ArticlePDF Available

Fabrication of metal-organic framework architectures with macroscopic size: A review

Authors:

Abstract and Figures

Metal-organic framework (MOFs) are potentially attractive porous materials for many industrial applications; however, they are not necessarily suitable for industrial scale implementations. MOFs are typically produced as polydisperse microcrystalline powders. Industrial applicability of MOF powders is limited by their mechanical, chemical and attrition resistance, mass transfer limitations, and/or poor handling properties. Formulating MOFs in macroscopic architectures is an essential step towards their successful implementations at industrial level. In this comprehensive Review, we extensively describe the most popular synthesis methods of macroscopic MOF architectures: mechanical densification, coordination replication, sol–gel approach, Pickering emulsion method, and 3D printing technology. The main objective of this Review is to systematize our knowledge about the synthesis methods of macroscopic MOF architectures. A thorough understanding of these synthesis techniques is expected to facilitate the preparation of innovative macroscopic MOF architectures. The excellent performance of macroscopic MOF architectures in comparison to polydisperse microcrystalline MOF powders is also revealed throughout the Review. Finally, we outline the advantages and drawbacks of each synthesis approach.
Content may be subject to copyright.
Review
Fabrication of metal-organic framework architectures with macroscopic
size: A review
Javier Fonseca
, Tenghua Gong
Department of Chemical Engineering, Northeastern University, 313 Snell Engineering Center, 360 Huntington Avenue, Boston, MA 02115-5000, United States
article info
Article history:
Received 29 October 2021
Accepted 14 March 2022
Keywords:
Metal-organic frameworks
Mechanical densification
Pseudomorphic replication
Sol–gel
Pickering emulsions
3D printing
abstract
Metal-organic framework (MOFs) are potentially attractive porous materials for many industrial applica-
tions; however, they are not necessarily suitable for industrial scale implementations. MOFs are typically
produced as polydisperse microcrystalline powders. Industrial applicability of MOF powders is limited by
their mechanical, chemical and attrition resistance, mass transfer limitations, and/or poor handling prop-
erties. Formulating MOFs in macroscopic architectures is an essential step towards their successful
implementations at industrial level. In this comprehensive Review, we extensively describe the most
popular synthesis methods of macroscopic MOF architectures: mechanical densification, coordination
replication, sol–gel approach, Pickering emulsion method, and 3D printing technology. The main objec-
tive of this Review is to systematize our knowledge about the synthesis methods of macroscopic MOF
architectures. A thorough understanding of these synthesis techniques is expected to facilitate the prepa-
ration of innovative macroscopic MOF architectures. The excellent performance of macroscopic MOF
architectures in comparison to polydisperse microcrystalline MOF powders is also revealed throughout
the Review. Finally, we outline the advantages and drawbacks of each synthesis approach.
Ó2022 Elsevier B.V. All rights reserved.
https://doi.org/10.1016/j.ccr.2022.214520
0010-8545/Ó2022 Elsevier B.V. All rights reserved.
Abbreviations: MOF, metal-organic framework; 3D, three-dimensional; PVA, polyvinyl alcohol; K, bulk modulus; G, shear modulus or rigidity modulus; H
2
BDC, benzene
dicarboxylic acid; H
3
BTB, 1,3,5-benzenetribenzoic acid; ENG, expanded natural graphite; H
3
BTC, benzene tricarboxylic acid; H
4
dhtp, 2,5-dihydroxyterephthalic acid; H
2
BDC-
NH
2
, 2-aminoterephthalic acid; H
2
BPDC, 2,2’-bipyridyl-4,4’-dicarboxylic acid; 4-m-5-ica, 4-methyl-5-imidazolecarboxaldehyde; TazBz, 3,3
0
,5,5
0
-azobenzene-tetracarbox
ylate; H
4
bptc, 3,3
0
,5,5
0
-biphenyl-tetracarboxylic acid; SBU, secondary building unit; H
2
BDC-NO
2
, 2-nitroterephthalic acid; H
2
BDC-acetamido, 2-acetamidoterephthalic acid;
H
2
BDC-Br, 2-bromoterephthalic acid; 1,4-NDC, 1,4-naphthalenedicarboxylate; 2,6-NDC, 2,6-naphthalenedicarboxylate; MTV, multivariate; 2D, two-dimensional; GNR, gold
nanorod; DMF, N,N-dimethylformamide; bpy, 4,4’-bipyridine; NP, nanoparticle; CHN, copper hydroxide nitrate; DBT, dibenzothiophene; PEC, photoelectrochemical; ORR,
oxygen reduction reaction; PEO, polyethylene oxide; H
2
sq, squaric acid; H
2
dhbq, 2,5-dihydroxybenzoquinone; MOXs, metal-organic xerogel; MOA, metal-organic aerogel; tt-
MA, trans,trans-muconic acid; FeOAc, iron(III) acetate; NPC, nanoporous carbon; aASC, aqueous asymmetric supercapacitor; EtOH, ethanol; H
2
FDC, 2,5-furandicarboxylic acid;
H
2
ADC, 9,10-anthracenedicarboxylic acid; CTAB, cetyltrimethylammonium bromide; TMB, 1,3,5-trimethylbenzene; MC, microcystins; SMZ, sulfamethazine; SDZ,
sulfadiazine; 2,4-DNT, 2,4-dinitrophenol; 2,6-DNT, 2,6-dinitrophenol; HPLC, high performance liquid chromatography; TEA, triethylamine; ACN, acetonitrile; DMSO,
dimethyl sulfoxide; MeOH, methanol; VE, viscoelastic; CMC, carboxymethylcellulose; MB, methylene blue; H
4
TBAPy, 1,3,6,8-tetrakis(p-benzoic acid)pyrene); FLIM,
fluorescence lifetime imaging microscopy; HIPE, high internal phase emulsion; SDS, sodium dodecyl sulfate; GO, graphene oxide; RHB, Rhodamine B; PAM, polyacrylamide;
V50, 2,2’-azobis(2-methylpropionamidine) dihydrochloride; DVB, divinylbenzene; PS, polystyrene; IL, ionic liquid; BSA, bovine serum albumin; AM, additive manufacturing;
CAD, Computer-Aided Design; ABS, acrylonitrile butadiene styrene; IJP, inkjet printing; SLS, selective laser sintering; SLA, stereolithography; FDM, fused deposition modeling;
DIW, direct ink writing; EG, ethylene glycol; dabco, 1,4-diazabicyclo[2.2.2]octane; Ln-MOFs, lanthanide-organic frameworks; Mell, mellitate; N12, Nylon-12; MMFs, mixed
matrix films; PA12, polyamide 12; UV, ultraviolet; DLP, digital light processing; DOPsL, dynamic optical projection stereolithography; PLA, polylactic acid; HIPS, high impact
polystyrene; CPE, chlorinated polyethylene; PBT, polybutylene terephthalate; PEEK, polyether ether ketone; PMMA, poly(methyl methacrylate); PP, polypropylene; T
g
, glass
transition temperature; EtOAc, ethyl acetate; CR, Congo red; PVDF-HFP, poly(vinylidene-co-hexafluoropropylene); 3DFD, 3D fibre deposition; CA-GE, calcium alginate and
gelatin; pzdc, pyrazine-2,3-dicarboxylate; pyz, pyrazine; HEC, 2-hydroxyethyl cellulose; PVP, polyvinylpyrrolidone; NMP, N-methyl-2-pyrrolidone; Torlon, polyamide
(imide); GPG, gel-print-grow; TOCNF, 2,2,6,6-tetramethylpiperidine-1-oxylradical-mediated oxidized cellulose nanofibers; PCL, polycaprolactone; DCPD, dicalcium
phosphate dihydrate; BMSC, bone mesenchymal stem cell;
r
y
, yield stress; Hcit, citric acid; HPC, hydroxypropyl cellulose; IPA, isopropyl alcohol; PEI, polyethylenimine;
TEPA, tetraethylenepentamine; MBG, macroporous bioactive glass; TMPPTA, trimethylolpropane propoxylate triacrylate; DCM, dichloromethane; FAST, field-assisted
sintering technique; PAN, polyacrylonitrile; Co-NC, cobalt-nitrogen-carbon.
Corresponding author.
E-mail address: fonsecagarcia.j@northeastern.edu (J. Fonseca).
Coordination Chemistry Reviews 462 (2022) 214520
Contents lists available at ScienceDirect
Coordination Chemistry Reviews
journal homepage: www.elsevier.com/locate/ccr
Contents
1. Introduction . . . ........................................................................................................ 2
2. Mechanical densification methods . . . . ..................................................................................... 3
2.1. Mechanical compression or pelletization. . . . ........................................................................... 3
2.2. Mechanical granulation. . ........................................................................................... 7
2.3. Mechanical extrusion . . . ........................................................................................... 8
2.4. Spray drying . . . . . . . . . . ........................................................................................... 9
3. Pseudomorphic replication . . . . . . . . . . .................................................................................... 10
3.1. Pseudomorphic replication of Al-based monolithic solids . . . . . . . . . ....................................................... 10
3.2. Pseudomorphic replication of Cu-based monolithic solids . . . . . . . . . ....................................................... 11
3.3. Pseudomorphic replication of Zn-based monolithic solids . . . . . . . . . ....................................................... 13
3.4. Pseudomorphic replication of V-based monolithic solids . . . . . . . . . . ....................................................... 14
3.5. Pseudomorphic replication of Ca-based monolithic solids . . . . . . . . . ....................................................... 15
4. Sol-gel method . ....................................................................................................... 15
4.1. Sol-gel processing of Fe-based monolithic gels . . . . . . . . . . . . . . . . . . ................ ....................................... 17
4.2. Sol-gel processing of Al-based monolithic gels . . . . . . . . . . . . . . . . . . ....................................................... 18
4.3. Sol-gel processing of Cu-based monolithic gels . . . . . . . . . . . . . . . . . .......... ............................................. 19
4.4. Sol-gel processing of Zn-based monolithic gels. . . . . . . . . . . . . . . . . . ....................................................... 22
4.5. Sol-gel processing of Zr-based monolithic gels . . . . . . . . . . . . . . . . . . ....................................................... 23
5. Pickering emulsion method . . . . . . . . . . .................................................................................... 25
5.1. Aerogel-type monoliths . .......................................................................................... 25
5.2. Polymer-supported monoliths . . . . . . . . . . . . .......................................................................... 27
6. 3D printing technology . . . . . . . . . . . . . .................................................................................... 28
6.1. 3D printing technologies. ......................................................................................... . 28
6.1.1. MOF architectures prepared by inkjet printing . . . . . ............................................................ 28
6.1.2. MOF architectures prepared by selective laser sintering . . . . . . . . . . . . . . . . . ......................................... 29
6.1.3. MOF architectures prepared by stereolithography . . . ............................................................ 31
6.1.4. MOF architectures prepared by fused deposition modeling . . . . . . . . . . . . . . ......................................... 31
6.1.5. Direct ink writing . . . . . . . . . . ............................................................................... 32
7. Conclusions. . . . ....................................................................................................... 40
Declaration of Competing Interest . . . . .................................................................................... 42
Acknowledgements . . . . . . . . . . . . . . . . .................................................................................... 42
Appendix A. Supplementary data . . . . . . . .................................................................................... 42
References . . . . ....................................................................................................... 42
1. Introduction
Metal-organic frameworks (MOFs) are crystalline or amorphous
coordination compounds formed through the self-assembly of
organic ligands with metal ions or clusters [1–3]. MOFs exhibit
high permanent porosity, large specific surface areas, and abun-
dant active sites. In addition, unlike other porous solids such as
zeolites, silica, or activated carbon, MOFs offer unique structural
tenability [4,5]. The chemical structures of MOFs can be designed,
modified and altered using different chemical building units, regu-
lating synthetic parameters and/or utilizing post-synthetic treat-
ments. The physicochemical properties of MOFs make them
excellent candidates for a wide range of applications such as
adsorption [6–8], separation [9–11], catalysis [12–14], sensing
[15–17], nonlinear optics [18], magnetism [19], drug delivery
[20–22], etc [23–27].
During the last two decades, most MOFs have been synthesized
as polydisperse microcrystalline powders with crystallite sizes
ranging from nanometers to hundreds of microns [28]. However,
MOF powders are not necessarily suitable for industrial scale
implementation. The industrial applicability of MOF powders is
limited by their poor handling properties, mass transfer limita-
tions, and mechanical instability [29,30]. For example, a MOF pow-
der adsorption column often experiences a significant pressure
drop over time, resulting in higher mass resistance within the col-
umn [31]. The pressure drop is due to the gradual compaction of
the powder with pressure. Gas storage requires materials with
the following properties: high volumetric gas storage density, high
thermal conductivity, and stability in humid and reactive
environments. Unfortunately, MOF powders lack these properties
[29,32–34]. In the field of catalysis, MOF microcrystalline powders
often present difficulties for separation and recycling [35,36].MOF
powders do not exhibit high mechanical strength, or good thermal
and chemical stability at high temperatures under reactive condi-
tions [37]. Furthermore, these catalysts can agglomerate and, in
turn, be deactivated during successive catalytic cycles [38]. There-
fore, a bottleneck for the use of MOFs is the preparation of architec-
tures with macroscopic sizes, whether they are architectures based
on high powder packing density or continuous and structurally
coherent architectures [39]. Attention is slowly shifting from
microcrystalline MOF powders for laboratory-scale applications
to macroscopic MOF architectures for industrial scale applications.
