640

630

Article

Figure 1. Catalysts that use second-sphere interactions to enhance catalytic performance: (A) the biological conversion of carbon dioxide to the

glucose precursor glyceraldehyde-3-phosphate by the enzyme RuBisCo beneﬁts from second-sphere interactions between the bound carbon dioxide

substrate and an amine functionality on a nearby amino acid residue and (B) a synthetic multicomponent host−guest catalyst system reported here for

the hydrogenation of carbon dioxide to methanol that beneﬁts from an ammonium functionality installed in close proximity to a transition metal

complex encapsulated in a metal−organic framework (MOF). In this system, the green ruthenium complex 1@UiO-66-NH carries out the

hydrogenation of carbon dioxide to formic acid, the blue Zr (OH) O nodes of the MOF converts formic acid to formate ester, and the orange 2@

UiO-66 carries out the hydrogenation of formate ester to methanol.

workers originally described a system in which separation of

incompatible catalysts increased its activity by an order of

Our inspiration for these investigations is the sophisticated

protein superstructures that dictate speciﬁc interactions

between the catalytically active sites in enzymes and their

substrates. Such interactions signiﬁcantly improve activity and/

or selectivity in enzymatic reactions (Figure 1A). Synthetic

chemists have developed excellent ways to control direct, inner-

sphere interactions between catalysts and substrates, but

magnitude from a TON of 2.5 to 21. More recently, Goldberg

and co-workers improved upon this system by using more stable

catalyst precursors so as to improve the TON of the cascade an

order of magnitude to 428. We reported a system built on this

foundation that made use of a porous framework to separate

incompatible catalysts, leading to another order of magnitude

manipulating indirect, outer-sphere interactions has not been

14−16

increase in activity. The reaction proceeded eﬃciently at

as extensively explored.

While recent reports have shown

temperatures lower than had previously been described and

without the need for super-stoichiometric additives. While this

system demonstrated activity and catalyst lifetimes that rivaled

that outer-sphere interactions can be introduced through ligand

17−19

design to enhance performance in molecular catalysts,

this

strategy is sometimes synthetically demanding. Moreover, the

geometric constraints incurred with covalent assemblies

designed to position the second-sphere element in close

proximity to the active site restrict catalyst mobility, which

may limit the beneﬁcial eﬀects from second-sphere interactions.

Alternatively, host−guest systems have provided a more

modular approach for engineering outer-sphere interactions,

and host−guest systems that do not feature covalent interactions

between the host and the guest provide more degrees of freedom

the best amine assisted reactions reported in the literature, we

recognized that the host−guest catalyst system was amenable

toward further manipulations that can capitalize on properties

common to biological systems, such as the inﬂuence from outer-

sphere interactions. In this work, we use this MOF-based system

as a platform to explore these additional strategies to enhance

the catalytic performance in the cascade hydrogenation of

carbon dioxide to methanol.

631

J. Am. Chem. Soc. 2021, 143, 1630−1640

Journal of the American Chemical Society

Article

for the guest catalyst to interact with the host. Promising host−

RESULTS AND DISCUSSION

■

guest systems have been developed using supramolecular

To test the hypothesis that catalyst activity for the cascade

hydrogenation of carbon dioxide to methanol could be aﬀected

by outer sphere inﬂuences arising from interactions between one

of the two transition metal complexes and the MOF, we ﬁrst

encapsulated 1 in various UiO-66 derivatives using our aperture-

opening approach. In addition to unfunctionalized UiO-66,

monosubstituted UiO-66-X frameworks (X = CH , F, Br, NO ,

0,21

22−24

cages

and zeolites

that have shown enhanced activity

and selectivity for a variety of reactions catalyzed by transition

metal complex-based catalysts. However, systematic engineering

of the hosts for further improvement in performance is often

synthetically demanding: modifying supramolecular cages

requires lengthy bottom-up synthesis of new organic ligands,

and modiﬁcation of zeolites is often diﬃcult due to the harsh

NH , and NH ) and a perﬂuorinated variant of UiO-66 (X =

conditions required to modify their exceptionally stable

4F) were assessed as host materials. Initial evaluation of the 1@

UiO-66-X constructs was carried out in the cascade hydro-

genation of carbon dioxide to methanol using the ester

hydrogenation catalyst 2 in solution (Figure 2 , red dots).

structures. The absence of these restrictions makes our host

of choice, MOFs, an attractive platform for systematic

engineering of outer-sphere interactions.

MOFs have been reported with a variety of functionalities

comprising the library of nodes and linkers, some of which can

27−32

also be modiﬁed postsynthetically.

Furthermore, we have

recently developed an “aperture-opening” encapsulation

method that permits the encapsulation of various guest

molecules into MOFs without speciﬁc electrostatic or covalent

,33−35

interactions between the host and guest.

Since the

encapsulation method is solvent dependent, catalytic reactions

can be carried out in solvents where “aperture opening” does not

occur so as to prevent leaching of the guest. Because guest and

host syntheses are decoupled, such an encapsulation method

grants access to a wide range of host−guest combinations. Due

to this advantage, and encouraged by recent reports from

Shustova and co-workers detailing the eﬀects of outer-sphere

interactions on ﬂuorescent guests, we recently investigated

how the ﬂuorescence properties of a dye are aﬀected by its

encapsulation into UiO-66-X MOFs containing diﬀerent

functionalities. Using a variety of physical organic techniques,

we discovered that the dye’s ﬂuorescence behavior was dictated

by an intimate interaction between the MOF host, solvent, and

guest.

Figure 2. Eﬀect of host functionality on cascade hydrogenation of

carbon dioxide to methanol using 2,2,2-triﬂuoroethanol (red, dots) or

ethanol (blue, stripes) as an alcohol additive. Turnover number (TON)

calculated relative to 1 as an average of three independent runs.

Excited by these observations, we pondered whether similar

outer sphere eﬀects could be translated to the aforementioned

MOF-based catalytic system. The tandem catalytic system

included a ruthenium complex 1 encapsulated in the Lewis

acidic MOF UiO-66, which converted carbon dioxide and

hydrogen to formate esters via formic acid. When used in

conjunction with a second ruthenium complex 2, formate ester

reduction occurred leading to methanol production from carbon

dioxide and hydrogen in a single reaction vessel (Figure 1B).

