Crystal Engineering of Hybrid Coordination Networks: From Form to Function

Abstract

Crystal engineering, the field of chemistry that studies the design, properties, and applications of crystals, is exemplified by the emergence of coordination networks (CNs). CNs are comprised of metal cations or metal clusters linked into 2D or 3D networks by organic and/or inorganic ligands. Early studies on CNs tended to focus upon design using self-assembly and/or reticular chemistry, but
interest in CNs has grown exponentially as a consequence of their potential utility in catalysis, purification, sensing, and gas storage. Whereas the use of organic ‘linker ligands’ has afforded N75 000 metal-organic frameworks (MOFs), hybrid CNs (HCNs) sustained by both inorganic and organic linkers remain understudied. We herein review design approaches to HCNs and recent developments
concerning their exceptional gas-separation performance.

Full text

Review
Crystal Engineering of Hybrid Coordination
Networks: From Form to Function
Soumya Mukherjee
1
and Michael J. Zaworotko
1,
*
Crystal engineering, the field of chemistry that studies the design, properties, and
applications of crystals, is exemplified by the emergence of coordination net-
works (CNs). CNs are comprised of metal cations or metal clusters linked into
2D or 3D networks by organic and/or inorganic ligands. Early studies on CNs
tended to focus upon design using self-assembly and/or reticular chemistry, but
interest in CNs has grown exponentially as a consequence of their potential utility
in catalysis, purification, sensing, and gas storage. Whereas the use of organic
‘linker ligands’has afforded N75 000 metal-organic frameworks (MOFs), hybrid
CNs (HCNs) sustained by both inorganic and organic linkers remain understudied.
We herein review design approaches to HCNs and recent developments
concerning their exceptional gas-separation performance.
Crystal Engineering
‘What would the properties of materials be if we could really arrange the atoms the way we want
them’[1].
Feynman answered his question as follows: ‘I can hardly doubt that when we have some control
of the arrangement of things on a small scale, we will get an enormously greater range of possible
properties that substances can have and of different things that we can do’.Inessence,
Feynman’s vision provides the motivation for crystal engineering, the field of chemistry that stud-
ies the design, properties, and applications of crystals [2]. The term crystal engineering was
coined by Pepinsky in 1955 [3] and first implemented in the 1960s, when G.M.J. Schmidt intro-
duced the topochemical principle in the context of solid-state photodimerization reactions [4].
Schmidt also asserted that there are ‘four phases of crystal engineering’:
1. Experimental determination of crystal structure and correlation with properties: phase of the
topochemical principle.
2. Establishing the limits of the topochemical principle.
3. Development of empirical rules.
4. Systematic solid-state (photo)chemistry.
The late 1980s saw significant growth in the scope of crystal engineering thanks to parallel de-
velopments in our understanding of: (i) molecular crystals, and (ii) coordination networks (CNs).
In both cases, progress was enabled by applying the concepts of supramolecular chemistry [5]
and data-mining using the Cambridge Structural Database [6]. With respect to molecular
crystals, contributions by Ermer [7], Desiraju [8], and Etter [9] resulted in better understanding
of hydrogen bonding patterns and introduction of concepts such as graph sets and supramo-
lecular synthons [9,10](i.e.,‘phase 3’crystal engineering). Ermer’sreportonthefivefold
interpenetrated diamondoid, dia, structure of adamantane-1,3,5,7-tetracarboxylic acid [7]is
noteworthy given the subsequent relevance of dia topology and interpenetration to CNs, as
discussed later. Ermer also suggested the use of organic nodes and condensation reactions
Highlights
The modular composition of hybrid co-
ordination networks (HCNs) means
that they are amenable to ‘bottom-up’
design using the principles of crystal
engineering.
Hybrid ultramicroporous materials
(HUMs), a subclass of HCNs with pore
diameter b0.7 nm, can combine strong
electrostatics with a high density of tight
binding sites to create energetic ‘sweet
spots’that enable: (i) benchmark selec-
tivity towards CO
2
,C
2
, and C
3
hydro-
carbons; and (ii) fast sorption kinetics.
Whereas the first generation of HUMs
offered orderof magnitudeimprovement
versus previous benchmark sorbents
such as zeolites and MOFs, second
generation variants have afforded even
stronger performance.
That bespoke HUMs now exist for
several important gases such as CO
2
,
C
2
H
2
, and C
2
H
6
can enable one-step,
low energy purification of multicompo-
nent gas mixtures, even for trace
impurities.
1
Bernal Institute, Department of Chemical
Sciences, University of Limerick, Limerick
V94 T9PX, Ireland
*Correspondence:
xtal@ul.ie (M.J. Zaworotko).
506 Trends in Chemistry, June 2020, Vol. 2, No. 6 https://doi.org/10.1016/j.trechm.2020.02.013
© 2020 Published by Elsevier Inc.
Trends in Chemistry