Microcrystalline MOFs are shaped into macroscopic architec-
tures to promote handling, mechanical strength, low mass transfer
resistance, volumetric adsorption capacities, BET surface areas,
packing density, reusability, and stability [37,40–42]. A macro-
scopic MOF architecture is successfully manufactured if it meets
the three following requirements: [43]
(1) Retention of the microcrystalline powder structure (i.e., particle
crystallinity, topography, etc.). This manufacturing criterion
is especially relevant for brittle and/or solvent-sensitive
MOFs that can break during shaping. Furthermore, MOF
functional groups can interact with chemicals, such as bin-
ders, required for shaping, thereby leading to differences in
active moieties between the pristine microcrystalline pow-
ders and the macroscopic architecture.
(2) Retention of the physiochemical properties of the microcrys-
talline powder. The physiochemical properties include textu-
ral properties (i.e., surface area, porosity, and pore volume)
J. Fonseca and T. Gong Coordination Chemistry Reviews 462 (2022) 214520
2
and chemical properties (i.e., surface acidity/basicity, and
redox properties).
(3) Achievement of optimal mechanical strength, thermal stability,
and/or chemical resilience of the macroscopic MOF architecture.
These properties are critical for the applicability of the
architectures.
Many methods have been developed to synthesize MOF archi-
tectures with macroscopic size [31]. MOFs can be mechanically
densified in macroscopic architectures by pressing (pelletization)
[44], granulation [45], extrusion [46], and spray drying [47]. These
approaches are also known as mechanical shaping methods. A crit-
ical goal is to reduce the volume of intergranular voids without
destroying the intrinsic microporous structure. Unfortunately,
these compacting processes often reduce the porosity of MOFs
and, in turn, the accessibility of the micropores to guest molecules.
The optimal conditions required for compaction without structural
collapse are highly dependent on the identity of the MOF [31].
Beyond the mechanical shaping methods, continuous and struc-
turally coherent MOF architectures with macroscopic size can be
prepared by coordination replication [48,49], sol–gel approach
[50,51], Pickering emulsion method [52,53], and three-
dimensional (3D) printing technology [54].
Macroscopic architectures can also be produced by embedding
and/or coating pre-synthesized macroporous architectures [55–58]
. Furthermore, host architectures can be created around pre-
synthesized MOFs, thus forming the macroscopic architectures
[59–60]. Strategies to incorporate MOFs into pre-synthesized
macroporous architectures and strategies to synthesized host
architectures around pre-synthesized MOFs are out of the scope
of this Review. Recent reviews on those synthesis approaches can
be found elsewhere [40,41,61,62]. Herein, we extensively discuss
the strategies that produce macroscopic MOF architectures com-
posed entirely (binderless strategies) or primarily by MOFs. MOF
films can be also considered as a MOF architecture. However, this
Review is mainly focused on the preparation of 3D macroscopic
MOF architectures. Previous reviews have discussed the topic of
MOF thin films [63–65].
The above-mentioned methods (mechanical shaping, coordina-
tion replication, sol–gel approach, Pickering emulsion method, and
3D printing technology) can prepare pure macroscopic MOF archi-
tectures or macroscopic composite architectures. Binders may pro-
vide the mechanical stability necessary to withstand stress during
the preparation of architectures, as well as improving the density,
mechanical strength, attrition resistance, and thermal conductivity
of the final composite architectures. Conversely, binders often
dilute the active component of MOFs by covering their surface
and/or blocking their pores, resulting in reduced performance per
unit mass (or volume) of the macroscopic architecture. In brief,
pure MOF architectures often exhibit superior performance than
composite architectures for those applications that depend on
MOF properties [54]. Therefore, methodologies for producing pure
(or binderless) MOF architectures are highly sought after.
Despite the importance of the subject, to the best of our knowl-
edge, no Review has comprehensively summarized all the principal
strategies for developing macroscopic MOF architectures. Thus, the
major aim of this Review is to systematize our knowledge about
those synthesis approaches (mechanical densification, coordina-
tion replication, sol–gel approach, Pickering emulsion method,
and 3D printing technology). We explain the most widespread
strategies to prepare macroscopic MOF architectures with a focus
on the design and control of their textural properties, mechanical
strength, thermal stability, and chemical resilience. The potential
application of the prepared MOF architectures is also revealed.
Finally, the advantages and drawbacks of each synthesis approach
are outlined.
2. Mechanical densification methods
Although the synthesis of MOFs is well documented, their den-
sification or shaping is rarely studied. However, the use of MOFs in
industrial-scale catalysis and separations often requires a shaping
step to generate macroscopic architectures of millimetric dimen-
sions with sufficient mechanical resistance and maintaining the
initial porosity of MOF powders. Mechanical densification meth-
ods, such as compression, granulation, extrusion, and spray drying,
are the simplest techniques for preparing macroscopic MOF archi-
tectures based on a high density of powder packing. These meth-
ods are based on adding MOF powder into a mold and
subsequently applying certain pressure to shape the materials. Bin-
ders are generally used to facilitate the shaping process and
improve the density and mechanical strength of densified MOFs.
Inorganic binders, such as alumina, silica, and silica/graphite, and
organic binders, such as graphite and polyvinyl alcohol (PVA), are
typically utilized. The final application and operating conditions
dictate the shape of the object and the choice of binders and heat
treatments.
MOFs generally exhibit weak mechanical stabilities. Many
MOFs suffer partial loss of textural properties, or even structural
collapses, under mechanical densification processes [1,66–69].
This limitation remains a key bottleneck for certain large-scale
applications that require dense packing of MOF powders. MOFs
must be mechanically stable enough to retain their crystallinity
and textural properties during shaping. Bulk modulus K (inverse
of compressibility) of benchmark MOFs (MOF-5 (Zn
4
O(BDC)
3
,H
2
-
BDC = benzene dicarboxylic acid) and ZIF-8 (Zn(MeIM)
2
, MeIM =
2-methylimidazolate)) ranges from 7 to 20 GPa, while their shear
modulus G (also called rigidity modulus) is an order of magnitude
smaller than the K values [66,70–73]. Therefore, shear modulus is
the most critical parameter to consider when evaluating the
mechanical stabilities of MOFs.
The fundamental mechanisms of mechanical compression,
granulation, extrusion, and spray drying are described in this
Section (Table S2). The major aim is to systematize our knowledge
about the mechanical densification of MOFs. Unfortunately,
parameters related to MOF shaping, such as compression ramp
speed and dwell time, are rarely reported, although they can be
of great importance for the final properties of densified MOFs.
2.1. Mechanical compression or pelletization
The oldest, easiest, and most common method of densifying
MOFs is pelletizing by mechanical or hydraulic pressing [32,37].
MOF powders are typically placed between two plates in a cylin-
drical die (Fig. 1). MOF pellets are formed by applying unilateral
hydraulic pressure to MOF powders. This mechanical compression
induces intraparticle binding. The intrusion between particles is
directly related to the applied force [43]. The final properties of
the densified MOFs depend on the density of powder packing,
which, in turn, depends on the strengths of the pressing. Unilateral
pressures can deactivate MOFs and/or limit mass transfer within
MOFs. The degree to which this deactivation occurs varies between
frameworks. Despite these disadvantages, pelletization is widely
used in the industry due to its simplicity [74–78]. Pelletization
can be carried out with or without a binder. As previously men-
tioned, binders typically improve the mechanical stability and/or
thermal conductivity of MOF pellets [79,80]. However, binders
can reduce the catalytic and/or adsorption capacities of MOFs. It
is worth noting that binders often exhibit poor adsorption and cat-
alytic capacity. In addition, binders can also decrease the surface
area of MOFs by blocking their pores. Therefore, binder-free
MOF pellets are highly recommended as densified structures in
J. Fonseca and T. Gong Coordination Chemistry Reviews 462 (2022) 214520
3
large-scale industrial applications. In mechanical compression, hot
pressing and MOF activation have often been employed to limit the
use of binders or, in some cases, avoid binders altogether, resulting
in densified MOFs with no loss of catalytic or adsorbent capacity
[32].
MOFs with ultrahigh porosity have been found to be favorable
materials for H
2
-storage systems based on cryo-adsorption, due
to their low heat evolution during loading and the high gravimetric
H
2
storage capacity [81,82]. The volumetric H
2
storage capacity of
MOFs has been significantly increased through densification [83–
89]. MOF-177 (Zn
4
O(BTB)
2
,H
3
BTB = 1,3,5-benzenetribenzoic acid)
powder has been compressed to prepare MOF-177 pellets with
bulk densities of 0.39, 0.42, 0.53, 0.62, 0.73, 0.90, 1.10 and
1.40 g cm
3
[83]. Manual compression was applied until a prede-
termined density was reached. Microcrystalline MOF-177 powder
gradually amorphized when subjected to compressive stress. Both
the micropore volume and the surface area were found to decrease
as the density of MOF-177 increased. Upon reaching densities
below 0.4 g cm
3
, some micropore volume was lost due to the
compression-induced distortion of MOF-177. However, micropore
volume decreased considerably when reaching densities higher
than 0.43 g cm
3
. This reduction was ascribed to a progressive
transition to the amorphous phase. The gravimetric H
2
storage
capacity of the MOF-177 pellet decreased as its density increased,
which was related to the progressive loss of micropore volume.
Conversely, the volumetric H
2
capacity of MOF-177 improved with
densification. The total volumetric H
2
storage capacity of the MOF-
177 pellet with a density of about 0.427 g cm
3
was 42 g L
1
at
about 8 MPa and 77 K [83].
Similar results have been shown for the compaction of MOF-5
[84–86]. The crystal structure of MOF-5 has been reported to
undergo amorphization at the relatively low applied pressure of
3.5 MPa [68]. However, MOF-5 has also been found to maintain a
considerable fraction of the crystalline phase at pressures higher
than 3.5 MPa [84–86]. This discrepancy between the reports is
attributed to the different methods of synthesis and desolvation
of MOF-5 before mechanical compression [68]. Siegel’s group has
densified MOF-5 by mechanical compression [84–86]. The desol-
vated MOF-5 powder was manually pressed for 1 min to achieve
MOF-5 bulk densities between 0.27 and 0.79 g cm
3
[84]. MOF-5
progressively transformed from crystalline to amorphous due to
compression. However, the MOF-5 pellets with a bulk density of
0.75 g cm
3
retained a considerable fraction of the crystalline
phase. Mechanical compression reduced the surface area of MOF-
5. There was a 2% decrease in surface area for MOF-5 pellets with
a density of 0.31 g cm
3
compared to MOF-5 powder. The surface
area of MOF-5 pellets with densities of 0.51 g cm
3
and 0.90 g cm
3
decreased by 18 and 57%, respectively. Mechanical compression
also reduced the total pore volume of MOF-5. This reduction in sur-
face area and total pore volume was due to amorphization. Com-
paction of MOF-5 produced modest reductions in its gravimetric
H
2
storage capacity, but large increases in its volumetric H
2
storage
capacity [84]. The total volumetric H
2
storage capacity of the MOF-
5 pellet with a density of about 0.5 g cm
3
was 44 g L
1
at 10 MPa
and 77 K. MOF-5-based composite pellets containing 1 to 10 wt%
expanded natural graphite (ENG) with densities of 0.3, 0.5, or
0.7 g cm
3
have been prepared to improve the thermal conductiv-
ity of MOF-5 (Fig. 2)[85,86]. It is worth mentioning that MOFs
often exhibit very low thermal conductivity, which limits their
applicability as cryo-adsorption-based H
2
storage systems [90].
For example, low thermal conductivity can hinder fast refueling
for adsorption that depend on a pressure or temperature-swing
[86]. MOF-5-based composite pellets with three bulk densities of
approximately 0.3, 0.5 and 0.7 g cm
3
were prepared by applying
approximately 9, 33 and 60 MPa, respectively. While the diffrac-
tion intensity of the MOF-5 pellet decreased by 50% from 0.49 to
0.67 g cm
3
, the diffraction intensity of composite pellet contain-
ing 10 wt% ENG decreased only 38% with a similar increase in den-
sity. Therefore, the addition of ENG protected the structural
crystallinity of MOF-5 during mechanical compression. ENG also
prevented MOF-5 from losing textural properties. Pellets made of
MOF-5 and 10 wt% ENG with density of 0.5 g cm
3
were found
to improve thermal conductivity by a factor of 5 relative to pure
MOF-5 pellets of the same density. Specifically, the thermal con-
ductivity increased from 0.10 W mK
1
for MOF-5 pellets to
0.56 W mK
1
for composite pellets at room temperature. At 77 K
and 10 MPa, the total volumetric H
2
storage capacity of the com-
posite pellet with a density 0.5 g cm
3
was 6% less than that of
the pure MOF-5 pellet of equal density, which, in turn, exhibited
23% higher total volumetric H
2
storage capacity than that of the
MOF-5 powder [85,86]. Therefore, the addition of ENG to MOF-5
can improve thermal conductivity while maintaining the favorable
H
2
storage properties of MOF-5.