Methanol was only observed in these cascade hydrogenation

reactions when the two transition metal complexes were

separated from one another, illustrating the importance of

component separation for eﬃcient catalysis. In this work, we

describe how diﬀerent UiO-66-X host materials aﬀect catalytic

While most variants of UiO-66 revealed similar turnover

numbers as observed with unfunctionalized UiO-66, an increase

in activity was observed for UiO-66-NH . Even higher activity

was observed with 1@UiO-66-NH . Since 1@UiO-66-NH

was obtained by protonating UiO-66-NH , we entertained the

possibility that the higher activity of 1@UiO-66-NH could be

a consequence of incomplete protonation of 1@UiO-66-NH₂.

However, titration experiments revealed that all amine linkers

within 1 @UiO-66-NH₃⁺were protonated (see Figure S1 and

Table S1 ), which ruled out the possibility for cooperative eﬀects

between protonated and unprotonated amines as the culprit for

the higher activity observed for 1@UiO-66-NH . Analysis of

@UiO-66-NH and 1@UiO-66-NH revealed no change in

performance. Our results uncovered outer-sphere inﬂuence

crystallinity or morphology during encapsulation or acid

treatment. Moreover, a similar particle size and distribution

were observed for these MOFs compared to UiO-66, which

further supported that the higher activity observed using these

MOFs was not due to particle size eﬀects (Figures S2 and S3 ).

from the functionalized host UiO-66-NH on the encapsulated

carbon dioxide hydrogenation catalyst 1 that was used in

conjunction with 2. Mechanistic experiments revealed that

conﬁnement of the catalyst was necessary to observe these outer-

sphere eﬀects, and they also indicated that the role of the

ammonium functionality is to facilitate the ﬁrst step of the

reaction, the hydrogenation of carbon dioxide to formic acid.

Further optimization highlighted the importance of removing

water from these reactions, which led to a catalyst system with

the highest reported turnover number (TON) and turnover

frequency (TOF) for the hydrogenation of carbon dioxide to

Due to the positive inﬂuence of basic (−NH ) and acidic

(

−NH ) functionality on catalytic turnover, we carried out

additional experiments to further understand the origin of the

high turnover observed with these substituents. One possibility

was that the basic amine functionality was protonated in situ by

the acidic 2,2,2-triﬂuoroethanol (TFE) additive (pK = 12.4),

meaning that the functional group leading to an increase in

methanol. Recycling of the catalyst yielded an unprecedented

cumulative TON of 100 000.

activity in both cases was actually −NH . To determine if this

was the case, ethanol, a less acidic additive (pK = 15.9), was

632

J. Am. Chem. Soc. 2021, 143, 1630−1640

Journal of the American Chemical Society

Article

used in the catalytic reactions in place of TFE (Figure 2, blue). In

these reactions, the increase in TON that had been observed for

@UiO-66-NH in the presence of TFE was no longer detected,

while a signiﬁcant activity increase persisted for 1@UiO-66-

NH . These results conﬁrmed that the activity increase resulted

from the inﬂuence of the positively charged ammonium

functional group (−NH ) on catalysis.

Although the beneﬁcial eﬀect of a protonated amine was

apparent from the above experiments, it was unclear whether

this inﬂuence was from the Brønsted acidity of the ammonium

functional group or from its electron-withdrawing character-

istics. It is noteworthy that other UiO-66-X variants with

electron-withdrawing groups (e.g., X = F, 4F, Br, and NO ) did

not demonstrate any beneﬁcial eﬀect on catalyst turnover

(Figure 2 and Figure S4 ). These results suggested that it is the

Brønsted acidity of the ammonium functionality that led to its

beneﬁcial eﬀects. Further supporting this hypothesis were

reactions carried out in UiO-66-X hosts containing primary (X

(

NH /NH ), secondary (X = NMeH/NMeH ), and tertiary

X = NMe /NMe H ) amine/ammonium substituents (Table

2 2

). We also tried to synthesize a MOF containing a quaternary

Table 1. Cascade Hydrogenation of Carbon Dioxide to

Methanol in the Presence of Ethanol as an Additive with

Diﬀerent Amine and Ammonium-Based Substitutions on the

Host Framework

Entry

pK /pK

TON

Figure 3. Eﬀect of the size of the MOF cage and external anilinium

additive on turnover for the hydrogenation of carbon dioxide to

methanol. The ruthenium complexes are roughly 13 Å in diameter,

while the diameter of the larger octahedral pores in UiO-66 and UiO-67

^N^H2

9.42

9.15

8.94

−

4.58

4.85

5.06

4.70

6200 ± 500

6100 ± 400

6000 ± 350

6000 ± 400

8900 ± 350

7200 ± 300

6600 ± 400

4600 ± 300

NHMe

NMe₂

are 11−14 Å (depending on the existence of missing linker defects) and

3 Å, respectively. Turnover number (TON) determined relative to

NH₃

NH Me

NHMe₂

ND₃⁺

catalytic reaction, as well as reports that the internal pore

environment in UiO-66 is relatively hydrophilic. Another

piece of evidence that supports the role of the host functional

pK and pK are determined in aqueous solution.

number calculated relative to 1 as an average of three runs with error

reported as average error. Reaction run using EtOD as an additive.

Turnover

^g^r^o^u^p^a^s^a^B^r^øⁿ^s^t^e^d^a^cⁱ^d^w^e^r^e^r^e^a^c^tⁱ^oⁿ^s^r^uⁿ^u^sⁱⁿ^g^Uⁱ^O^-⁶⁶^-^N^D3

as the host (entry 8). If the role of the ammonium functionality

was to serve as an electron-withdrawing group to accelerate

ammonium substituent (i.e., X = NMe ), which does not

contain a Brønsted acidic proton but has similar electron-

catalysis, then reactions run with 1@UiO-66-ND₃would

have the same TON as the natural abundance 1@UiO-66-NH ,

withdrawing eﬀects as the other ammonium substituents.

but a Brønsted acid catalyzed process should demonstrate an

isotope eﬀect. In the event, we found that reactions run using

1@UiO-66-ND₃⁺had lower turnover after 2 h (TON = 4600)

than 1@UiO-66-NH₃⁺(TON = 8900), consistent with a

Unfortunately, our attempts to encapsulate 1 into UiO-66-

NMe were unsuccessful. Encapsulation of 1 in the other three

substituents, however, was successful, which enabled us to

evaluate this series during catalysis. Ethanol was used as an

additive in these studies to ensure that complications from in situ

protonation did not occur. Whereas MOFs substituted with

diﬀerent amines (i.e., X = NH , NHMe, NHMe , entries 1−3)

Brønsted acid facilitated process with an isotope eﬀect of 1.93.