to form polymeric networks, the approach typically used today to prepare covalent organic
frameworks (see Glossary)[11].
Whereas the end goal of crystal engineers has always been to reach ‘phase 4’of crystal engineering,
which can be more broadly defined as systematic enhancement of functional properties, it has only
been in the past 15–20 years that design principles became reliable enough to enable phases 2 and
3. Today there are two primary exemplifications of the utility of crystal engineering that could be
regarded as having reached phase 4: (i) pharmaceutical cocrystals [12] offer a low cost, low
risk route to better medicines, and several drug products based upon cocrystal drug substances
have reached the market [13,14]; (ii) CNs are a subclass of coordination polymers (CPs).CPsare
defined by IUPAC as ‘coordination compounds with repeating coordination entities extending in 1,
2, or 3 dimensions’[15].CNsaredistinctfrommostexistingclasses of materials since their inherent
modularity makes them amenable to design using Robson’s‘node-and-linker’approach [16]or
Yaghi and O’Keeffe’s reticular chemistry concept [17]. Crystal engineering enables the ‘bottom-
up’approaches promoted by Feynman and has resulted in new families of CNs. Perhaps more
importantly, such approaches represent a paradigm shift from the high-throughput methods tradi-
tionally utilized in materials discovery.
CNs
‘A coordination network is a compound extending, through repeating coordination entities, in
1 dimension, but with cross-links between two or more individual chains, loops, or spiro-
links, or a coordination compound extending through repeating coordination entities in 2 or
3 dimensions’[15].
CNs, trace back to Alfred Werner, who introduced coordinate covalent bonding [18]andits
directionality, thereby enabling the first generation of CNs [19]. However, the prototypal CN is
even older since it is the naturally occurring material Prussian blue (Fe
4
[Fe(CN)
6
]
3
), a pigment
manufactured and used commercially since the early 18th century. The ‘primitive cubic’(pcu)
Glossary
Azolate ultramicroporous material:
a subclass of metal azolate frameworks
(MAFs) with pore sizes b0.7 nm.
Cocrystals: multicomponent crystalline
single-phase solids comprised of two or
more different molecular and/or ionic
compounds, generallyinastoichiometric
ratio, which are neither solvates nor
simple salts.
Coordination polymer (CP): a
coordination compound with repeating
coordination entities extending in one,
two, or three dimensions.
Covalent organic framework: aclass
of crystalline or semicrystalline porous
organic polymers with permanent
porosity and ordered structures.
Dynamic column breakthrough
(DCB) experiments: adsorption and
diffusion experiments that provide
information on the macroscopic
separation performance of the adsorption
column. A typical DCB experiment
comprises ‘saturating the adsorption
column with a gas (or gas mixture) of a
known composition and switching the inlet
to a gas stream that is different from the
one used to saturate the column. The exit
gas phase composition and flow rate is
measured with suitable detectors’[100].
Metal-organic framework (MOF): a
coordination network with organic
ligands containing potential voids.
Molecular building block (MBB): a
discrete coordination complex or
multifunctional organic ligand that is
suited to serve as a node for the
generation of a coordination network.
Pharmaceutical cocrystal: a
multicomponent crystal in which at least
one of the constituents is an active
pharmaceutical ingredient.
S
AC
:adsorption selectivity of C
2
H
2
over
CO
2
.
S
AE
:adsorption selectivity of C
2
H
2
over
ethylene, C
2
H
4
.
S
CA
:adsorption selectivity of CO
2
over
acetylene, C
2
H
2
.
S
CM
:adsorption selectivity of CO
2
over
methane, CH
4
.
S
CN
:adsorption selectivity of CO
2
over N
2
.
S
CW
:adsorption selectivity of CO
2
over
water, H
2
O.
Synergistic sorbent separation
technology (SSST): the use of highly
selective physisorbents in a synergistic
manner. Each gas-specificsorbent
covers trace/bulk gas capture with ultra-
high selectivity for one of the impurities,
thereby enabling one-step purification of
multicomponent gas mixtures.
Trends
Trends
in
in
Chemistry
Chemistry
Figure 1. Design of Coordination Networks, Simple yet Seminal. The ‘node-and-linker’design strategy is exemplified
by the coordination network [Zn(4,4′-bipy)
2
(H
2
O)
2
]
n
[SiF
6
]
n
(Cambridge Structural Database refcode: JEZRUB, extra-
framework SiF
6
2–
anions omitted for clarity; color codes: grey, C; blue, N; red, O; green, F; saffron, metal cation; yellow
sphere, potential void) [22].
Trends in Chemistry
Trends in Chemistry, June 2020, Vol. 2, No. 6 507