Fig. 2. Cylindrical pellets of MOF-5 and MOF-5/ENG composites. (a) MOF-5/ENG
composite pellet with 6.35 mm diameter and 4.9 mm thickness. Pure MOF-5 Pellets
with (b) 12 mm diameter and 1 mm thickness and (c) 12 mm diameter and 2 mm
thickness. Adapted with permission from ref. [86]. Copyright 2012 Elsevier.
Fig. 1. Schematic illustration of MOF powder compression.
J. Fonseca and T. Gong Coordination Chemistry Reviews 462 (2022) 214520
4
Most Zn-based MOFs, such as MOF-177 and MOF-5, are sensi-
tive to moisture [91–93]. However, the Cr-based MOF, MIL-101
(Cr) (Cr
3
O(OH)(H
2
O)
2
(BDC)
3
), is very stable when exposed to
humidity and high temperature. This MOF can be readily regener-
ated by activation if adsorbed water decreases its H
2
storage capac-
ity. Furthermore, MIL-101(Cr) can be synthesized using just water.
Anhydrous organic solvents are not required for the purification of
MIL-101(Cr), which favors scaling-up its synthesis. Thus, from a
technological point of view, MIL-101(Cr) is considered a promising
material for H
2
storage. In various studies, MIL-101(Cr) powder has
been manually compressed to produce MIL-101(Cr) pellets with
remarkable mechanical and thermal stability [87,88]. There were
no notable structural alterations with compression up to densities
of 0.47 g cm
3
. The textural properties of densified MIL-101(Cr)
were found to be closer to those of MIL-101(Cr) powder when
the compression rate was low. In other words, the textural proper-
ties of MIL-101(Cr) were closer to those of MIL-101(Cr) powder
when the time to reach a bulk density of 0.467 g cm
3
was
6 min than when it was 2 min. Specifically, the surface area and
pore volume decreased by 10.2 and 10.8% when the compression
time was approximately 6 min. However, when the compression
time was approximately 2 min, the surface area and pore volume
decreased 38.8 and 38.3%, respectively. The total volumetric H
2
storage capacity of MIL-101(Cr) pellets (high compression rate)
with a density varying between 0.45 and 0.47 g cm
3
was
40 g L
1
at 8 MPa and 77 K [87]. The total volumetric H
2
storage
capacity of MIL-101(Cr) pellets (low compression rate) with a den-
sity of 0.467 g cm
3
was 43.7 g L
1
at 15 MPa and 77 K [88].
UiO-66 (Zr
6
O
4
(OH)
4
(BDC)
6
) exhibits unusually high shear sta-
bility [94]. The minimum shear modulus of UiO-66 (G
min
= 13.7 GPa)
is an order of magnitude higher than that of other benchmark
MOFs, such as MOF-5, ZIF-8, and HKUST-1 (Cu
3
(BTC)
2
,H
3
BTC = ben-
zene tricarboxylic acid). The exceptional mechanical stability of
UiO-66 is attributed to its high framework connections [94]. There-
fore, UiO-66 is an outstanding candidate for large-scale applica-
tions. UiO-66 powder has been compacted at 150, 290, 440, 590,
and 665 MPa for 15 min without structural collapse of the crystal
lattice [95]. At a pressure as high as 665 MPa, the compacted UiO-
66 showed a BET surface area of 1707 m
2
g
1
, which is less than 2%
decrease compared to the UiO-66 powder (1737 m
2
g
1
). The total
pore volume of UiO-66 powder and compacted UiO-66 was
0.96 cm
3
g
1
and 0.81 cm
3
g
1
, respectively. There was a 16%
reduction in total pore volume, which was attributed to the reduc-
tion of macropores after compaction. In other words, the interpar-
ticle voids were eliminated during compaction due to the close
packing of the particles. Total gravimetric H
2
uptake was not
compromised by densification. At 100 bar and 77 K, the total gravi-
metric H
2
uptake of the compacted UiO-66 and the UiO-66 powder
was 5.1 wt% and 5.0 wt%, respectively. Therefore, the textural and
adsorption properties of the UiO-66 powder were preserved after
compaction. The total volumetric H
2
capacity was higher for the
compacted UiO-66 than for UiO-66 powder. The densified UiO-66
exhibited a volumetric H
2
capacity of up to 74 g L
1
(100 bar,
77 K) and 13 g L
1
(100 bar, 298 K), whereas UiO-66 powder
showed 29 g L
1
(100 bar, 77 K) and 6 g L
1
(100 bar, 298 K). These
capacities were determined by using the packing density of the
adsorbents. Using the model developed by Ahmed et al. [96] the
volumetric H
2
capacity was 43 g L
1
(100 bar, 77 K) and 6 g L
1
(100 bar, 298 K) for compacted UiO-66 and 35 g L
1
(100 bar,
77 K) and 6 g L
1
(100 bar, 298 K) for UiO-66 powder. In brief,
the total volumetric H
2
storage capacity improved by densification,
regardless of the model to determine this parameter [95]. Hence,
the compaction of highly robust MOFs can improve their total vol-
umetric H
2
storage capacity, without causing anisotropic changes
in the crystal topology [97] and/or reducing the surface area [98]
and therefore without compromising the total gravimetric uptake.
MOFs are considered exceptional systems for the storage of CH
4
[81,99,100]. However, there are still challenges to overcome:
(1) to find MOFs that pack efficiently, but can withstand sub-
stantial mechanical pressure and retain full adsorption
capacity, and
(2) to find non-destructive ways to densify high-porosity MOFs
[101].
HKUST-1 has been densified into pellets by compression [101].
The crystal structure of HKUST-1 partially collapsed under pres-
sure (Fig. 3b). The micropore volume of the HKUST-1 pellets was
found to decrease significantly as the compression pressure
increased (Fig. 3a). Although the density of the HKUST-1 pellets
increased with pressure, their total volumetric CH
4
uptake
decreased due to their loss of porosity (Fig. 3c). In brief, the densi-
fication of HKUST-1 partially amorphized the framework, signifi-
cantly reducing both volumetric and gravimetric CH
4
uptake [101].
MOF-5 pellet has also been evaluated for CH
4
uptake. MOF-5
was densified by mechanical compression at different pressures
in a dry inert atmosphere [33]. MOF-5 pellets with a density of
0.2569, 0.3151, 0.4612, and 0.4890 g cm
3
were prepared by
applying pressures of 0, 0.5, 1.0, and 2.0 MPa, respectively. The
evolution of the structural crystallinity of MOF-5 with pressure
was not reported. The BET surface area of MOF-5 decreased as
the applied pressure increased. For example, while the BET specific
Fig. 3. (a) N
2
isotherms of HKUST-1 powder and pressed pellets. The pore volume decreased as the pressure increased. (b) X-ray diffraction patterns of HKUST-1 powder and
pellets. As pressure increased, peak intensities decreased, and peak widths increased. (c) Total volumetric CH
4
uptake by HKUST-1 for different bulk densities. Insets: top:
photograph of HKUST-1 powder packed in a 1 ml syringe; bottom: pressed wafer. Adapted with permission from ref. [101]. Copyright 2013 American Chemical Society.
J. Fonseca and T. Gong Coordination Chemistry Reviews 462 (2022) 214520
5
surface area of the MOF-5 powder was 2777 m
2
g
1
, the BET sur-
face area of the pellet after pressing 2.0 MPa was 1450 m
2
g
1
.
Mechanical compression also decreased the gravimetric CH
4
stor-
age capacity of MOF-5. This was attributed to the reduction of
adsorption sites due to the partial collapse of the micropore net-
work of MOF-5. The pellet (pressed at 1.0 MPa) exhibited an effec-
tive volumetric CH
4
storage capacity of 81 v(STP)/v at 3.69 MPa
and 300 K, which is much lower than that of the theoretically per-
fect MOF-5 sample (135 v(STP)/v) with crystallographic density
(0.59 g cm
3
)[33].
CPO-27-Ni (Ni
2
(dhtp), H
4
dhtp = 2,5-dihydroxyterephthalic acid)
has been shaped by applying a mechanical pressure of 0.1 and
1 GPa for 30 min [102]. The crystal structure of CPO-27-Ni did
not change at a pressure of 0.1 GPa. However, there was a strong
loss of crystallinity at a pressure of 1 GPa. This decrease in crys-
tallinity was attributed to the breakdown of the MOF structure.
The specific surface area of the CPO-27-Ni pellet prepared at
0.1 GPa (1050 m
2
g
1
) was only slightly lower than that of the
CPO-27-Ni powder (1060 m
2
g
1
). Therefore, the textural proper-
ties were mainly preserved after mechanical compression at
0.1 GPa. However, the textural properties were completely lost
after mechanical compression at 1 GPa. At 303 K and 34 bara,
the CH
4
adsorption of the CPO-27-Ni pellet prepared at 0.1 GPa
(129 Ncm
3
g
1
(9% wt/wt)) was slight lower than that of the
CPO-27-Ni powder (157 Ncm
3
g
1
(11% wt/wt)). Therefore, the
small structural and textural modifications of the CPO-27-Ni pellet
was found to produce a reduction of its CH
4
uptake capacity [102].
The adsorption of NH
3
and CNCl in UiO-66-NH
2
(Zr
6
O
4
(OH)
4
(BDC-NH
2
)
6
,H
2
BDC-NH
2
= 2-aminoterephthalic acid) pellets has
been evaluated by Peterson et al. [103] UiO-66-NH
2
was pelletized
by mechanical compression at pressures ranging from 5000 to
100,000 psi. The crystal structure remained intact up to
25,000 psi. The MOF underwent amorphization at pressures above
25,000 psi. The BET surface area decreased 5.5, 13.9, 25.9 and 83%
by pressing the powder at 5000, 10,000, 25,000 and 100,000 psi,
respectively. Similarly, the total pore volume was found to
decrease as the pelletizing pressure increased. In contrast, the pore
size distribution remained constant even up to 100,000 psi. In both
dry and humid conditions, the NH
3
and CNCl uptake capacity of the
densified UiO-66-NH
2
decreased as the palletization pressure
increased. These results were attributed to mass transfer limita-
tions within the material due to the lack of a hierarchical pore
structure [103]. This research group has also evaluated NH
3
adsorption in HKUST-1 pellet. HKUST-1 was pelletized by pressing
at 1000 and 10,000 psi for 1 min [98]. HKUST-1 was unable to form
a pellet when pressed at 1000 psi. That pressure was not enough to
provide an engineering material. In fact, the density was kept con-
stant for HKUST-1 powder and HKUST-1 pressed at 1000 psi. How-
ever, the density of HKUST-1 pellet (pressed at 10,000 psi) was 37%
higher than that of HKUST-1 powder. The crystal structure
remained almost intact after pelletization. The surface area
decreased from 1698 m
2
g
1
for HKUST-1 powder to 892 m
2
g
1
for HKUST-1 pellet. This reduction in surface area was attributed
to a localized structural collapse generated during the densification
process. The total NH
3
loading of HKUST-1 pellet was similar to
that of HKUST-1 powder. Therefore, the active sites were still
accessible for NH
3
adsorption although the surface area and pore
volume decreased. In addition, octane adsorption in UiO-66 pellet
has also been explored in this study. UiO-66 was pelletized by
pressing at 1000 and 10,000 psi for 1 min. The crystal structure
and textural properties were maintained during pressing. UiO-66
powder exhibited higher octane adsorption capacity than UiO-66
pellet pressed at 1000 psi, which, in turn, showed higher adsorp-
tion than UiO-66 pellet pressed at 10,000 psi. Therefore, the
adsorption capacity decreased as the pelletization pressure
increased [98]. UiO-66 pellet has also been evaluated for xylene
separation in liquid phase [104]. UiO-66 powder was first uni-
formly mixed with 1 wt% graphite powder and subsequently com-
pressed to form a UiO-66 pellet with a density of 0.8 g cm
3
. The
crystal structure was found to be partially amorphous after com-
pression. Moreover, the surface area decreased by 22% during pel-
letization, from 1140 m
2
g
1
for the UiO-66 powder to 885 m
2
g
1
for the UiO-66 pellet. Both UiO-66 powder and UiO-66 pellet
exhibited preference for o-xylene over m- and p-xylene in the liq-
uid phase. However, UiO-66 pellet showed less adsorption capacity
than UiO-66 powder. This can be attributed to the loss of structural
and textural properties during the densification process [104].
Dhainaut et al. have investigated in detail the shaping of UiO-
66, UiO-66-NH
2
, UiO-67 (Zr
6
O
4
(OH)
4
(BPDC)
6
,H
2
BPDC = 2,2
0
-bipyri
dyl-4,4
0
-dicarboxylic acid), and HKUST-1 powders into tablets by
applying pressure [105]. The UiO-66 powder (BET surface
area = 1426 m
2
g
1
; bulk density = 0.17 g cm
3
) was compressed
at 18 MPa, producing a macroscopic UiO-66 architecture with a
BET surface area of 1459 m
2
g
1
and a bulk density of
0.43 g cm
3
. Similarly, UiO-66-NH
2
(BET surface area = 839 m
2
g
1
;
bulk density = 0.41 g cm
3
), UiO-67 (BET surface area = 2034 m
2
g
1
;
bulk density = 0.25 g cm
3
), and HKUST-1 (BET surface
area = 1288 m
2
g
1
; bulk density = 0.48 g cm
3
) powders were
compressed at 164, 63, and 121 MPa, respectively, thus
yielding UiO-66-NH
2
(BET surface area = 625 m
2
g
1
; bulk
density = 0.93 g cm
3
), UiO-67 (BET surface area = 1549 m
2
g
1
;
bulk density = 0.62 g cm
3
), and HKUST-1 (BET surface
area = 1091 m
2
g
1
; bulk density = 0.90 g cm
3
) macroscopic archi-
tectures. Therefore, no loss of textural properties was shown in
densified UiO-66. Discrepancies have been observed in the case
of HKUST-1. Herein, it was reported that up to 85% of the BET sur-
face area of the powder was preserved in densified HKUST-1 [105].