Finally, a decrease in TON was observed as the amount of

ammonium was decreased in partially protonated UiO-66-NH₂

(Table S3). This result suggested that the ammonium

functionality was acting as a general acid catalyst, interacting

directly with a reactive intermediate rather than indirectly as a

speciﬁc acid catalyst though solvent-aided proton transfer.

Our previous studies involving encapsulated dyes indicated

that close proximity of the MOF functional group to the guest is

resulted in similar activity as when unfunctionalized UiO-66 was

used as the host (entry 4), activity decreased monotonously for

the ammonium-functionalized 1@UiO-66-X in the order X =

NH , X = NH Me , and X = NHMe (entries 5−7). This trend

parallels the relative acidity of the anilinium functional groups in

water (pK for PhNH = 4.58, PhNMeH = 4.85, PhNMe H

5.06). This observation is consistent with the ammonium

critical to observe outer-sphere eﬀects. To evaluate whether

this factor was also important in catalysis, two sets of control

experiments were performed (Figure 3). First, reactions were

functional group serving as a Brønsted acid to facilitate the

633

J. Am. Chem. Soc. 2021, 143, 1630−1640

Journal of the American Chemical Society

was in solution (Figure 4 A). No increase in TON was observed

Article

evaluated using 1@UiO-67-X as the host−guest construct. UiO-

the reaction, the hydrogenation of formate esters to methanol,

was not facilitated by the Brønsted acid. To provide further

insight, the established system (1@UiO-66 with 2 in solution)

was compared to a system in which 2 was encapsulated and 1

7 is an isomorph of UiO-66 with larger cages that reduce

proximity between the functional group and catalyst. Consistent

with proximity being a factor, catalytic reactions using 1@UiO-

7-NH (top right) exhibited a lower TON than 1@UiO-66-

NH₃(top left). Although an increase in TON was observed

when compared to 1@UiO-67 (bottom right), this increase in

activity was less signiﬁcant than that observed for 1@UiO-66-

NH₃compared to 1@UiO-66 (middle left). Importantly,

despite their larger pore size, no leaching of the ruthenium

complex was observed in any of the UiO-67 variants as evidence

by the similar ruthenium loading observed in the UiO-67 before

and after catalysis. To ensure that the diﬀerence observed was a

result of the functional group being anchored in close proximity

to the encapsulated catalyst, 2,6-diisopropylanilinium chloride

was tested as an additive in conjunction with 1@UiO-66

(

bottom left). This additive resulted in no increase in turnover

even at elevated concentrations, a likely consequence of the

anilinium’s steric bulk, which prevented its interaction with the

encapsulated CO hydrogenation catalyst 1. Overall, the above

results underline the importance of proximity between the

encapsulated catalyst and ammonium functionality, which could

allow for more immediate interaction between intermediates

that are formed at the metal center of the encapsulated catalyst

and the functional group and thus enhance catalytic eﬃciency.

The results obtained with the anilinium additive were also

helpful in elucidating which step in the reaction sequence is most

aﬀected by the substitution of 1@UiO-66 with 1@UiO-66-

NH . As mentioned previously, the hypothesized mechanistic

pathway involves three simultaneously occurring catalytic

reactions (Scheme 1). That no increase in TON was observed

for reactions proceeding with 1@UiO-66 in the presence of 2,6-

diisopropylanilinium chloride suggested that the third step of

Scheme 1. Representation of the Simultaneous

Transformations Occurring in the Cascade Hydrogenation of

Carbon Dioxide to Methanol and the Catalysts Required for

Each Step in the Process

Figure 4. Reactions carried out to assess which step in the cascade

hydrogenation of carbon dioxide to methanol is aﬀected most by the

ammonium functional group in UiO-66-NH . (A) Activity in the

cascade reaction with 1 (red) or 2 (blue) as the encapsulated catalyst

with turnover (TON) determined relative to 1. (B) Esteriﬁcation of

formic acid to 2,2,2-triﬂuoroethyl formate using diﬀerently function-

alized UiO-66 derivatives.

in the latter case, consistent with the ammonium functional

group not enhancing the rate of the ester reduction step of the

reaction (Figure S5).

These experiments leave the possibility that the ammonium

functional group positively inﬂuences the ﬁrst and/or second

steps involved in the hydrogenation of carbon dioxide to

methanol: CO₂hydrogenation to formic acid and/or

esteriﬁcation of formic acid (Scheme 1). Various functionalized

634

J. Am. Chem. Soc. 2021, 143, 1630−1640

Journal of the American Chemical Society

Article

derivatives of UiO-66 were tested as catalysts for the

esteriﬁcation of formic acid with TFE to evaluate how this

step is aﬀected by the nature of the MOF used in the cascade

propose that the beneﬁt of the ammonium functional group is to

protonate the formate species so that it can more easily

dissociate from F to form C. Noteworthy is the beneﬁt of amine

additives in carbon dioxide hydrogenation to methanol, which

39,42

reaction (Figure S6).

These experiments revealed that

,5,8

esteriﬁcation reactions proceeded with no observable diﬀerence

has been observed previously.

These eﬀects have been

in reaction rate resulting from changing MOF functionality

justiﬁed by the formation of a carbamate intermediate that

prevents formate binding to the metal center. However,

formation of such a carbamate intermediate is not likely the

origin of the rate enhancements observed here because the

presence of an acidic ammonium functionality is beneﬁcial

rather than a basic amine. Under reaction conditions where the

amine functionality is not being protonated (i.e., using EtOH as

the additive), reaction rates did not increase. Consequently, we

propose a similar but distinct hypothesis for the beneﬁcial role of

(Figure 4B). In addition to ruling this step out as the step most

likely facilitated by the ammonium functionality in the cascade

hydrogenation of carbon dioxide to methanol, the results also

illustrated that the esteriﬁcation was likely not shifting from a

Lewis acid catalyzed process to a Brønsted acid catalyzed

process. This conclusion was based on similar rates observed

for all UiO-66 variants evaluated, including UiO-66-NH and

UiO-66-ND . Since the beneﬁcial eﬀects observed in the

cascade reaction were shown to be a result of Brønsted acid

activation, these results provide additional evidence that the

the ammonium functional group in the hydrogenation of carbon

dioxide to methanol promoted by 1@UiO-66-NH , namely

esteriﬁcation reaction was not aided by the ammonium

protonation of the anionic formate as it is formed. Introducing

the second sphere inﬂuence of the ammonium functionality on

catalysis was possible primarily as a result of the modularity and

stability of the MOF host. The facile aperture opening catalyst

encapsulation method was instrumental in identifying this outer

sphere eﬀect by allowing for eﬃcient screening of various

functionalized MOFs.

functionality in the 1@UiO-66-NH /2 reaction system.