crystal structure of Prussian blue was reported in 1936 [20], and in 1949 Powell reported
Hofmann’s clathrate ([Ni(NH
3
)
2
Ni(CN)
4
]·guest) [21]. Both materials are sustained by cyano linker
ligands. Prussian blue and Hofmann’s clathrate are highlighted herein because, while they do
not exhibit the modularity that is a feature of many CN families, they are prototypal for two
of the most commonly reported structural classes of CN, pcu and square lattice (sql) network
topology, respectively. When a CN does have a high degree of modularity then phase 4 crystal
engineering based upon a parent CN is enabled by known coordination chemistry. An important
development in this context came from Hoskins and Robson, who introduced the node-and-
linker approach [22], which led to the design of dia [23], sql [24], and pcu [25] networks. Using
the same approach, Kitagawa’s group reported ‘honeycomb’(hcb) CNs [26]. CN design can
thereby be represented as a type of molecular Lego®, since it uses a metal, metal cluster, or
organic building block with three or more connection points (or nodes) linked by ligands (linkers)
that connect these nodes (Figure 1). Shortly thereafter, Moore and Lee coined the term
molecular building block (MBB) [27], which is conceptually important as it allows for fine-
tuning of structure and composition using known chemistry. The existence today of a library of
N75 000 such materials [28]reflects the diversity of MBBs and linker ligands, yet this diversity is
not even close to being fully tapped, an issue we address later.
The creation of empty space by design resulted in the emergence of two phenomena that were
previously understudied: interpenetration of CNs [29–32] and CNs that exhibit permanent porosity
[i.e., porous coordination networks (PCNs)]. These phenomena can occur together, in which case
porosity is reduced, or separately. The potential utility of permanent porosity provided motivation
for the design, synthesis, and characterization of new classes of PCNs. It was around this time
that Kitagawa coined the term ‘porous coordination polymers’(PCPs) [33], Yaghi introduced
‘metal-organic frameworks’(MOFs) [34], and Zhou used PCNs [35]. A plethora of three-letter
codes now exist.
The first PCNs with extra-large gravimetric surface area were reported in 1999 by the groups of
Williams [36] (HKUST-1, [Cu
3
(1,3,5-benzenetricarboxylate)
2
]
n
, ca. 1900 m
2
g
–1
) and Yaghi [37]
(MOF-5, [Zn
4
O(1,4-benzenedicarboxylate)
3
]
n
, ca. 3800 m
2
g
–1
). The search for extra-large
surface area MOFs continues, as exemplified by DUT-60 (7839 m
2
g
–1
)[38]andNU-110
(7140 m
2
g
–1
)[39]. However, as discussed herein, smaller is typically better with respect to selec-
tive binding of sorbates. This is because the lead PCNs for capture of important small molecules
such as CO
2
[40], C
2
H
2
[41], and H
2
O[42,43] perform thanks to strong sorbent-sorbate binding
that is enabled by: (i) optimal pore size for tight binding, and (ii) the right pore chemistry. It is there-
fore unsurprising that microporous CNs (pore size b2.0 nm), in particular ultramicroporous
variants (pore size b0.7 nm), have emerged as the top-performing PCNs for gas separation
and purification [44]. Further, whereas interpenetration was once considered to the nemesis of
porosity [45], as detailed below, it can also result in tight binding sites and exceptional sorption
performance.
Hybrid Coordination Networks (HCNs)
If linker ligands are classified as being either inorganic or organic, then there are three possible
classifications for PCNs: inorganic only (e.g., Prussian blue or ALPOs [46]); organic only
(e.g., HKUST-1 or MOF-5); and combined inorganic and organic (HCNs). That N75 000 MOFs
sustained by organic linkers exist [28] tends to obscure that, at least in principle, combinations
of inorganic linkers and organic linkers offer as much compositional diversity. Figure 2 illustrates
these classifications through the parents of three pcu topology PCN families: (i) Prussian blue,
Fe
4
[Fe(CN)
6
]
3
.xH
2
O, comprises alternating Fe(III) and Fe(II) cations linked by cyano ligands
(Figure 2,aboveleft)[47] and is the parent of Prussian blue analogs in which Fe cations are
Trends in Chemistry
508 Trends in Chemistry, June 2020, Vol. 2, No. 6