However, Kim et al. reported BET surface area losses ranging
between 34 and 96% in densified HKUST-1, which was prepared
by applying pressures ranging from 2.5 to 34 MPa [34]. These dis-
crepancies can be attributed to:
(1) Different compression protocols. Dhainaut et al. used a rela-
tively slow compression rate that could allow the polydis-
perse microcrystalline powders to rearrange during
compression.
(2) Kim et al. activated HKUST-1 powder before compression,
while Dhainaut et al. did not. The presence of solvent within
a framework can improve its mechanical resilience, which
helps to maintain its integrity during compression [106].
(3) The presence of defects within a MOF structure can reduce
its resistance during compression.
In addition, Dhainaut et al. also investigated the role of ENG as a
binder [105]. HKUST-1 powder and ENG were compressed to pre-
pare a densified HKUST-1 tablet. Up to 2 wt% ENG was found to
reduce the pressure required to achieve certain robustness without
impacting the BET surface area [105].
Bazer-Bachi et al. have also explored the pelletization of ZIF-8,
HKUST-1 and SIM-1 (Zn(4-m-5-ica)
2
, 4-m-5-ica = 4-methyl-5-imi
dazolecarboxaldehyde) at different pressures [107]. These frame-
works lost crystallinity during compression. Amorphization was
consistent with a decrease in the BET surface area. Compression
caused a partial structural collapse in these frameworks. The struc-
tural collapse was reported to be proportional to the applied com-
pression force. The impact of compression was higher in HKUST-1
than in ZIF-8 and SIM-1. Thus, the degree of amorphization was
found to depend on the nature of MOF. Furthermore, to prepare a
resilient ZIF-8 pellet, ZIF-8 was pelletized together with an organic
binder (cellulose ester, methocel
TM
K15M) at different pressures.
This densified ZIF-8 pellet exhibited higher mechanical properties
and crush strength than pellets prepared without the binder.
J. Fonseca and T. Gong Coordination Chemistry Reviews 462 (2022) 214520
6
Amorphization similar to that of samples prepared without binder
was followed. However, the structural and textural changes did not
modify the acid-base properties of ZIF-8. While the surface area of
ZIF-8 was reduced, its catalytic activity for the conversion of veg-
etable oil into monoglyceride remained constant. This observation
was consistent with the hypothesis of an external surface reactiv-
ity of ZIF-8 [107].
MOF pellets have also been investigated as adsorbents for the
separation of paraffin isomers [108], to remove low concentrations
of Xe and Kr from spent nuclear fuels [109], and for many other
important issues in the petrochemical industry [110,111]. How-
ever, the influence of the densification of these MOFs on their
structural and textural properties and how this densification
affects their applicability has not been explored in many studies.
2.2. Mechanical granulation
Granulation follows the same principles as pelletization. MOF
granulates are formed by applying pressure to the MOF powder
with or without adding a binder. The crystal structure of MOFs
can be destroyed during the process. The difference between
pelletization and granulation relies on the shape of the resulting
densified MOFs, spherical for granulates and cylindrical for pel-
lets. The shape of densified MOFs can play a crucial role in some
industrial processes [39]. The granulation process is referred to
as wet or dry granulation depending on the addition or not of
binder. In wet granulation, a binder, such as PVA [34], silica
[45], sucrose [77], graphite [104], chitosan [112], cellulose ester
[107], and silicone resin [113], may be used to agglomerate MOF
powder particles into handleable granulates [114]. The surface
tension and viscosity of the binder and the binder-MOF interac-
tions strongly influence the properties of the MOF granulates
[39]. Therefore, the choice of binder is critical. Three successive
steps are distinguished in wet granulation (Fig. 4): (1) wetting
and nucleation, (2) consolidation and coalescence, and (3) attri-
tion and breakage. As mentioned above, binders can improve the
mechanical stability and/or thermal conductivity of densified
MOFs. However, they can also decrease the accessible surface
area of MOFs and thus hinder mass transfer within MOFs
[45,76]. Therefore, binders can reduce the catalytic and/or
adsorption capacities of MOFs. In dry granulation, densified
materials are formed by applying only high pressure. Dry
Fig. 4. Schematic illustration of wet granulation.
Fig. 5. SEM images of (a) MIL-125-NH
2
granulate, (b) surface of the granulate, and (c) interior of the granulate. (d) Photograph of MIL-125-NH
2
granulate. Adapted with
permission from ref. [115]. Copyright 2015 American Chemical Society.
J. Fonseca and T. Gong Coordination Chemistry Reviews 462 (2022) 214520
7
granulation could be used when the MOF is sensitive to solvents
and/or heat.
UiO-66, UiO-66-NH
2
, MIL-100(Fe) (Fe
3
O(H
2
O)
2
OH(BTC)
2
), and
MIL-127(Fe) (Fe
3
O(OH)(H
2
O)
3
(TazBz)
1.5
, TazBz = 3,3
0
,5,5
0
-azoben
zene-tetracarboxylate) have been densified by wet granulation
[115,116]. Specifically, mixtures of MOF and polyvinyl-based bin-
der (3 wt%) were shaped by a granulator. To induce agglomera-
tion of microcrystalline powders, solid mixtures were
periodically wetted with ethanol (EtOH). The BET surface area of
the MOF granulates was less than that of the MOF powders. Micro-
porosity decreased approximately 3% in the UiO-66 and MIL-127
(Fe) samples, which was related to the addition of the binder. How-
ever, the UiO-66-NH
2
and MIL-100(Fe) samples showed greater
decreases in microporosity, which was attributed to the partial
degradation of their structure. The bulk density of the MOFs
increased with this granulation. These MOF granulates were used
as adsorbents for a series of gases (N
2
, CO, CO
2
,CH
4
,C
2
H
6
,C
3
H
8
,
C
3
H
6
,C
4
H
10
)at30°C. The granulates were found to exhibit lower
total gravimetric uptakes and higher total volumetric uptakes than
MOF powders. The wet granulation procedure has also been
applied to prepare a MIL-125(Ti)-NH
2
(Ti
8
O
8
(OH)
4
(BDC-NH
2
)
6
)
granulate (Fig. 5). This densified MOF was studied for syngas treat-
ment [115]. Ren et al. have also shaped UiO-66 into granulates
[77]. UiO-66 powder was mixed with sucrose (10 wt%), which
acted as a binder. This mixture was granulated under centrifugal
force with water spray. The crystal structure of UiO-66 was
retained after centrifugal granulation. The size of the UiO-66-
based granulates depended on the shaping time under the cen-
trifugal force. The porosity of the UiO-66-based granulates
decreased compared to the UiO-66 powder. BET surface area and
pore volume were found to decrease by 50.7% and 39.3%, respec-
tively. Sucrose was suggested to block the pores of UiO-66. The
decrease in porosity was higher than that reported in the previous
study when the binder was a polyvinyl based material [115]. The
granulates were found to have interparticle macropores between
the UiO-66 crystals. UiO-66-based granulates exhibited mechani-
cal strength to resist abrasions in a real H
2
storage environment.
However, the H
2
storage capacity decreased by 44.8% from powder
(1.54 wt%) to granulates (0.85 wt%), which was correlated with the
decrease in porosity [77]. Kim et al. have also prepared MIL-100
(Fe)-based granulates via conventional wet granulation [45]. These
architectures were developed by mixing MIL-100(Fe) powder and a
silica sol solution (10 wt%) in a mixing granulator. Unlike the pre-
viously mentioned MIL-100(Fe)-based granulates where the binder
was a polyvinyl based material [115], the crystalline structure of
MIL-100(Fe) was not affected by the granulation process
when silica sol acted as a binder. The porosity of the granulates
decreased slightly compared to that of MIL-100(Fe) powder. BET
surface area and total pore volume decreased from 1772 m
2
g
1
and 0.93 cm
3
g
1
for MIL-100(Fe) powder to 1619 m
2
g
1
and
0.82 cm
3
g
1
for granulates. While the bulk density of the MIL-
100(Fe) powder was 331 g L
1
, the bulk density of the granulates
was 498 g L
1
. MIL-100(Fe)-based granulates were tested as SF
6
adsorbents. SF
6
adsorption capacity only slightly decreased from
MIL-100(Fe) powder (1.673 mmol g
1
) to MIL-100(Fe)-based
granulates (1.658 mmol g
1
). Multiple cycles of adsorption and
desorption were carried out for the separation of SF
6
/N
2
. These
granulates were found to regenerate easily compared to MIL-100
(Fe) powder [45].
2.3. Mechanical extrusion
Mechanical extrusion has been widely used to manufacture
structures based on zeolites, silicates, metals, and metal oxides,
with loads of up to 100% (materials without binders) [117–122].
This method relies on pushing a material powder, herein MOF
powder, through a die in the shape of the desired structure at high
temperature (Fig. 6)[32]. The heat treatment step must be
adjusted to avoid burning the MOF powders. Optimizing the rheol-
ogy of MOF-based putty is essential to forming extruded architec-
tures. Mechanical extrusion produces higher volumes of
macroscopic MOF architectures than mechanical compression
and granulation. However, the mechanical extrusion of MOF pow-
ders has been limited to a few reports. It is worth noting that the
architecture of MOFs depends on the extrusion molds used. There-
fore, the molds must be pre-manufactured to generate the desired
geometry in the MOF architecture. Furthermore, ultra-complex
geometries cannot be produced. This manufacturing approach is
only capable of formulating simple structures [43].
Monolithic HKUST-1 has been prepared by extrusion (Fig. 7)
[46]. HKUST-1 powder was first mixed with a binding agent (the
methoxy functionalized siloxane ether SILRES
Ò
MSE 100 (CH
3
Si
(O)
1.1
(OCH
3
)
0.8
)) and a plasticizer (the methyl hydroxyl propyl cel-
lulose Culmial MHPC 20,000P), forming a homogenous paste. This
paste was extruded to fabricate HKUST-1-based monolithic struc-
tures. These structures were treated by microwave drying for
20 min and finally dried at 120 °C for at least 8 h. The plasticizer
was not removed at 120 °C; however, the porosity decreased at
higher temperatures probably due to the partial degradation of
HKUST-1. The resulting monolithic HKUST-1-based architectures
exhibited a surface area of 484 m
2
g
1
and a high mechanical sta-
bility of 320 N [46].
Fig. 6. Schematic illustration of mechanical extrusion.
Fig. 7. Picture of the HKUST-1-based monolith. Adapted with permission from ref.
[46]. Copyright 2010 John Wiley and Sons.
J. Fonseca and T. Gong Coordination Chemistry Reviews 462 (2022) 214520
8
MIL-101(Cr)-based monoliths have also been manufactured by
extrusion [123]. First, MIL-101(Cr) powder and bentonite clay were
mixed in water to form an extrudable paste. MIL-101(Cr)-based
monoliths were then prepared by extruding the paste. The
bentonite clay acted as a binder. The monoliths were dried at
10 °C for several days and then treated at 150 °C for approximately
33 h to form a strong, solid monolithic structure. The heat treat-
ment was suggested to favor the linking MIL-101(Cr) crystals-
binder. The heat-treated monolithic structures contained 60 wt%
of MIL-101(Cr). The crystalline structure of the MIL-101(Cr) pow-
der was maintained in the monolithic structures. The porosity of
the monoliths was lower than that of the as-synthesized MIL-101
(Cr) powder. Specifically, the BET surface area of the monoliths
(182.84 m
2
g
1
) was 47.08% less than that of the powder
(345.49 m
2
g
1
). This was attributed to a partial degradation of
MIL-101(Cr) during treatment and purification, as well as the pres-
ence of non-porous binder. The presence of bentonite also
increased the bulk density of MIL-101(Cr) monoliths compared to
MIL-101(Cr) powders. Furthermore, a higher proportion of binder
in the monoliths produced mechanically stronger structures.
MIL-101(Cr)-based monoliths were tested for CO
2
adsorption.
These monoliths exhibited a CO
2
adsorption capacity comparable
to that of the MIL-101(Cr) powder. In addition, MIL-101(Cr) mono-
liths were regenerated at a moderate temperature of 150 °C for
repeated cycles of CO
2
adsorption without loss of performance
[123].
MIL-53(Al(OH)(BDC))-based architectures have also been pre-
pared by extrusion [124]. Two different types of methyl cellulose
(differing in molecular weight, viscosity, etc.) were used as binders.
First, MIL-53 and methyl cellulose powders were dry mixed. A mix-
ture of deionized water and EtOH (50/50 vol%) was added drop-
wise to obtain a viscous and homogeneous paste. This paste was
extruded to yield MIL-53-based architectures. The crystallinity of
MIL-53 was maintained in the architectures. The MIL-53 crystals
within the architectures were able to reversibly undergo lp-np
phase transition (breathing effect). These architectures exhibited
porosity. Their pore volumes decreased only slightly with increas-
ing binder fraction. The reduction in pore volume was moderate
even with the highest binder content of 10 wt%. The type of methyl
cellulose had a minor impact on the textural properties of the MIL-
53-based architectures obtained. The architectures with a binder
content of 5 or 10 wt% achieved mechanical stabilities comparable
to those reported in the literature for other extruded porous mate-
rials such as titanium silicalite or SBA-15 [124–126].