With the other steps ruled out, we hypothesize that the

beneﬁcial eﬀect of the ammonium functionality in the cascade

hydrogenation of carbon dioxide aﬀects the eﬃciency of the ﬁrst

step in the cascade process, the hydrogenation of carbon dioxide

to formic acid. Past research has revealed many mechanistic

features of this step in the cascade hydrogenation of carbon

While the initial activity of the system had been optimized by

modifying the functional groups attached to the host, we

suspected that the partially homogeneous catalyst system would

have a limited lifetime. This expectation was based on our

previous studies, which showed that completely heterogeneous

catalyst systems involving encapsulation of 1 and 2 in MOFs had

lower initial activities but also had elongated catalyst lifetimes

compared to partially heterogeneous catalysts systems involving

45−48

dioxide to methanol.

In the absence of base, the catalytic

cycle is believed to proceed through three steps (Scheme 2):

Scheme 2. Proposed Catalytic Cycle for the Hydrogenation of

Carbon Dioxide to Formic Acid Catalyzed 1, and the Role of

Ammonium Functionality in UiO-66-NH₃⁺To Promote

Catalysis

encapsulation of either 1 or 2. To test whether completely

heterogeneous catalyst systems extended catalyst lifetimes, we

encapsulated both complexes (1 and 2) into MOFs (i.e., 1@

UiO-66-NH /2@UiO-66 and 1@UiO-66/2@UiO-66), and

we compared their performance over time with the analogous

partially homogeneous systems (i.e., 1@UiO-66-NH /2 and

@UiO-66/2, respectively) (Figure 5A). For the partially

homogeneous constructs (dotted lines), 1@UiO-66-NH /2

outperformed 1@UiO-66/2 with the diﬀerence in activity

reﬂected by a higher initial turnover frequency for 1@UiO-66-

NH /2. Consistent with our expectations, the activity of the

partially homogeneous catalysts signiﬁcantly decreased at

extended reaction times, producing very little methanol after

the ﬁrst 4 h. We hypothesized that this result was due to the

deactivation of free complex 2 in solution, which we have

observed previously. Others have also commented on the

instability of similar complexes in solution, which has limited

1,12,45−48

catalyst turnover.

For the fully heterogeneous

constructs (Figure 5A, solid lines), 1@UiO-66-NH /2@UiO-

66 also outperformed 1@UiO-66/2@UiO-66. Furthermore,

although the initial turnover for the fully heterogeneous systems

was lower, they outperformed their partially homogeneous

counterparts at elongated reaction times. This result supported

the hypothesis that the partially homogeneous systems under-

went catalyst decomposition, which could be prevented by

making the reaction completely heterogeneous. Further

veriﬁcation for this supposition was obtained by resubjecting

the catalyst systems to the reaction conditions (Figure 5B).

Whereas the partially homogeneous catalyst systems failed to

produce additional methanol in such experiments (red), the

fully heterogeneous system exhibited almost identical turnover

in the second cycle as the ﬁrst (Figure 5B, blue). These results

(

i) outer sphere attack of a hydride ligand on the electron-

deﬁcient carbon of CO through transition state B forms

formate and the coordinatively unsaturated intermediate C; (ii)

coordination of dihydrogen to the unsaturated metal center

forms intermediate D; (iii) deprotonation of dihydrogen

through transition state E by formate forms formic acid and

regenerates active species A.

The resting state of the catalyst is F, an oﬀ-cycle species that

45−48

inhibits the rate of CO hydrogenation.

from binding formate to the coordinatively unsaturated species

C. While we cannot deﬁnitively rule out other possibilities, we

This species results

635

Journal of the American Chemical Society

J. Am. Chem. Soc. 2021, 143, 1630−1640

Article

Figure 5. Comparison of partially homogeneous (dashed lines) and

fully heterogeneous (solid lines) systems for the cascade hydrogenation

of carbon dioxide to methanol: (A) in batch reactions carried out for

diﬀerent periods of time with turnover (TON) represented relative to 1

as an average of three diﬀerent reactions. Error bars indicate average

error, and (B) after resubjecting the same catalysts to the reaction

conditions for 16 h. Lines are a guide for the eye, and they do not

represent a mathematical ﬁt to the data.

Figure 6. (A) Turnover number as a function of time in the cascade

hydrogenation of carbon dioxide to methanol in the presence of

catalytic amounts of TFE with a fully heterogeneous multicomponent

system. (B) Eﬀect of the addition of 3 Å molecular sieves on turnover

(

solid lines) overlaid on the reactions without molecular sieves (dotted

suggest that catalyst decomposition had been circumvented by

making the catalyst system fully heterogeneous (vide infra).

Closer inspection of how the heterogeneous reactions evolved

over time revealed a much more complicated reaction proﬁle

lines). Turnover number (TON) determined relative to 1 as an average

of three diﬀerent reactions. Error bars depict average error. Lines are a

guide for the eye, and they do not represent a mathematical ﬁt to the

data.

that involved multiple phases, each responding diﬀerently to the

functionality installed in the MOF. While 1@UiO-66-NH /2@

ammonium functionality for the 1@UiO-66-NH /2@UiO-66

UiO-66 (Figure 5A, blue line) outperformed 1@UiO-66/2@

UiO-66 (Figure 5A, green line) as expected, the major

diﬀerences between the two occurred only at early time points

were most prevalent when autocatalysis was most evident (i.e.,

short reaction times). Therefore, we explored methods to extend

the autocatalytic regime so as to maximize the beneﬁcial eﬀect of

(

time <4 h). After this initial stage of the reaction, the two

the ammonium functionality in 1@UiO-66-NH₃(Figure 6,

heterogeneous reactions produced methanol at a similar,

constant rate regardless of the framework functionality.