substituted by other metal cations [48]; (ii) the prototypal HCN, [Zn(4,4′-bipyridine)
2
(SiF
6
)],
SIFSIX-1-Zn, was reported in 1995 and exhibits microporous (0.8 × 0.8 nm) square channels
[25](Figure 2A), whereas its ultramicroporous (b0.4 × 0.4 nm) variant SIFSIX-3-Zn (Figure 2,
above center) is the parent hybrid ultramicroporous material (HUM) [38]; and (iii) DMOF-1
(Figure 2, above right) is the parent of a large family of related MOF structures that are accessed
through linker, pillar, or metal substitution. DMOF-1 was reported in 2004 and is comprised of
1,4-benzenedicarboxylate (1,4-bdc) linked ‘paddlewheel’sql nets pillared by 1,4-diazabicyclo
[2.2.2]octane (dabco) linkers [49].
The cationic nature of dipyridyl or pyrazine linked sql topology networks means that appropriate
inorganic dianions can serve as pillars to generate families of neutral HCNs with pcu [25] or mmo
topology [50]. The Cu(II) variant of SIFSIX-1-Zn, SIFSIX-1-Cu, was studied by Kitagawa’s group
[51] and determined to be microporous (BET surface area = 1337 m
2
g
–1
) with high CH
4
satura-
tion uptake at 298 K and 35 bar (145.6 cm
3
g
–1
)[52]. This uptake far exceeds that of zeolite 5A
(82.9 cm
3
g
–1
), the leading CH
4
adsorbent at the time of publication in 2000. SIFSIX-1-Zn is also
modular with respect to the inorganic pillar and the organic linker. Substitution of SIFSIX was
demonstrated in Cu(II) analogs with GeF
6
2–
(GEFSIX) pillars [51,53]. Other anionic pillars used to
date include MFSIX (e.g., SiF
6
2–
,TiF
6
2–
,SnF
6
2–
), FOXY (e.g., NbOF
5
2–
), M'FFIVE (e.g., AlF
5
2–
)[54],
and DICRO (dichromate, Cr
2
O
7
2–
,Figure 2C) [55]. Pore size can be expanded by using
longer dipyridyl linkers, although pore size can be reduced if interpenetration [56]occurs
Trends
Trends
in
in
Chemistry
Chemistry
Figure 2. Composition of Primitive Cubic (pcu) Networks and Prototypal Hybrid Coordination Networks (HCNs). Above: the three classifications of pcu
networks (inorganic only, hybrid, and organic only) are exemplified by Prussian blue, SIFSIX-3-Zn and DMOF-1, respectively (color codes: red, inorganic linker; blue,
organic linker; green, metal nodes). Below: three examples of pcu topology HCNs formed from cationic square lattice (sql) nets and inorganic pillars: (A) SIFSIX-1-Zn
(non-interpenetrated pcu net, fluorometallate pillar); (B) SIFSIX-2-Cu-i (twofold interpenetrated pcu net, fluorometallate pillar, F atoms are disordered); (C) DICRO-3-Ni-i
(interpenetrated pcu net, oxometallate pillar) (color codes: grey, C; blue, N; red, O; green, F; orange, Si; yellow, Cr; saffron, metal cation; cyan, interpenetrated
network). Abbreviations: MOF, Metal-organic framework.
Trends in Chemistry
Trends in Chemistry, June 2020, Vol. 2, No. 6 509