2.4. Spray drying
In spray drying, a solution (or slurry) of MOF reagents is ato-
mized (generally by a nozzle) and then dried to form MOF granu-
lates (Fig. 8). The diameter of the MOF granulates is varied by
adjusting the flow rate at the spray nozzle and the speed of the
drum. The granulates are sieved when they leave the drum.
Whereas the smaller granulates are reintroduced into the drum
to increase their size, the larger granulates are ball milled to reduce
their diameter [43]. This technique allows the rapid production of
densified MOF granulates with fine control of their size. However,
it can only form small architectures, which limits its applicability.
Nano-sized MOFs have been assembled by spray-drying, form-
ing sub-5
l
m spherical hollow superstructures with controllable
morphologies and sizes [47]. A MOF precursor solution (containing
metal ions and organic ligands) was atomized into microdroplets.
After evaporation of the solvent, the precursors diffused radially
to the droplet surface, which favored the nanoMOFs to crystallize
at the surface. The nanoMOFs assembled into a well-packed shell
referred to as the hollow MOF superstructure. This approach was
performed to prepare hollow superstructures based on the follow-
ing MOFs: HKUST-1; CuBDC; NOTT-100 (also called MOF-505;
Cu
2
(bptc), H
4
bptc = 3,3
0
,5,5
0
-biphenyl-tetracarboxylic acid); MIL-
88A (Fe
3
O(fumarate)
3
); MIL-88B (Fe
3
O(BDC-NH
2
)
3
); MOF-14
(Cu
3
(BTB)
2
); MOF-74 (also called CPO-27; M
2
(dhtp), where M is
Zn, Ni and Mg); UiO-66; ZIF-8; Cu(II) Prussian blue analogue (CuPB,
Cu
3
[Fe(CN)
6
]
2
); MOF-5; and IRMOF-3 (Zn
4
O(BDC-NH
2
)
3
). The MOF
superstructures were disassembled into discrete, homogeneous
nanoMOFs crystals by sonication. Hollow multicomponent MOF
superstructures were also prepared by spray-drying. Specifically,
a hollow superstructure made of nanoMOF-74 containing both
Ni
2+
and Zn
2+
ions was synthesized. All the hollow superstructures
were found to retain the textural properties of their constituent
nanoMOFs. These hollow MOF superstructures allowed the encap-
sulation of active species [47].
In a subsequent study, microspherical MOF beads have been
fabricated by incorporating a continuous flow reactor at the inlet
of the spray-drier [127]. First, the precursor solution was injected
into a continuous flow reactor, which was heated to a certain tem-
perature to promote secondary building unit (SBU) formation and
nucleation. The heated solution was then automatically injected
into the spray-drier. The solution was atomized at a certain tem-
perature. Therefore, the growth of MOFs was limited to individual
microreactors (microdroplets). This process produced MOF archi-
tectures in the form of beads. These microscale beads were made
of multiple MOF nanoparticles (NPs). They were highly compact,
unlike the previously mentioned hollow superstructures [47]. This
was attributed to the formation, inside the reactor, of a suspension
containing a primary nucleus. The rate at which the nucleus dif-
fused to the surface was less than the rate at which the nucleus
grew during the drying-evaporation process. Several MOF, such
as UiO-66, UiO-66-NH
2
, UiO-66-NO
2
(Zr
6
O
4
(OH)
4
(BDC-NO
2
)
6
,
H
2
BDC-NO
2
= 2-nitroterephthalic acid), UiO-66-acetamido
(Zr
6
O
4
(OH)
4
(BDC-acetamido)
6
,H
2
BDC-acetamido = 2-acetamidoter
ephthalic acid), UiO-66-Br (Zr
6
O
4
(OH)
4
(BDC-Br)
6
,H
2
BDC-Br = 2-b
romoterephthalic acid), UiO-66-(OH)
2
(Zr
6
O
4
(OH)
4
(dhtp)
6
),
UiO-66–1,4-NDC (Zr
6
O
4
(OH)
4
(1,4-NDC)
6
, 1,4-NDC = 1,4-naphthale
nedicarboxylate), UiO-66–2,6-NDC (Zr
6
O
4
(OH)
4
(2,6-NDC)
6
, 2,6-N
DC = 2,6-naphthalenedicarboxylate), MIL-100(Fe) and [Ni
8
(OH)
4
(H
2
O)
2
(L)
6
]
n
(L = 1H-pyrazole-4-carboxylic acid), were assembled
into microspherical beads using this spray-drying continuous
Fig. 8. Schematic illustration of spry drying process.
J. Fonseca and T. Gong Coordination Chemistry Reviews 462 (2022) 214520
9
flow-assisted synthesis. The microporosity of these beads was sim-
ilar to that of their constituent MOFs. In addition, microspherical
multivariate (MTV) MOF beads were also prepared using the
spray-drying continuous flow-assisted approach [127].
3. Pseudomorphic replication
Pseudomorphic mineral replication processes are geological
transformations, in which one mineral phase is converted into
another through a chemical reaction. While the parent phase is
out of equilibrium, the new phase is more thermodynamically
stable. This phenomenon is characterized by the retention of the
morphology of the parent phase. It involves dissolution and re-
precipitation sub-processes [128,129]. In fact, pseudomorphic
replication occurs when the dissolution kinetics of the parent
phase are coupled with the nucleation and crystallization kinetics
of the new phase. Applying this natural process using pre-shaped
sacrificial supports is an attractive approach to introducing a
structural hierarchy into materials [130,131].
Pseudomorphic replication, which is also known as coordination
replication, is a promising method to synthesize porous macro-
scopic MOF architectures [132–135]. A pre-shaped metal source,
such as metal oxide, metal hydroxide, and metal oxyhydroxide
monoliths, becomes a thermodynamically more stable MOF archi-
tecture in the presence of organic ligands. This phenomenon occurs
when the dissolution kinetics of the sacrificial metal source are cou-
pled with the nucleation and crystallization kinetics of the MOF.
Local dissolution of the pre-shaped metal phase provides metal
ions, which are immediately consumed by the formation of MOFs
at metal phase/solution interfaces. Thus, the morphology of the par-
ent phase is preserved in the MOF architecture. Therefore, the pre-
shaped metal phase not only acts as a source of metal, but also as an
‘architecture-directing agent’ (Fig. 9).
Although pseudomorphic replication can expand the scientific
ability to create macroscopic MOF architectures with exceptional
functionalities, this technique has so far been limited to Al
2
O
3
[136–138], Cu(OH)
2
[139–142],Cu
2
O[143],Cu
2
(OH)
3
(NO
3
)[144],
ZnO [48,49,145–147],V
2
O
5
[148], and CaCO
3
[149] monolithic par-
ent sources. Therefore, the scope of coordination replication is
restricted by the proper preparation of pre-shaped metal-based
architectures.
In this Section, we extensively explore the formation of macro-
scopic MOF architectures via coordination replication (Table S3).
Many studies are reviewed paying special attention to the synthe-
sis protocol. The applicability of these reported macroscopic MOF
architectures is also mentioned to show the tremendous possibili-
ties of these materials.
3.1. Pseudomorphic replication of Al-based monolithic solids
In an early example, Kitagawa, Furukawa, and coworkers pre-
pared various Al-based MOF architectures with hierarchical poros-
ity via pseudomorphic replication [136], Alumina phases were
selected as metal reservoir and architecture-directing agent. A
two-dimensional (2D) honeycomb alumina pattern [150], exhibit-
ing well-ordered hexagonal macro-voids, reacted with 1,4-NDC in
an aqueous solution under microwave irradiation at 180 °C for
10 min, forming an aluminum naphthalene dicarboxylate frame-
work, [Al(OH)(1,4-NDC)]
n
. A 3D alumina inverse opal structure
was similarly used as metal source and template [151]. Coordina-
tion replication led to the formation of MOF architectures with the
morphology of their alumina pattern. They exhibited hierarchical
porosity, combining the microporosity of the MOF and the macro-
porosity of the alumina architectures. Furthermore, the 3D alumina
inverse opal architecture also reacted with H
3
BTC and H
2
BDC pro-
ducing MIL-100(Al) (Al
3
O(H
2
O)
2
OH(BTC)
2
) and MIL-53, respec-
tively. MIL-100(Al) and MIL-53 architectures were analogous to
that of the alumina. The morphology preservation was explained
by the ‘dissolution-reprecipitation’ mechanism. In brief, the meta-
stable alumina pattern dissolved and immediately crystallized as a
MOF at the solid/liquid interface [152]. Therefore, the production
of MOF architectures through pseudomorphic replication occurred
when the dissolution kinetics of the metastable alumina phase
were coupled with the nucleation and crystallization kinetics of
the Al-based MOF phase [136]. This study also explored the
water/EtOH separation in two [Al(OH)(1,4-NDC)]
n
architectures
[136]. Two highly open alumina aerogels with different porosities
were chosen as metal reservoirs and architecture-directing agents
[153,154]. One aerogel exhibited mesoporosity and the other
meso- and macroporosity. Two [Al(OH)(1,4-NDC)]
n
architectures
with different hierarchical porosity were prepared by pseudomor-
phic replication. Whereas the coordination replication of the meso-
porous alumina aerogel resulted in the formation of a MOF
architecture with micro- and mesoporosity, the coordination repli-
cation of the meso- and macroporous alumina aerogel led a MOF
architecture with micro-, meso- and macroporosity. Therefore,
the MOF architectures preserved the meso- and macroporosity of
the aerogels while possessing the inherent microporosity of
[Al(OH)(1,4-NDC)]
n
. Both MOF architectures exhibited significant
water/EtOH separation. Thus, hierarchical porosity was found to
improve material selectivity and mass transfer for water/EtOH sep-
aration [136].
This research group has also prepared Al-based MOF nanofibers,
[Al(OH)(L)]
n
where L = 1,4-NDC or H
2
BDC, from alumina nanofibers
by coordination replication [137]. The alumina nanofiber was
heated in an aqueous solution of 1,4-NDC or H
2
BDC under micro-
wave irradiation at 180 °C for 1 min. In doing so, Al-based MOF
architectures were synthesized. They maintained the macroscopic
fibrous morphology of alumina. Therefore, the alumina nanofibers
acted as a solid-state metal source and structure-directing agent.
As occurs in polydisperse microcrystalline [Al(OH)(BDC)]
n
[155],
the as-synthesized closed-form of Al(OH)(BDC)]
n
crystals that con-
formed the nanofibers was transformed into open-form after ther-
mal treatment at 400 °C for 72 h in air. The pH of the ligand
Fig. 9. Schematic illustration of the pseudomorphic replacement (or coordination replication) process. The pseudomorphic replication process is based on the replacement of
a metastable monolithic metal phase (illustrated as a blue honeycomb pattern) by more stable MOF crystals (purple honeycomb pattern).
J. Fonseca and T. Gong Coordination Chemistry Reviews 462 (2022) 214520
10
solution affected the morphology of the Al-based MOF fibers.
Nanofiber morphology was not retained at low pH (pH 2.1). The
acidic medium accelerated the dissolution of the alumina becom-
ing faster than the crystallization of MOFs. Coordination replica-
tion was successful in a mild acidic environment (pH 4.3).
Deprotonation of dicarboxylate ligands was favored in a high pH
medium (pH 5.4). This resulted in a less anisotropic crystal mor-
phology. The MOF nanofibers exhibited hierarchical micro- and
mesoporosity. Therefore, the mesoporosity of the alumina nanofi-
bers was preserved after coordination replication. The microporos-
ity of the MOF nanofibers was provided by the MOF crystals [137].
A core–shell composite, made of a gold nanorod (GNR) core and
an aluminum-based MOF shell, has been prepared by coordination
replication of the precursor (GNR@alumina core–shell composite)
[138]. First, a hydrated amorphous alumina layer was deposited
on the surface of a GNR by a sol–gel process, thus producing a
GNR@alumina core–shell composite. This composite was
immersed in an aqueous solution of 1,4-NDC with a pH value of
2. Subsequently, this mixture was treated in the microwave reactor
at 180 °C for 1 min. In doing so, GNR@alumina core–shell compos-
ite became GNR@[Al(OH)(1,4-NDC)]
n
core–shell composite
(Fig. 10). The new composite consisted of a few intergrown MOF
nanocrystals aggregated around individual GNRs with a mean size
of 300 nm. The alumina species adhered to the GNR surface acted
as aluminum source. The amorphous alumina dissolved during the
microwave treatment. This dissolution, which occurred in the pres-
ence of 1,4-NDC, was immediately followed by the formation of [Al
(OH)(1,4-NDC)]
n
. Heterogeneous nucleation was localized on the
surface of the GNRs. Local Joule heating, which resulted from the
adsorption of microwave energy by the GNR, probably also con-
tributed to the nucleation of the MOF. A kinetic coupling between
the dissolution of the alumina and the crystallization of MOFs
allowed the precise localization of the MOF nucleation and, there-
fore, the preservation of the shape and dimension of the parent
phase. This GNR@[Al(OH)(1,4-NDC)]
n
core–shell composite was
used as an optical switch to control the release of molecules by
irradiation of light. The composite combined the unique loading
capacity of MOFs with the photothermal properties of GNRs.