Importantly, this outcome did not appear to be a consequence

of catalyst degradation (Figure 5B, blue). Instead, it suggested

that the reaction produced methanol through two diﬀerent

phases with the beneﬁts of the ammonium functionality being

most prominent in the initial phase of the reaction.

dotted lines). In order to achieve this goal, we ﬁrst decreased the

amount of TFE used in the reactions to a catalytic quantity (TFE

−

= 1.0 × 10 mmol). This change resulted in a decrease in overall

activity, which we have previously attributed to changes in

solvent polarity, but the beneﬁts of 1@UiO-66-NH were

maintained (Figure 6A). Moreover, the initial TON versus time

was consistent with conversion versus time plots typically

A feature of the catalyst system not discussed to this point is

observed in autocatalytic reactions.

50−54

the network autocatalysis

product participating in the esteriﬁcation step of the reaction

Figure 1B). The initial phase of the reaction where TON

that results from the alcohol

A puzzling and unexpected feature of these experiments was

that the reduction in turnover rate previously observed at long

reaction times at high TFE loadings was also observed at low

TFE loading. Since catalyst decomposition was not a culprit

(Figure 5B), one potential explanation was catalyst inhibition by

(

rapidly increased with time is consistent with autocatalytic

behavior (Figure 6A). We hypothesized that the beneﬁts of the

636

Journal of the American Chemical Society

J. Am. Chem. Soc. 2021, 143, 1630−1640

Article

Figure 7. Recycling of the fully heterogeneous cascade hydrogenation of carbon dioxide to methanol over ten cycles. Turnover number (TON)

represented relative to 1.

the product or byproduct produced during the reaction.

UiO-66 and 1@UiO-66/2@UiO-66 catalyst systems. This

behavior supported the hypothesis that functionalized and

unfunctionalized systems were primarily diﬀerentiated during

this short rapid-turnover period.

Product inhibition has been linked to catalyst productivity

previously by Saouma and co-workers in similar CO hydro-

genation systems. In order to balance the chemical equation, a

necessary byproduct formed in the hydrogenation of carbon

dioxide to methanol is water. We hypothesized that over time,

the water produced in the reaction inhibits esteriﬁcation (Figure

S6 , step vi), which could lead to the marked change in turnover

frequency observed at long reaction times. Since the

esteriﬁcation step then becomes more kinetically important as

the reaction proceeds, the beneﬁcial eﬀect of the ammonium

Surprisingly, a decrease in turnover frequency at longer time

points was still observed for the heterogeneous catalyst systems

in the presence of molecular sieves (Figure 6B, time >8 h). The

persistence of this phase of the reaction was puzzling considering

that the adsorptive capacity of the added molecular sieves far

exceeded the amount of water produced during the reaction. It is

noteworthy that UiO-66-NH₃⁺and UiO-66 are fairly hygro-

scopic (Table S5 ), so it was hypothesized that water trapped

within the MOF pores could inhibit turnover even in the

presence of molecular sieves. Several control experiments were

carried out to test this hypothesis (Table S6 ). First, recycling was

attempted without prior removal of residual water that may be

trapped in the MOF pores, which is typically done with extensive

washing of the heterogeneous catalyst with fresh methanol

followed by fresh DMF. In this reaction, the heterogeneous

mixture of 1@UiO-66-NH₃⁺and 2@UiO-66 was subjected to

reaction conditions for 16 h and resubjected without washing,

leading to residual water within the MOF pores. This sample

showed a decrease in TON from 4900 to 3800 upon

resubjection to reaction conditions, suggesting that some

inhibitory species remained after the ﬁrst reaction cycle.

Washing this catalyst with fresh methanol followed by fresh

DMF and drying overnight was found to restore the original

activity of the catalyst, suggesting that such a washing process

removed inhibitor from the system. The second control reaction

functionality in 1@UiO-66-NH₃is less pronounced.

In order to minimize the inhibitory eﬀect of water in the

reactions and to maximize the time that the reaction beneﬁts

from its autocatalytic feature, the heterogeneous reactions were

carried out in the presence of 3 Å molecular sieves (Figure 6B,

solid lines). Control experiments revealed that, despite their

Lewis acidity, the sieves are catalytically inactive for the

esteriﬁcation of formic acid (Figure S8), were inactive for the

conversion of carbon dioxide to methanol under reaction

conditions shown in Figure 6, and were superior desiccants

compared to the hygrosocopic MOF (Table S5). Satisfyingly,

adding molecular sieves to the reaction mixture for the cascade

hydrogenation of carbon dioxide led to an approximately 4-fold

increase in turnover for the production of methanol after 16 h

when 1@UiO-66-NH /2@UiO-66 was used as the catalyst

system (Figure 6B). Once again, reaction characteristics

consistent with autocatalysis were observed in the presence of

molecular sieves at short reaction times. However, the major

diﬀerence between the reactions carried out in the presence and

absence of molecular sieves was at intermediate time points

where the rapid turnover period was extended from 2 to 6 h,

that was carried out was to saturate a heterogeneous mixture of

1@UiO-66-NH and 2@UiO-66 with deionized water prior to

catalysis. When such a catalyst was employed, the observed

turnover number decreased from 4900 to 1200, a more than 75%

decrease in activity for the hydrated MOF compared to the dry

leading to a large increase in overall TON and resulting in an

even larger diﬀerence in TON between 1@UiO-66-NH /2@

637

J. Am. Chem. Soc. 2021, 143, 1630−1640

Journal of the American Chemical Society

https://pubs.acs.org/doi/10.1021/jacs.0c08957.

Article

MOF. To verify that hydration did not destroy the precatalyst, a

second batch of the catalyst was saturated with deionized water

and then subjected to standard washing procedures. This

catalyst gave a TON of 4800, conﬁrming pore-conﬁned water’s

role as a reversible inhibitor.

Finally, assembling all of the insights gained in the above

experiments allowed us to design catalyst recycling so as to

Consequently, the resulting catalyst system provides access to

reactivity that is diﬃcult to achieve by manipulating inner-

sphere interactions historically employed to improve catalyst

performance. Importantly, the beneﬁcial eﬀects of the second

coordination sphere were only extensively investigated for the

ﬁrst catalyst in this cascade process. While it is beyond the scope

of this paper, it is possible that second sphere interactions from

diﬀerent functionalized MOFs will beneﬁt the last step in the

cascade process, which would further enhance catalytic

eﬃciency. Moreover, the modularity of the method employed

is attractive for its application to other similar MOF-based

catalytic host−guest systems to improve activity and/or

selectivity. Future eﬀorts will be focused on developing such

systems.