[pcu-interpenetrated topology (pcu-i)] [40](Figure 2B,C). The first uninodal 6-c mmo topology
nets, the HCNs [M(bpe)
2
(M′O
4
)] (M = Co or Ni; bpe = 1,2-bis(4-pyridyl)ethene; M′=MoorCr),
form the isostructural family [Ni(bpe)
2
(M′O
4
)], MOOFOUR-1-Ni, CROFOUR-1-Ni, and
WOOFOUR-1-Ni [50]. Pore size can also be reduced with shorter linkers, most commonly
through the use of pyrazine linkers [57]. The ligand libraries that have been used for the most
widely explored HCN platforms are presented in Figure S1 (see the supplemental information
online) and are exemplified by the prototypal HCNs for each pillar in Figure S2. Figure 3
presents a chronology of the key developments with respect to HCN design and properties.
HUMs: Separation of Gases Done Differently
The chemical industry accounts for 7% of the world GDP, $5.7 trillion per annum [58], and ca. 40%
of industrial energy consumption arises from separation/purification of chemical commodities [59].
This energy footprint represents ~15% of global energy consumption [60]. Ominously, a threefold
increase in demand for commodities has been projected to occur by 2050 [59]. The high energy
footprint of commodity purification processes results from their reliance upon energy-intensive
methods such as cryogenic separation, distillation, chemisorption, and solvent extraction [61].
To address energy sustainability in what is now the ‘Age of Gas’[59], physisorbents are attrac-
tive because they require relatively low energy for recycling/regeneration. Physisorbents have
therefore been studied with respect to their potential utility for a number of commodity purifica-
tions, including carbon capture, natural gas (NG) upgrading, environmental remediation
(negative emission technologies, e.g., trace CO
2
capture), and gas storage (e.g., H
2
and NG
storage) [62]. Unfortunately, many existing physisorbents, such as zeolites and MOFs/PCPs, are
handicapped by cost and/or performance limitations (e.g., poor chemical stability, interference
from water vapor, low selectivity, low working capacity, and energy-intensive regeneration). Further,
Trends
Trends
in
in
Chemistry
Chemistry
Figure 3. A Timeline of Key Discoveries in the Development of Hybrid Coordination Networks (HCNs) as Adsorbents. Chronology of the evolution of HCNs
in terms of their design (1995–2012) and properties as adsorbents (2013–present). See [25,40,41,50,51,54,57,64–66,69,70,76,99]. Abbreviations: CN, Coordination
network; HUM, hybrid ultramicroporous materials.
Trends in Chemistry
510 Trends in Chemistry, June 2020, Vol. 2, No. 6

most physisorbents exhibit binding energies of b30 kJ mol
-1
even at low loading and MOFs can suf-
fer from poor mechanical, thermal, and/or hydrolytic stability. Although chemisorbents often feature
high selectivity, the energy needed to break sorbate–sorbent chemical bonds results in a high energy
footprint and their kinetics is slow, especially for trace impurities. In essence, a catch-22
situation typically exists between uptake and selectivity, which tend to be inversely related [62].
The inherent modularity of HUMs enables fine-tuning of pore size and pore chemistry across
families of related materials. This fine tuning has enabled the discovery of energetic ‘sweet
spots’in binding sites that can overcome both the high energy penalty of chemisorbents and
the low selectivity of traditional physisorbents. Further, HUMs are ultramicroporous (b7Å[63]),
which means that when designed properly they can offer a high density of single-site, strong
(45–60 kJ mol
–1
) sorbate binding sites. The combination of these features was reported in a
2013 study of SIFSIX-3-Zn from Zaworotko and colleagues, a HUM that exhibits selectivity of
ca.2500 for CO
2
over N
2
across its loading [40]. This selectivity represents around an order of
magnitude improvement versus the previous benchmarks, zeolite 13X and Mg-MOF-74 [40]. In
2016, Zaworotko and colleagues reported a similar enhancement for C
2
H
2
over C
2
H
4
selectivity
versus the previous benchmark sorbents for SIFSIX-2-Cu-i [41]. The authors attributed the
extraordinary performance of these two HUMs to tight single binding sites enabled by strong
electrostatics from surrounding SiF
6
2–
(SIFSIX) anions (Figure 4). Further, that these ‘generation
1’(gen-1) HUMs are inherently modular enabled facile fine-tuning of pore size and pore chemistry
Trends
Trends
in
in
Chemistry
Chemistry
Figure 4. Energetic Sweet Spots Are Behind the Trace Gas Performance of HUM Physisorbents. The binding
sites in generation-1 (gen-1) HUMs. (A) CO
2
in SIFSIX-3-Zn and (B) C
2
H
2
in SIFSIX-2-Cu-i. Strong sorbent–sorbate
interactions are driven by tight binding and strong electrostatics. Abbreviations: HUM, hybrid ultramicroporous material;
S
AE
, adsorption selectivity of C
2
H
2
over ethylene, C
2
H
4
;S
CN
, adsorption selectivity of CO
2
over N
2
.
Trends in Chemistry
Trends in Chemistry, June 2020, Vol. 2, No. 6 511