Specifically, the light triggered the release of molecules attached
to the surface of the gold [138].
3.2. Pseudomorphic replication of Cu-based monolithic solids
Pseudomorphic replication has extended to copper chemistry. A
macroporous Cu(OH)
2
-based monolith was converted into a
HKUST-1 monolith with macro-, meso- and microporosity by coor-
dination replication [139]. The macroporous Cu(OH)
2
-based mono-
lith was immersed in a solution of H
3
BTC in N,N-
dimethylformamide (DMF) and EtOH (1:1 v/v) and treated at
80 °C, producing a HKUST-1 monolith. The conversion of Cu
(OH
2
))-based monolith to HKUST-1 (Cu
3
(BTC)
2
) monolith was as
follows:
3Cu OHðÞ
2
þ2H
3
BTC !Cu
3
BTCðÞ
2
þ6H
2
O (Eqn. 1).
Densely packed HKUST-1 crystals with the parent macroporous
architecture were obtained after a reaction time of 6 min. The BET
surface area of the MOF architecture was as high as 1315 m
2
g
1
.
The strength decreased from 2.5 MPa for the monolithic Cu
(OH)
2
to 1.5 MPa for the monolithic HKUST-1 (Cu
3
(BTC)
2
). How-
ever, the value was comparable to that of the periodic mesoporous
silica monolith. The kinetics of coordination replication was found
to depend on the dimension of the monolith. The adequate replica-
tion time increased as the dimension of the Cu(OH)
2
monolith
increased. This was related to the time required for the diffusion
of the ligand inside the monolith [139].
Kitagawa, Furukawa, et al. have prepared flexible Cu-based MOF
architectures using coordination replication [140]. A monolithic Cu
(OH)
2
-polyacrylamide (PAM) composite with meso- and macrop-
orosity became 3D architectures consisting of flexible MOFs, Cu
2
(-
BDC)
2
(MeOH)
2
and Cu
2
(BDC)
2
(bpy) (bpy = 4,4
0
-bipyridine). The
composite was first transformed into a Cu
2
(BDC)
2
(MeOH)
2
mono-
lith by coordination replication. Specifically, the Cu(OH)
2
-PAM
monolith was heated with a methanol (MeOH) solution of H
2
BDC
at 60 °C, thus forming the Cu
2
(BDC)
2
(MeOH)
2
monolith after 7 days.
The external dimensions and mechanical integrity of the Cu(OH)
2
-
PAM monolith were preserved in the Cu
2
(BDC)
2
(MeOH)
2
monolith.
The macroporosity of the Cu(OH)
2
-PAM monolith was lost due to
the increase in volume of Cu
2
(BDC)
2
(MeOH)
2
with respect to that
of Cu(OH)
2
. The BET surface area of the Cu
2
(BDC)
2
(MeOH)
2
mono-
lith was 520 m
2
g
1
. The PAM content of the Cu
2
(BDC)
2
(MeOH)
2
monolith was approximately 15.0 wt%. This polymer was found
to be a crucial component for the integrity of the 3D architecture.
The Cu
2
(BDC)
2
(MeOH)
2
monolith was converted to Cu
2
(BDC)
2
(bpy)
by the pillar ligand (bpy) insertion process. Cu
2
(BDC)
2
(MeOH)
2
was
treated with an excess of bpy in MeOH, resulting in the interpene-
trated Cu
2
(BDC)
2
(bpy) framework. The BET surface area of
Cu
2
(BDC)
2
(bpy) monolith was 1030 m
2
g
1
. The flexibility of
Cu
2
(BDC)
2
(bpy) was preserved, resulting in a flexible monolith sys-
tem. In brief, the framework changed from the open pore form to
the closed pore form by activation under a He flow at 150 °C.
Resolvation of the material in MeOH led to the return of open form
with retention of the monolithic architecture [140].
Hierarchical Cu-based MOF nanorod arrays have also been
developed by pseudomorphic replication [141]. MOF-2 (Zn
2
(-
BDC)
2
), Cu(BDC-Br), Cu(BDC-NH
2
), Cu(dhtp), Cu(1,4-NDC), Cu(2,6-
NDC) and Cu(BPDC) nanorod arrays were prepared by pseudomor-
phic replication using H
2
BDC, H
2
BDC-Br, H
2
BDC-NH
2
,H
4
dhtp, 1,4-
H
2
NDC, 2,6-H
2
NDC and H
2
BPDC as organic ligands, respectively.
Hierarchical nanorod@nanoplate core–shell architectures with
MOF nanoplates epitaxially grown on Cu(OH)
2
nanorods were
formed (Fig. 11). The diameter of the whole nanorod was found
to increase with pseudomorphic replication. In brief, local dissolu-
tion of Cu(OH)
2
nanorod arrays provided Cu
2+
ions at the Cu(OH)
2
/-
solution interfaces. The Cu
2+
ions coordinated with the ligands thus
forming MOFs. The MOF nanocrystals grew on the surface of the
nanorod arrays. Therefore, Cu(OH)
2
nanorod arrays served as a
source of Cu
2+
ions and as a template to support MOF growth.
Fig. 10. Schematic illustration of the preparation of GNR@[Al(OH)(1,4-NDC)]
n
core–
shell composites capable of light-controlled molecular release. Adapted with
permission from ref. [138]. Copyright 2013 American Chemical Society.
J. Fonseca and T. Gong Coordination Chemistry Reviews 462 (2022) 214520
11
The organic ligands etched the templates and coordinated with
Cu
2+
to form MOFs. In particular, the hierarchical MOF-2 nanorod
array exhibited excellent catalytic activity (90% conversion in
6 min) and cycling stability for the reduction of 4-nitrophenol by
NaBH
4
[141].
Yolk-shell NP@MOF heterostructures have been synthesized by
coordination replication of NP@M
x
O
y
core–shell nanostructures
[143]. Specifically, an Au@Cu
2
O core–shell nanostructure reacted
with H
3
BTC in mildly acidic solution at 80 °C overnight, producing
a yolk-shell Au@HKUST-1 petalous heterostructure (Fig. 12a-c).
Cu
2
O shell acted as a template to direct MOF growth, provided
the metal ions for synthesis, and protected the NPs against acid-
induced corrosion or agglomeration during synthesis. Cu
2
O under-
went an oxidative dissolution (2Cu
2
OþO
2
þ8H
þ
!4Cu
2þ
þ4H
2
O)
[156] in the mildly acidic medium. The nucleation of the HKUST-1
crystals started on the surface of Cu
2
O. The crystals grew to cover
the entire surface and expanded outward. The coordination repli-
cation depended on the polarity of the solvent used. High-
polarity solvents such as water, EtOH, and DMF promoted the dis-
sociation of H
3
BTC, increasing the acidity of the reaction mixture.
Consequently, the dissolution rate of Cu
2
O exceeded the nucleation
rate of HKUST-1. A less polar solvent, BnOH, was chosen to control
the dissolution rate of Cu
2
O, while still exceeding the coordination
kinetics of the ligands. Once a continuous layer of HKUST-1 was
wrapped around the Cu
2
O shell, the Cu
2+
ions, which moved faster
than the ligands, diffused through the MOFs. Thus, the subsequent
coordination occurred only on the outer surface of the MOF, pre-
serving the hollow core. The HKUST-1 nanocrystals aggregated
together into flowerlike shape with a single 13 nm Au NP residing
at the center. There was no aggregation, dissolution, or morpholog-
ical variation of the Au NPs during the coordination replication.
Cu
2
O was found to be completely transformed into HKUST-1. The
BET surface area of yolk-shell Au(13 nm)@HKUST-1 petalous
heterostructure was 916 m
2
g
1
. This surface area was lower than
that of HKUST-1 (1482 m
2
g
1
) due to the heavier and nonporous
metal cores of yolk-shell Au(13 nm)@HKUST-1 petalous
heterostructure. The shell thickness of the petalous heterostruc-
ture was found to depend on the Cu salt concentration during syn-
thesis of the Au@Cu
2
O core–shell nanostructure. In addition, the
type and number of NPs encapsulated in each petalous
heterostructure depended on the type and concentration of NPs
used during synthesis of the Au@Cu
2
O core–shell nanostructure
(Fig. 12d-g). Pt-on-Au dendritic NPs were similarly encapsulated
in HKUST-1 petalous structure. This PtAuDNPs@HKUST-1 petalous
heterostructure was tested as a catalyst for the liquid-phase hydro-
genation of olefin [143].
Fig. 12. (a and b) TEM images of yolk-shell Au(13 nm)@HKUST-1 petalous heterostructures. (c) STEM dark-field image of yolk-shell Au(13 nm)@HKUST-1 petalous
heterostructures. TEM images of (d) yolk-shell Au nanorod@HKUST-1 petalous heterostructure, (e) yolk-shell Au(100 nm)@HKUST-1 petalous heterostructures, (f) yolk-shell
Pd nanocube(20 nm)@HKUST-1 petalous heterostructure, and (g) yolk-shell multiple Pd nanocubes@HKUST-1 petalous heterostructure. Adapted with permission from ref.
[143]. Copyright 2014 American Chemical Society.
Fig. 11. SEM images of the hierarchical MOF-2 nanorod arrays at different reaction
times: (a) 0 h, (b) 3 h, (c) 5 h. (d) SEM and (e) TEM images of a hierarchical MOF-2
nanorod at 5 h. Adapted with permission from ref. [141]. Copyright 2017 John Wiley
and Sons.
J. Fonseca and T. Gong Coordination Chemistry Reviews 462 (2022) 214520
12
Although it is beyond the scope of this review, it is worth men-
tioning that flexible fibrous materials have been coated with MOFs
by coordination replication [144]. As a model to demonstrate the
coordination replication coating process, cotton fiber was coated
with HKUST-1. First, copper hydroxide nitrate (CHN) was prede-
posited on the cotton fiber. The CHN-coated fiber was then
immersed in an EtOH/water (9:1 v/v) solution of H
3
BTC for 1 h. In
doing so, coordination replication occurred at the interface between
CHN and the ligand solution, producing HKUST-1 crystals on the
fiber surface (Fig. 13a). CHN acted as a precursor of Cu
2+
ions and
as a template for the growth of HKUST-1. MOF crystals with diame-
ters below 200 nm were formed after only 1 min of treatment. The
fiber surface was almost covered after 5 min of reaction. The
HKUTS-1 crystals on the fiber surface grew over time. Crystal growth
was facilitated by the diffusion of reagents through the intercrys-
talline voids and/or pores of HKUST-1 [157]. The HKUST-1 crystals
reached a maximum diameter of approximately 2
l
m after 30 min
of reaction. The CHN precursor was completely consumed at this
time of reaction. The reaction medium was a water/EtOH (9:1 v/v)
solution. Only small HKUST-1 crystals formed on the fiber surface
when EtOH was the reaction medium. Therefore, a small amount
of water promoted the release of cupric ions from the precursor for
HKUST-1 growth [145,158,159]. On the other hand, a scattered coat-
ing was achieved when the volume percent of water in the solvent
mixture was above 30% (Fig. 13b). It was attributed to a rapid disso-
lution of the precursor (CHN). Other fibrous materials, such as elec-
trospun polyacrylonitrile (PAN) nanofiber (500 nm) mat and a
commercial polyester microfiber (5
l
m) textile, were similarly
coated with HKUST-1. Furthermore, other MOFs, such as CuBDC,
ZIF-8 and ZIF-67, also similarly coated a cotton fiber by altering the
organic ligands and/or the predeposited insoluble layered hydroxide
salt [160,161]. It is worth mentioning that the HKUST-1-coated cot-
ton fiber was used for the continuous liquid-phase removal of the
organosulfur compound dibenzothiophene (DBT) from simulated
gasoline [144].
3.3. Pseudomorphic replication of Zn-based monolithic solids
ZnO@ZIF-8 nanorods, ZnO@ZIF-8 nanorod arrays and ZnO@ZIF-
8 nanotube arrays with core–shell heterostructures have been
developed by pseudomorphic replication [145]. ZnO nanorods
and MeIM were added to a solution of DMF/H
2
O (3:1 v/v). The mix-
ture was sonicated for 5 min and then heated at 70 °C for 24 h, thus
forming ZnO@ZIF-8 core–shell nanorods. ZnO nanorods became
rough and some ZIF-8 NPs appeared after 4 h of reaction
(Fig. 14a). ZIF-8 NPs grew and covered the surface of the ZnO
nanorods after 8 h of reaction (Fig. 14b). The transformation of
ZnO nanorods into ZnO@ZIF-8 nanorods was explored through
the thickness ratio, T
ZIF 8
/D
ZnO
(T
ZIF 8
= thickness of ZIF-8 shells;
D
ZnO
= diameter of ZnO cores) (Fig. 14e). T
ZIF 8
/D
ZnO
increased as
the reaction progressed. Therefore, ZIF-8 shells became thicker
and ZnO nanorod cores thinner over time (Fig. 14). The MeIM
ligands passed through the pores of ZIF-8, reached the surface of
the ZnO nanorod, and coordinated with the dissolved Zn
2+
ions
to form ZIF-8. Zn
2+
ions also diffused from the ZnO nanorod
through the pores of ZIF-8 and reacted with MeIM on the outer sur-
face of ZIF-8 layer. The growth of the ZIF-8 shells was found to stop
within 48 h of reaction due to diffusion limitation. The growth of
ZIF-8 depended on the release rate of Zn
2+
ions and the coordina-
tion rate with MeIM, which, in turn, depended on the composition
of the solvent and the reaction temperature [48,145].At70°C, ZIF-
8 was not formed when DMF was used as the solvent. In DMF, the
etching ability of MeIM could not release a sufficient concentration
of Zn
2+
ions for the formation of ZIF-8. ZIF-8 was not formed when
H
2
O was used as the solvent. In H
2
O, the dissolution rate of ZnO
nanorods was faster than the coordination rate. ZnO@ZIF-8 core–
shell nanorods were obtained when the solvent was H
2
O and
DMF (i.e., DMF/H
2
O = 3:1, 2:1, 1:1). This solution provided the
proper balance between dissolution rate and coordination rate.