maximize catalyst eﬃciency. Three factors were utilized for this

purpose: (1) 1@UiO-66-NH /2@UiO-66 was used as the

catalyst system to capitalize on the beneﬁcial second sphere

interactions between the Brønsted acidic UiO-66-NH and the

encapsulated ruthenium complex 1 as well as to minimize

decomposition of 2; (2) reactions were carried out in the

presence of 3 Å molecular sieves to eﬀectively remove water

from the reaction; and (3) reactions were carried out for 5 h

instead of 16 h so as to maximize the beneﬁts from using 1@

ASSOCIATED CONTENT

Supporting Information

The Supporting Information is available free of charge at

■

UiO-66-NH and to minimize the inhibitory eﬀects associated

sı

with water building up in the MOF pores. Satisfyingly, turnover

observed in these recycling studies was predictable from the time

course studies shown in Figure 4B and negligible change in

activity was observed through 10 cycles (Figure 7, red bars),

leading to an unprecedented cumulative TON of 100 000 (blue

bars) after 10 cycles (for comparison to relevant catalytic

Procedures and additional data (PDF)

AUTHOR INFORMATION

Corresponding Authors

■

systems for hydrogenation of CO to methanol, see Table S7).

ICP analysis of the MOF material after the recycling experiment

revealed a similar ruthenium loading (2.22 ppm) after the

recycling experiment as observed before the recycling experi-

ment (2.23 ppm). The recyclability and negligible decay in

activity make the system amenable for its adaption for catalysis

in ﬂow.

Jeﬀery A. Byers − Department of Chemistry, Boston College,

Chestnut Hill, Massachusetts 02467, United States;

orcid.org/0000-0002-8109-674X ; Email: je ﬀ ery.byers@

bc.edu

†

Chia-Kuang Tsung − Department of Chemistry, Boston

College, Chestnut Hill, Massachusetts 02467, United States;

orcid.org/0000-0002-9410-565X ; Email: frank.tsung@

bc.edu

CONCLUSION

■

In this paper, we describe an exceptionally eﬃcient multi-

component, host−guest catalyst system for the hydrogenation of

carbon dioxide to methanol that beneﬁts from second sphere

interactions between an MOF host and an organometallic guest.

Systematic manipulation of host functionality was enabled by

the modularity of the MOF and the aperture-opening

encapsulation method employed. This method facilitated

rapid evaluation of a variety of host−guest constructs, which

Authors

Thomas M. Rayder − Department of Chemistry, Boston

College, Chestnut Hill, Massachusetts 02467, United States

Adam T. Bensalah − Department of Chemistry, Boston College,

Chestnut Hill, Massachusetts 02467, United States

Banruo Li − Department of Chemistry, Boston College, Chestnut

Hill, Massachusetts 02467, United States

led to the discovery that the UiO-66-NH₃host signiﬁcantly

Complete contact information is available at:

https://pubs.acs.org/10.1021/jacs.0c08957

increased activity compared to reactions that utilized UiO-66 as

the host: the partially homogeneous catalyst system using 1@

UiO-66-NH and 2 dissolved in solution resulted in a very high

Funding

−

initial TOF of 9100 h . Mechanistic insight into the role of the

ammonium group suggested that it served as a general Brønsted

acid that facilitated the reduction of carbon dioxide to formic

acid, likely by preventing the formation of an oﬀ-cycle resting

species. Time course studies revealed that the beneﬁcial eﬀect of

the ammonium functionality was maximal when the reaction

demonstrated autocatalytic features. Removing water from the

reaction extended the autocatalytic regime, which led to a

further improvement in the catalyst system. When combined

with making the catalyst completely heterogeneous by

This research was supported by the U.S. Department of Energy,

Oﬃce of Science, Oﬃce of Basic Energy Sciences, Catalysis

Science program, under Award DE-SC0019055.

Notes

The authors declare no competing ﬁnancial interest.

†

Chia-Kuang “Frank” Tsung, deceased, January 5, 2021.

DEDICATION

■

This paper is dedicated to our co-author, Chia-Kuang “Frank”

Tsung, a great collaborator, mentor, and friend.

^eⁿ^c^a^p^s^u^l^a^tⁱⁿ^g¹^aⁿ^d²ⁱⁿ^Uⁱ^O^-⁶⁶^-^N^H3 and UiO-66,

respectively, a high TON of 19 000 was obtained in a single

reaction carried out for 16 h. Recycling experiments carried out

REFERENCES

■

for 5 h each to maximize the beneﬁcial eﬀects of the UiO-66-

1) Brown, S. D.; Babbitt, P. C. New insights about enzyme evolution

NH host led to an unprecedented cumulative TON of 100 000

from large scale studies of sequence and structure relationships. J. Biol.

Chem. 2014, 289, 30221−30228.

2) Raynal, M.; Ballester, P.; Vidal-Ferran, A.; van Leeuwen, P. W. N.

when the heterogeneous catalyst was recycled ten times. The

developments described in this work represent how host−guest

catalyst systems can beneﬁt from systematic manipulation of

noncovalent interactions designed to maximize performance.

M Supramolecular catalysis. Part 2: artificial enzyme mimics. Chem. Soc.

Rev. 2014, 43, 1734−1787.

638

J. Am. Chem. Soc. 2021, 143, 1630−1640

Journal of the American Chemical Society

Article

3) Rayder, T. M.; Adillon, E. H.; Byers, J. A.; Tsung, C.-K. A

(23) Kosinov, N.; Liu, C.; Hensen, E. J. M.; Pidko, E. A. Engineering of

Transition Metal Catalysts Confined in Zeolites. Chem. Mater. 2018,

30, 3177−3198.

Bioinspired Multicomponent Catalytic System for Converting Carbon

Dioxide into Methanol Autocatalytically. Chem. 2020 , 6 , 1742 −1754.

4) Khusnutdinova, J. R.; Garg, J. A.; Milstein, D. Combining Low-

(24) Dai, C.; Zhang, A.; Song, C.; Guo, X. Advances in the synthesis

and catalysis of solid and hollow zeolite-encapsulated metal catalysts.

Adv. Catal. 2018, 63, 75−115.

Pressure CO Capture and Hydrogenation to Form Methanol. ACS

Catal. 2015, 5, 2416−2422.

(5) Kothandaraman; Goeppert, A.; Czaun, M.; Olah, G. A.; Prakash,

(25) Mollick, S.; Fajal, S.; Mukherjee, S.; Ghosh, S. K. Stabilizing

Metal-Organic Polyhedra (MOP): Issues and Strategies. Chem. - Asian

J. 2019, 14, 3096−3108.

(26) Moshoeshoe, M.; Nadiye-Tabbiruka, M. S.; Obuseng, V. A

Review of the Chemistry, Structure, Properties, and Applications of

Zeolites. Am. J. Mater. Sci. 2017, 7, 196−221.

(27) Moghadam, P. Z.; Li, A.; Wiggin, S. B.; Tao, A.; Maloney, A. G.

P.; Wood, P. A.; Ward, S. C.; Fairen-Jimenez, D. Development of a

Cambridge Structural Database Subset: A Collection of Metal-Organic

Frameworks for Past, Present, and Future. Chem. Mater. 2017, 29,

2618−2625.

(28) Cohen, S. M. The Postsynthetic Renaissance in Porous Solids. J.

Am. Chem. Soc. 2017, 139, 2855−2863.

G. K. S. Conversion of CO from air into methanol using a polyamine

and a homogeneous ruthenium catalyst. J. Am. Chem. Soc. 2016, 138,

6) Ribeiro, A. P. C.; Martins, L. M. D. R. S.; Pombeiro, A. J. L. Carbon

78−781.

dioxide-to-methanol single-pot conversion using a C-scorpionate

iron(II) catalyst. Green Chem. 2017, 19, 4811−4815.

Low-Temperature Hydrogenation of Carbon Dioxide to Methanol with

a Homogeneous Cobalt Catalyst. Angew. Chem., Int. Ed. 2017, 56,

7) Schneidewind, J.; Adam, R.; Baumann, W.; Jackstell, R.; Beller, M.

(

890−1893.

8) Kar, S.; Sen, R.; Kothandaraman, J.; Goeppert, A.; Chowdhury, R.;

Munoz, S. B.; Haiges, R.; Prakash, G. K. S. Mechanistic insights into

ruthenium -pincer-catalyzed amine-assisted homogeneous hydro-

genation of CO to methanol. J. Am. Chem. Soc. 2019 , 141 , 3160 −3170.

Snurr, R. Q.; Hupp, J. T.; Farha, O. K. A facile synthesis of UiO-66,

(29) Katz, M. J.; Brown, Z. J.; Colon, Y. J.; Siu, P. W.; Scheidt, K. A.;

UiO-67 and their derivatives. Chem. Commun. 2013, 49, 9449 −9451.

(30) Kalaj, M.; Palomba, J. M.; Bentz, K. C.; Cohen, S. M. Multiple

functional groups in UiO-66 improve chemical warfare agent simulant

degradation. Chem. Commun. 2019, 55, 5367−5370.

9) Everett, M.; Wass, D. F. Highly productive CO hydrogenation to

methanol − a tandem catalytic approach via amide intermediates.

Chem. Commun. 2017, 53, 9502−9504.

(

10) Wesselbaum, S.; Moha, V.; Meuresch, M.; Brosinski, S.; Thenert,

K. M.; Kothe, J.; vom Stein, T.; Englert, U.; Ho

lscher, M.;

(31) Katayama, Y.; Bentz, K. C.; Cohen, S. M. Defect-Free MOF-

Based Mixed-Matrix Membranes Obtained by Corona Cross-Linking.

ACS Appl. Mater. Interfaces 2019, 11, 13029−13037.

Klankermeyer, J.; Leitner, W. Hydrogenation of carbon dioxide to

methanol using a homogeneous ruthenium −triphos catalyst: from

mechanistic investigations to multiphase catalysis. Chem. Sci. 2015, 6,

(32) Doonan, C. J.; Morris, W.; Furukawa, H.; Yaghi, O. M.

Isoreticular Metalation of Metal-Organic Frameworks. J. Am. Chem. Soc.

2009, 131, 9492−9493.

11) Huff, C. A.; Sanford, M. S. Cascade catalysis for the

93−704.

homogeneous hydrogenation of CO to methanol. J. Am. Chem. Soc.

(33) Li, Z.; Rayder, T. M.; Luo, L.; Byers, J. A.; Tsung, C.-K. Aperture-

opening encapsulation of a transition metal catalyst in a metal −organic

12) Chu, W.-Y.; Culakova, Z.; Wang, B.-T.; Goldberg, K. I. Acid-

011, 133, 18122−18125.

framework for CO

8085.

hydrogenation. J. Am. Chem. Soc. 2018, 140, 8082−

Assisted Hydrogenation of CO to Methanol in a Homogeneous

Catalytic Cascade System. ACS Catal. 2019, 9, 9317−9326.

(34) Morabito, J. V.; Chou, L.-Y.; Li, Z.; Manna, C. M.; Petroff, C. A.;

Kyada, R. J.; Palomba, J. M.; Byers, J. A.; Tsung, C.-K. Molecular

encapsulation beyond the aperture size limit through dissociative linker

exchange in metal-organic framework crystals. J. Am. Chem. Soc. 2014,

136, 12540−12543.

(35) Li, Z. Design of a Host-guest Hybrid Catalytic System Through

Aperture-opening Encapsulation Using Metal-Organic Framework.

PhD Thesis, Boston College, 2019.

14) Li, Z.; Wang, D.; Wu, Y.; Li, Y. Recent advances in the precise

13) Cuesta, S. M.; Rahman, S. A.; Furnham, N.; Thornton, J. M. The

classification and evolution of enzyme function. Biophys. J. 2015, 109,

082−1086.

(15) Liu, X.-M.; Lu, G. Q.; Yan, Z.-F.; Beltramini, J. Recent advances

control of isolated single-site catalysts by chemical methods. Natl. Sci.

Rev. 2018, 5, 673−689.

in catalysts for methanol synthesis via hydrogenation of CO and CO ₂ .

(36) Dolgopolova, E. A.; Berseneva, A. A.; Faillace, M. S.; Ejegbavwo,

O. A.; Leith, G. A.; Choi, S. W.; Gregory, H. N.; Rice, A. M.; Smith, M.

D.; Chruszcz, M.; Garashchuk, S.; Mythreye, K.; Shustova, N. B.

Confinement-Driven Photophysics in Cages, Covalent-Organic Frame-

works, Metal-Organic Frameworks, and DNA. J. Am. Chem. Soc. 2020,

142, 4769−4783.

Ind. Eng. Chem. Res. 2003, 42, 6518−6530.

(

16) Valdes, H.; García-Eleno, M. A.; Canseco-Gonzalez, D.; Morales-

Morales, D. Recent advances in catalysis with transition-metal pincer

compounds. ChemCatChem 2018 , 10 , 3136 −3172.