Some independent ZIF-8 particles appeared together with
ZnO@ZIF-8 core–shell nanorods when the H
2
O content increased
further (i.e., DMF/H
2
O = 1:2 or 1:3). Independent ZIF-8 particles
and ZnO@ZIF-8 nanorods coexisted when the solvent mixture
was DMF/H
2
O = 1:1 and the reaction temperature was above
70 °C. ZnO@ZIF-8 nanorod arrays and ZnO@ZIF-8 nanotube arrays
were prepared similarly to ZnO@ZIF-8 nanorods but using ZnO
nanorod arrays and ZnO nanotube arrays as sacrificial templates,
respectively (Fig. 14f). ZnO acted as a template and as a source of
Zn
2+
ions independently of the prefabricated ZnO architecture.
Interestingly, the ZnO@ZIF-8 core–shell nanorod arrays showed
Fig. 13. (a) Schematic illustration of the coating of a flexible fibrous material with a MOF by coordination replication. (b) Schematic illustration of the coordination replication
mechanism. Adapted with permission from ref. [144]. Copyright 2019 American Chemical Society.
J. Fonseca and T. Gong Coordination Chemistry Reviews 462 (2022) 214520
13
distinct photoelectrochemical (PEC) response toward different hole
scavengers (e.g., H
2
O
2
and ascorbic acid) in the electrolyte solution
owing to the limitation of the aperture of ZIF-8 [145].
Pt NPs have been encapsulated in a ZIF-8/SiO
2
composite
(Pt@ZIF-8/SiO
2
composite) with a macroporous architecture by
coordination replication of Pt/ZnO/SiO
2
composite [48]. First,
Pt/ZnO/SiO
2
composite was prepared by a nanocasting method.
The macropore diameter of Pt/ZnO/SiO
2
composite was approxi-
mately 800 nm (Fig. 15a,b). Pt NPs were found to be dispersed
within the framework (Fig. 15c,d). It is worth noting that SiO
2
was the skeleton that preserved the architecture from collapsing.
Pt@ZIF-8/SiO
2
composite was synthesized via coordination replica-
tion by immersing Pt/ZnO/SiO
2
composite in a mixed solvent of
DMF and H
2
O (3:1 v/v) at 70 °C for 24 h. The solvent contained
MeIM ligands. These ligands reacted with ZnO at the solid/liquid
interface, thus forming ZIF-8 crystals attached to the host scaffold.
The ZnO in Pt/ZnO/SiO
2
acted as a template and as a precursor
releasing Zn
2+
ions. After the reaction, the Pt NPs were found to
be encapsulated and dispersed within the ZIF-8/SiO
2
matrix
(Fig. 15g,h). Pt@ZIF-8/SiO
2
composite preserved the architecture
and macroporosity of Pt/ZnO/SiO
2
composite after the transforma-
tion (Fig. 15e,f). BET surface area and total pore volume increased
from 97.2 m
2
g
1
and 0.14 cm
3
g
1
for Pt/ZnO/SiO
2
composite to
912.6 m
2
g
1
and 0.475 cm
3
g
1
for Pt@ZIF-8/SiO
2
composite,
respectively. Therefore, ZIF-8 shell not only protected Pt NPs from
agglomeration but also provided high porosity. Pt@ZIF-8/SiO
2
com-
posite exhibited excellent size-selective catalytic properties for the
hydrogenation of n-hexene and cis-cyclooctene due to the molecu-
lar sieve effect of the ZIF-8 shell [48].
In a recent study, cobalt-nitrogen-carbon (Co-NC) hollow
spheres have been prepared by applying pseudomorphic replica-
tion [49]. First, Co-doped ZnO hollow spheres were synthesized
by spray pyrolysis. Subsequently, Co-ZnO@ZIF-8 hollow spheres
were formed by pseudomorphic replication of Co-doped ZnO hol-
low spheres. The coordination replication procedure was similar
to the others mentioned above. It is worth mentioning that ZnO
was not completely converted to ZIF-8 during pseudomorphic
replication. Finally, Co-ZnO@ZIF-8 hollow spheres were calcined
at 800 °C under Ar atmosphere, producing Co-NC hollow spheres.
The remaining ZnO sublimated during the heat treatment. The
Co-NC hollow spheres were applied to the oxygen reduction reac-
tion (ORR) in alkaline media. They exhibited excellent half-wave
potential of 0.904 V and outstanding performance in single-cell
tests (power density of 271 mW cm
2
) without the requirement
of any leaching process for activation [49].
Although it is beyond the scope of this review, it is worth men-
tioning that coordination replication has been applied to synthe-
size ZIF-8 thin films [146,147]. For example, Stassen et al. have
demonstrated the reaction of ZnO films and melted MeIM at
160 °C, producing ZIF-8 thin films.[146] MeIM completely wetted
the ZnO surface, causing a homogeneous reaction. In less than
1 min, a layer of ZIF-8 crystals was formed covering the ZnO layer.
The morphology of the synthesized ZIF-8 film was found to repli-
cate that of the ZnO precursor film. The ZnO layer was only par-
tially transformed. It acted as a bridging layer between the
silicon support and the synthesized ZIF-8 crystals [146].
3.4. Pseudomorphic replication of V-based monolithic solids
Kitagawa, Furukawa, and coworkers have extended the coordi-
nation replication process to vanadium chemistry, synthesizing the
first macroscopic architecture made of V-based MOF crystals [148].
It is worth mentioning that a variety of vanadium oxide morpholo-
gies are accessible [162–165], making vanadium oxides particu-
Fig. 14. Low-magnification TEM images of ZnO@ZIF-8 core–shell nanorods obtained after reaction for (a) 4, (b) 8, (c) 24, and (d) 48 h, respectively. (e) Thickness ratio
(T
ZIF 8
/D
ZnO
;T
ZIF 8
= thickness of ZIF-8 shells and D
ZnO
= diameter of ZnO cores) of ZnO@ZIF-8 nanorods as a function of reaction time. (f) Schematic illustration of ZnO@ZIF-8
nanorod arrays synthesized by coordination replication. The prefabricated ZnO nanorod arrays acted as sacrificial templates in the mixed solvent of DMF and H
2
O, while
MeIM acted as ligand and etching reagent. Adapted with permission from ref. [145]. Copyright 2013 American Chemical Society.
J. Fonseca and T. Gong Coordination Chemistry Reviews 462 (2022) 214520
14
larly suitable candidates as sacrificial templates. Furthermore, they
exhibit promising electronic properties for heterogeneous catalysis
and sensing applications. In the study of Kitagawa et al., a V
2
O
5
-
nH
2
O pattern coated with polyethylene oxide (PEO) was added
to an aqueous solution of 1,4-NDC and ascorbic acid [23]. This mix-
ture was microwaved at 180 °C for 1 s. In doing so, the V
2
O
5
nH
2
O
pattern became a [V(OH)1,4-NDC]
n
architecture with identical
morphology. The V
2
O
5
nH
2
O parent phase was totally consumed
during the coordination replication process. Longer reaction times
resulted in the loss of macroscopic morphology. The PEO layer was
found to be critical to preserving the morphology of the parent V
2
-
O
5
nH
2
O phase during the dissolution process. It was suggested
that PEO acted as a trap, retarding dissolution of V
2
O
5
and diffusion
of the ionic vanadium species at the solid–liquid interface. There-
fore, PEO helped to keep the dissolution kinetics of V
2
O
5
lower than
the crystallization kinetic of MOF, thus ensuring that MOF nucle-
ation occurred at the vicinity of the dissolution front. Ascorbic acid
was critical for the successful conversion of V
2
O
5
to [V(OH)1,4-
NDC]
n
. This reducing agent promoted the dissolution of the sacrifi-
cial V
2
O
5
phase [166] and generated the V
3+
species required for
the construction of [V(OH)1,4-NDC]
n
[148].
3.5. Pseudomorphic replication of Ca-based monolithic solids
Kitagawa, Furukawa, et al. have also prepared Ca-based MOF
architectures by coordination replication using biomineralized
forms of calcium carbonate as templates [149]. A sample of
biomineralized calcium carbonate derived from Baculogypsina
sphaerulata (Fig. 16a) was immersed in an aqueous solution con-
taining an excess of squaric acid (H
2
sq). This mixture was treated
in a microwave reactor at 120 °C for 12 h. The biomineralized sam-
ple was partially converted to a highly crystalline Ca(sq)(H
2
O)
phase with retention of the original architecture. A rougher macro-
pore morphology was found in Ca(sq)(H
2
O) (Fig. 16c) compared
with that of biomineralized calcium carbonate (Fig. 16b). Longer
reaction times did not lead to complete conversion of the biomin-
eralized sample but favored the formation of large crystals of Ca
(sq)(H
2
O) up to 500
l
m in length. In addition, a new flexible
MOF, Ca(dhbq)(H
2
O)
2
, based on the 2,5-dihydroxybenzoquinone
(H
2
dhbq) linker was also developed in this study [149].
4. Sol-gel method
The term ‘sol-gel’ is associated with the exploitation of the solu-
tion state in the synthesis of a solid material. The sol–gel process is
a wet chemical approach to synthesize inorganic and organic–inor-
ganic materials from liquid sources (Fig. 17). Complete homogene-
ity, which is ensured by the solution state, favors the synthesis of
pure products and lower synthesis temperatures [167]. In the
sol–gel method, the precursors are dispersed in a solvent to pro-
duce a suspension of colloidal submicrometer-sized particles
(1
l
m > colloids greater than 10 nm), which is called sol. Gelation
is the process in which colloids link together, forming branched
fractal chains, thus filling the entire volume homogenously with
a coherent network of particles [168]. Therefore, the gel is a non-
fluid 3D network that extends through a fluid phase [167].It
adopts the macroscopic shape of the reaction vessel. Gel aging
may increase stiffness and lead to a rigid structure [169]. Gels
are referred to as xerogels when the solvent phase within the
resulting gel is removed by evaporative drying [170]. They are
called alcogels when the solvent phase to be removed by evapora-
tive drying is alcohol-based. It should be mentioned that simple
evaporation of liquid under ambient conditions is usually accom-
panied by shrinkage. Liquid-vapor interfaces within the gel net-
work produce strong surface tensions, which can destroy the
network and cause the structure to collapse [168]. Aerogels are
produced by removing the solvent phase as a gas phase by super-
critical drying. This drying can avoid capillary forces present dur-
ing ambient drying [171,172]. Therefore, aerogels are typically
materials in which the initial gel network is largely maintained
while the liquid in the pores is replaced by air [173]. The generic
term gel is often applied to xerogels or alcogels, while aerogels
are generally referred to as such.
Fig. 15. (a) SEM image of Pt/ZnO/SiO
2
composite. (b, c) TEM images of Pt/ZnO/SiO
2
composite. (d) STEM image of Pt/ZnO/SiO
2
composite. (e) SEM image of Pt@ZIF-8/SiO
2
composite. (f, g) TEM images of Pt@ZIF-8/SiO
2
composite. (h) STEM image of Pt@ZIF-8/SiO
2
composite. Adapted with permission from ref. [48]. Copyright 2018 John Wiley and
Sons.
J. Fonseca and T. Gong Coordination Chemistry Reviews 462 (2022) 214520
15
In the context of MOFs, sol–gel processing provides an alterna-
tive route to new materials with improved properties for specific
applications [170]. A MOF gel is a non-fluid colloidal network, con-
sisting solely of discrete MOF NPs linked by weak non-covalent
interactions (Van der Walls forces, H-bonding or
p
-
p
stacking),
that expands throughout a liquid phase [31]. The formation of
MOF gels is based on the supramolecular aggregation of discrete
MOF NPs preventing their continuous growth and, therefore,
avoiding the formation of microcrystalline MOF powder
[174,175]. Logically, the nucleation of discrete nanocrystalline
MOFs is essential before gelation. The challenge is to find the con-
ditions in which initial nucleation and subsequent gelation domi-
nate against the more thermodynamically favorable crystal
growth and precipitation processes [31]. In other words, synthetic
conditions such as temperature, metal source, ratio and concentra-
tion of precursors, or additives that disturb coordination [176–
178], must be controlled to favor the preparation of nanocrys-
talline MOFs. The choice of solvent is also an important parameter
for gelation to occur [179].