(17) Trost, B. M.; Bunt, R. C. On Ligand Design for Catalytic Outer

Sphere Reactions: A Simple Asymmetric Synthesis of Vinylglycinol.

(37) March, J. Advanced Organic Chemistry: Reactions, Mechanisms,

and Structure, 3rd ed.; Wiley: New York, 1985.

Angew. Chem., Int. Ed. Engl. 1996, 35, 99−102.

(18) Carboni, S.; Gennari, C.; Pignataro, L.; Piarulli, U. Supra-

(38) Ghosh, P.; Colon, Y. J.; Snurr, R. Q. Water adsorption in UiO-66:

molecular ligand-ligand and ligand-substrate interactions for highly

the importance of defects. Chem. Commun. 2014, 50, 11329 −11331.

(39) Cirujano, F. G.; Corma, A.; Llabres i Xamena, F. X. Conversion of

levulinic acid into chemicals: synthesis of biomass derived levulinate

selective transition metal catalysis. Dalton Trans. 2011 , 40 , 4355 −4373.

(19) Prokopchuk, D. E.; Morris, R. H. Inner-Sphere Activation,

Outer-Sphere Catalysis: Theoretical Study on the Mechanism of

Transfer Hydrogenation of Ketones Using Iron(II) PNNP Eneamido

Complexes. Organometallics 2012, 31, 7375−7385.

esters over Zr-containing MOFs. Chem. Eng. Sci. 2015, 124, 52−60.

(40) When EtOD was used in place of EtOH to ensure that the ND

functional group remained deuterated during catalysis, an isotope eﬀect

of 1.83 was observed (Table S2). While this is a nonnegligible change, it

indicates the same behavior as described in the text.

(

20) Fang, Y.; Powell, J. A.; Li, E.; Wang, Q.; Perry, Z.; Kirchon, A.;

Yang, X.; Xiao, Z.; Zhu, C.; Zhang, L.; Huang, F.; Zhou, H.-C. Catalytic

reactions within the cavity of coordination cages. Chem. Soc. Rev. 2019,

(41) Hanton, M. J.; Tin, S.; Boardman, B. J.; Miller, P. Ruthenium-

catalysed hydrogenation of esters using tripodal phosphine ligands. J.

Mol. Catal. A: Chem. 2011, 346, 70−78.

(42) Similar experiments designed to isolate the ﬁrst step of the

reaction could not be carried out because the hydrogenation of carbon

dioxide to formic acid requires base to drive the reaction to completion.

This base would deprotonate the ammonium functional group that is

necessary to facilitate the ﬁrst step of the reaction.

(22) Corma, A.; Garcia, H. Supramolecular host-guest systems in

8, 4707−4730.

21) Tan, C.; Chu, D.; Tang, X.; Liu, Y.; Xuan, W.; Cui, Y.

Supramolecular Coordination Cages for Asymmetric Catalysis. Chem. -

Eur. J. 2019, 25, 662−672.

zeolites prepared by ship-in-a-bottle synthesis. Eur. J. Inorg. Chem. 2004,

004, 1143−1164.

639

J. Am. Chem. Soc. 2021, 143, 1630−1640

Journal of the American Chemical Society

(45) Filonenko, G. A.; Hensen, E. J. M.; Pidko, E. A. Mechanism of

Article

(43) No conversion was observed without the addition of alcohol

additive, suggesting that no side reactions were occurring during the

esteriﬁcation reaction.

(44) Lowering the alcohol additive loading resulted in diﬀerent rates

of esteriﬁcation for the various linkers tested with increased

esteriﬁcation rates being observed with linkers containing electron

withdrawing groups, but there was still no diﬀerence in rate for UiO-66-

NH₃⁺and UiO-66-ND₃⁺(Figure S7 ).

CO hydrogenation to formates by homogeneous Ru-PNP pincer

catalyst: from a theoretical description to performance optimization.

Catal. Sci. Technol. 2014, 4, 3474−3485.

(46) Ahlquist, M.; Fabrizi, G.; Cacchi, S.; Norrby, P.-O. The

Mechanism of the Phosphine-Free Palladium-Catalyzed Hydro-

arylation of Alkynes. J. Am. Chem. Soc. 2006, 128, 12785−12793.

(47) Mathis, C. L.; Geary, J.; Ardon, Y.; Reese, M. S.; Philliber, M. A.;

VanderLinden, R. T.; Saouma, C. T. Thermodynamic Analysis of

Metal-Ligand Cooperativity of PNP Ru Complexes: Implications for

CO Hydrogenation to Methanol and Catalyst Inhibition. J. Am. Chem.

Soc. 2019, 141, 14317−14328.

48) Huff, C. A.; Sanford, M. S. Catalytic CO Hydrogenation to

Formate by a Ruthenium Pincer Complex. ACS Catal. 2013, 3, 2412−

416.

49) Fully heterogeneous systems were assembled by encapsulating 1

(

in UiO-66-X and 2 in UiO-66 in separate reactions and then physically

mixing samples of each species together (See Supporting Information).

M.; Campbell, C. E.; Kang, K.; Fox, J. M.; Whitesides, G. M.

50) Plasson, R.; Brandenburg, A.; Jullien, L.; Bersini, H.

Autocatalyses. J. Phys. Chem. A 2011 , 115 , 8073 −8085.

51) Bissette, A. J.; Fletcher, S. P. Mechanisms of Autocatalysis. Angew.

Chem., Int. Ed. 2013, 52, 12800−12826.

52) Semenov, S. N.; Kraft, L. J.; Ainla, A.; Zhao, M.; Baghbanzadeh,

(

Autocatalytic, bistable, oscillatory networks of biologically relevant

organic reactions. Nature 2016, 537, 656−660.

(53) Semenov, S. N.; Belding, L.; Cafferty, B. J.; Mousavi, M. P. S.;

Finogenova, A. M.; Cruz, R. S.; Skorb, E. V.; Whitesides, G. M.

Autocatalytic Cycles in a Copper-Catalyzed Azide-Alkyne Cyclo-

addition Reaction. J. Am. Chem. Soc. 2018, 140, 10221 −10232.

(54) McMurchie, E. J.; McGlasson, W.; Eaks, I. Treatment of Fruit

with Propylene gives Information about the Biogenesis of Ethylene.

Nature 1972, 237, 235−236.

(55) Reversible inhibition from carbon monoxide could not be fully

ruled out, as a small amount of carbon monoxide exists in the reaction

from the water−gas shift reaction catalyzed by the surface of the reactor.

640