The sol–gel method is a promising route for the formation of
pure monolithic MOFs [180–182]. The solvent is removed from a
MOF gel provoking MOF NPs aggregation and, thus, the forma-
tion of a monolithic MOF. The formation of MOF NPs is an
important requirement for the formation of MOF gels and there-
fore also for the formation of monolithic MOFs [31]. Metal-
organic xerogels (MOXs) are obtained by solvent evaporation
of the pore liquid from the MOF gels with no or minor changes
in structure. Metal-organic aerogels (MOAs) are produced by
gas-drying the MOF gels [183]. MOX are much denser and exhi-
bit lower porosity than MOA [184]. Therefore, it is possible to
prepare hierarchically porous MOFs by adjusting the drying step.
It should be noted that both MOXs and MOAs often show high
fragility [185–187].
Monolithic MOFs prepared by sol–gel method have advantages
over other macroscopic MOF architectures:
(1) Hierarchical porosity. Hierarchically porous monolithic
MOFs can be prepared by controlling the drying step. It is
worth mentioning that the macroporous structure of MOF
monoliths is difficult to control [31].
(2) Control over monolithic MOF morphology. MOF gels, which
often exhibit viscoelastic behavior, allow the preparation of
monolithic MOFs with predefined shapes [181,188]. Further-
more, the morphological structures of MOF gels can be mod-
ified by varying the synthesis conditions. For example, the
morphological structure of HKUST-1 changes from granular
to fibrous by modifying the molar ratio between metal and
ligand [189].
(3) No pore blocking. The sol–gel synthesis approach does not
require the use of binders that normally block the pores of
MOFs.
Fig. 17. Schematic illustration of the sol–gel process.
Fig. 16. (a) Field-emission SEM image of a biomineralized calcium carbonate
sample derived from Baculogypsina sphaerulata (‘‘star sand”) from Hoshizuna-no-
hama (‘‘Star Sand Beach”), Iriomote Island, Okinawa, Japan. It has a star-shaped
shell approximately 1.5 mm in size. The inset image shows an enlarged view of the
macropores. The scale bar within the inset is 10
l
m. Field-emission SEM images of
the biomineralized calcium carbonate sample (b) before and (c) after microwave
treatment in an aqueous solution containing H
2
sq at 120 °C for 12 h. Adapted with
permission from ref. [149]. Copyright 2016 American Chemical Society.
J. Fonseca and T. Gong Coordination Chemistry Reviews 462 (2022) 214520
16
(4) No structural collapse [185]. Unlike many mechanical densi-
fication processes, the sol–gel synthesis approach does not
cause structural collapse.
However, the formation mechanism of MOF gels, and their
resultant monolithic MOFs, is still poorly understood. The principal
aim in this area is to generalize the synthesis of monolithic MOFs.
In this Section, we extensively explore the formation of mono-
lithic MOFs by sol–gel approach (Table S4). State-of-the-art work
is reviewed focusing on the mechanistic aspects of sol–gel process-
ing. The application of these MOF architectures is specified as was
done in previous Sections. In doing so, the potentials of MOF mono-
liths developed by sol–gel is revealed.
4.1. Sol-gel processing of Fe-based monolithic gels
In an early study, Horcajada et al. have prepared a monolithic
MIL-89 (Fe
3
O(CH
3
OH)
3
(tt-MA)
3
, tt-MA = trans,trans-muconic acid)
xerogel [190]. Iron(III) acetate (FeOAc) and tt-MA were dissolved
in EtOH and subsequently heated at 60 °C for 10 min. In doing
so, a homogeneous colloidal solution was formed. This sol con-
sisted of nanocrystalline MIL-89 with sizes varying from 20 to
40 nm. The size of nanocrystalline MIL-89 in the sol increased to
300–500 nm after two days at room temperature. A perfect mono-
lithic MIL-89 xerogel appeared after three months at room temper-
ature. In contrast, other iron precursors, such as iron(III) nitrate
and iron(III) chloride, produced polydisperse microcrystalline
MIL-89 powder at 60 °C for 10 min. FeOAc was suggested to control
nucleation and growth of MIL-89 by reducing the functionality of
the organic ligand. It is worth noting that FeOAc itself carries the
inhibiting ligand (Fe
3
O(CH
3
CO
2
)
6
(H
2
O)
3
CH
3
CO
2
) and therefore
can act as a chemical-growth-inhibiting agent. Remarkably,
homogeneous thin films of flexible porous MIL-89 were prepared
from MIL-89 sol (60 °C for 10 min) by dip-coating [190].
Zou’s research group has developed monolithic monometallic
and bimetallic Fe-based MOF gels [174]. These gels were synthe-
sized by mixing ethanolic solutions of metal source(s) and H
3
BTC
at room temperature and/or 120 °C for 24 h. The metal source(s)
in EtOH were Fe(NO
3
)
3
9H
2
O and/or Al(NO
3
)
3
9H
2
O. While the
Fe
3+
ions coordinated with BTC
3-
at room temperature, the Al
3+
ions coordinated at 120 °C. Temperature was shown to be a critical
parameter to control MOF gel synthesis [177,191]. At room tem-
perature, Fe
3+
ions coordinated with BTC
3-
to form nanocrystalline
MOFs until all Fe
3+
ions were consumed (Fig. 18). The gels were
formed due to growth perturbation by fast reduction of Fe
3+
ions
[176,192]. Then, Al
3+
ions coordinated with the excess BTC
3-
at
120 °C, producing a heterogeneous gelation. Al
3+
ions were intro-
duced to disturb the growth of the Fe-based monoliths, thus
obtaining gels with heterogeneous nanocrystalline MOFs. The dif-
ference in ion sizes (Fe
3+
= 0.067 nm, Al
3+
= 0.055 nm) also induced
heterogeneity, enhanced mismatched growth, and therefore
favored gelation. Moreover, the same BTC
3-
molecule was able to
react with both Fe
3+
and Al
3+
. This could also have triggered the
gelation process. The nanocrystalline MOFs, which acted as build-
ing blocks in these MOF gels, were reported to be MIL-100 (Fe/Al).
In other words, these MOF gels were monolithic architectures
formed by polymerization of MIL-100(Fe) and/or MIL-100(Al) NPs
[193,194]. As the heterogeneity of the gels increased, the size of
the nanocrystalline MOFs, which conformed such gels, increased.
The MOF gels were subjected to air drying and supercritical drying
which led to MOXs and MOAs, respectively. Various monometallic
and bimetallic MOAs and MOXs were synthesized by varying the
concentration of metals: Fe-BTC (MOA-1 and MOX-1), 0.75Fe-
0.25Al-BTC (MOA-2 and MOX-2), 0.5Fe-0.5Al-BTC (MOA-3 and
MOX-3), 0.25Fe-0.75Al-BTC (MOA-4 and MOX-4) and Al-BTC
(MOA-5 and MOX-5), respectively. MOXs and MOAs showed
micro-, meso- and macroporosity. Their pore volume increased as
their heterogeneity increased. MOXs had lower porosity than their
Fig. 18. Schematic illustration of the synthesis of a monolithic bimetallic MOX. Notation: MOFP = MOF particle. MOFP is referred to as nanocrystalline MOF or MOF
nanoparticle throughout the Review. Adapted with permission from ref. [174]. Copyright 2015 Springer Nature.
J. Fonseca and T. Gong Coordination Chemistry Reviews 462 (2022) 214520
17
analogous MOAs, suggesting a partial shrinkage of the porous net-
work upon air drying. Notably, MOA-3 exhibited the highest BET
surface area (1861 m
2
g
1
) among all these MOXs and MOAs. These
gels showed high adsorption of dye molecules (290 mg g
1
rho-
damine B (RHB) and 265 mg g
1
methyl orange) under acidic con-
ditions. They also presented good stability in this medium where
most MOFs/MOF gels tend to decompose [174]. In another study,
this research group has prepared nanoporous carbon (NPC) with
and without Fe
3
O
4
/Fe NPs from MOX-1 and MOX-5, respectively
[193]. MOX-1 was used as a sacrificial material to prepare NPC
with homogenously dispersed Fe
3
O
4
/Fe NPs (Fe
3
O
4
/Fe/C). Fe
3
O
4
/
Fe/C exhibited a high specific capacitance of 600F g
1
at a current
density of 1 A g
1
and excellent capacity retention for 5000 cycles.
Similarly, MOX-5 was carbonized and treated with NaOH to pre-
pare NPC. An aqueous asymmetric supercapacitor (aASC) was
assembled using Fe
3
O
4
/Fe/C as positive electrode and NPC as neg-
ative electrode (Fe
3
O
4
/Fe/C//NPC) (Fig. 19). This construction
showed a high energy density of 17.496 Wh Kg
1
at the power
density of 388.8 W Kg
1
[193].
4.2. Sol-gel processing of Al-based monolithic gels
Ultralight monolithic Al-based MOAs with hierarchical porosity
have been developed by gelation of MOF NPs [191]. Al-based MOF
gels were prepared by mixing ethanolic (or DMF-EtOH) solutions
of Al(NO
3
)
3
9H
2
O and carboxylic acid ligand (H
2
BDC, H
2
FDC (2,5-
furandicarboxylic acid), 1,4-NDC, H
3
BTC, H
3
BTB, H
2
ADC (9,10-
anthracenedicarboxylic acid) or H
2
BDC-NH
2
) and subsequent heat-
ing. MOF nucleation occurred in the early stage of this reaction,
where discrete but periodic nanocrystalline MOFs formed as pre-
cursors [195,196]. In the second stage of this reaction, the coordi-
nation equilibria [197] were disturbed by other competing
interactions. Thus, non-crystallographic branching of the
nanocrystalline MOFs occurred, which favored gelation
[176,198,199]. The gel matrix consisted of nanocrystalline MOF
subunits (Fig. 20). The nanocrystalline MOFs in the MOF gels were
irregularly interconnected creating voids and sustaining the
matrix. Proper temperature was critical in triggering gelation and
not crystallization and/or precipitation. Other factors, such as
metal-to-ligand ratio and concentrations, only influenced the gela-
tion time and porosity of the resulting MOF gels. After a certain
aging time, the pore liquid of the Al-based MOF gels was exchange
with liquid CO
2
under supercritical conditions to produce porous
Al-based MOAs (Fig. 20). Direct drying of MOF gels led to MOXs,
which had less porosity than MOAs. Al-based MOAs were porous
monolithic wormhole-like networks made of cross-linked
nanocrystalline MOFs. Their structures were not conclusively
determined and could also have contained mixed or unknown
phases [200]. They were found to exhibit micro- and mesoporosity.
The textural properties of MOAs were strongly influenced by the
reactant concentrations [191].
The sol–gel method in conjunction with templates, such as sur-
factant micelles, has been widely used to prepare mesoporous sil-
ica oxide structures, such as MCM-41 and SBA [201,202]. Similarly,
this approach can produce MOFs [203]. MOF gels have also been
prepared in the presence of cetyltrimethylammonium
bromide/1,3,5-trimethylbenzene (CTAB/TMB) micelles [191]. The
Al-based MOF gels mentioned above were synthesized as previ-
ously described but in the presence of CTAB and TMB. While CTAB
was used as a structure directing agent, TMB was an auxiliary
agent to swell the CTAB micelles [202]. The corresponding MOAs,
which were also obtained by drying under supercritical conditions,
exhibited more homogeneous wormhole-like mesostructures com-
pared to non-templated MOAs. Therefore, the CTAB/TMB micelles
modulated the agglomeration and gelation of the nanocrystalline
MOFs. Furthermore, the BET surface areas of these template-
prepared MOAs increased significantly [191].
Zou’s research group has also prepared a monolithic Al-based
aerogel [204]. First, an ethanolic solution of Al(NO
3
)
3
9H
2
O and
H
3
BTC was heated at 120 °C for 1 h, forming a monolithic gel.
The gel and MIL-100(Al) were found to have a closely related struc-
ture (Fig. 21a). However, the gel showed low crystallinity. It was
suggested that the gel resulted from interconnected MIL-100(Al)
NPs. A xerogel was prepared by drying the wet gel in air at low
temperature for a few hours. The wet gel shrank and spread into
small xerogel particles. The monolithic aerogel was formed from
the wet gel by supercritical CO
2
drying process. The wet gel struc-
ture was largely maintained. The density of the xerogel was higher
than that of the aerogel. In contrast, the aerogel exhibited higher
macroporosity. The BET surface area of the xerogel and the aerogel
was 1761 and 1795 m
2
g
1
, respectively. The aerogel showed a
broader pore size distribution than the xerogel. The xerogel and
the aerogel were tested for the removal of microcystins (MC),
specifically MC-LR, which is the most toxic member of the MC fam-
ily. The xerogel and the aerogel removed over 96.3 wt% of the MC-
LR in water, and their adsorption capacities were as high as 6861
and 9007
l
gg
1
at an initial MC-LR concentration of 10000 ppb,
respectively. They exhibited an adsorption of MC-LR superior to
their analogous porous silica [204]. In a subsequent study, porous
carbons were prepared from this Al-based xerogel, and this Al-
Fig. 19. Schematic illustration of the preparation of an aASC made of Fe
3
O
4
/Fe/C as positive electrode and NPC as negative electrode. Adapted with permission from ref. [193].
Copyright 2016 American Chemical Society.
J. Fonseca and T. Gong Coordination Chemistry Reviews 462 (2022) 214520
18