Trends
Trends
in
in
Chemistr
yChemistry
[64,94]
[64,95]
[96]
[76,77,97,98]
(See figure legend at the bottom of the next page.)
Trends in Chemistry
512 Trends in Chemistry, June 2020, Vol. 2, No. 6

to create ‘generation 2’(gen-2) HUMs with further selectivity enhancements of up to another
order of magnitude across a wider range of sorbates [64–67]. The ‘molecular trap’-like binding
sites in these HUMs involve O=C
δ+
···F
δ–
and C–H
δ+
···F
δ–
electrostatic interactions, as illustrated
in Figure 4A,B for SIFSIX-3-Zn and SIFSIX-2-Cu-i, respectively. Figure 5B compares the perfor-
mance of gen-1 and gen-2 HUMs versus previous benchmark sorbents.
Such understanding of binding sites enables crystal engineers to build families or ‘platforms’of
closely related HUMs. The gas sorption performance of each member of the platform can then
be determined to gain further understanding about how to tune performance. HUMs and related
ultramicroporous materials have quickly evolved to the point that they now represent the perfor-
mance benchmarks (Figure 5B) for several other important commodity purifications: CO
2
capture
from flue gas [40,68]; trace CO
2
capture from air (both outdoor [69] and indoor [65,70]); Xe re-
movalfromKr[71]; C
2
H
2
removal from C
2
H
4
[66]; and CO
2
/C
2
H
2
and C
2
H
2
/CO
2
separation
[64]. The binding sites that enable these performance benchmarks are further discussed later.
Commodity Gas Separation Performance Benchmarks Enabled by Binding Sites
Trace (b10 000 ppm) CO
2
Capture and Removal
Energy efficient CO
2
capture is perhaps the most important global challenge and multiple needs
for capture of trace CO
2
exist, including the following: anthropogenic emissions of CO
2
are a sig-
nificant risk to global climate; CO
2
is an undesirable trace component in commodities (e.g., NG,
biogas, and acetylene); and elevated CO
2
levels cause health problems in confined spaces.
Direct Air Capture (DAC) of CO
2
from CO
2
/N
2
Dry and Wet Streams
DAC of CO
2
represents a potentially disruptive technology to address increased CO
2
levels in the at-
mosphere and in confined spaces [e.g., indoor air quality (IAQ) control]. To enable what would in ef-
fect be ‘carbon-negative’separation technologies, new selectivity benchmarks are needed to
address the relatively low (at least in the context of separations) concentration of CO
2
in the atmo-
sphere. Whereas almost all physisorbents are selective for CO
2
over N
2
, DAC requirements are be-
yond the capabilities of traditional physisorbents such as MOFs: low manufacturing cost; S
CN
N
2500; performance unaffected by competing sorbates, including H
2
O(S
CW
N100); fast kinetics;
and long-term chemical and thermal stability [62]. The gen-2 HUMs TIFSIX-3-Ni [72] and SIFSIX-
3-Ni [69] are presently the top-performing sorbents for DAC by physisorption. Whereas their CO
2
binding sites resemble that of SIFSIX-3-Zn, their S
CN
values and uptakes at low partial pressure
are much higher. However, co-adsorption by water vapor reduces performance. The HUM
SIFSIX-18-Ni-β[70] addresses co-adsorption of water vapor and is now the benchmark for IAQ con-
trol [65]. The S
CW
of 16 in SIFSIX-18-Ni-βis enabled by a hydrophobic CO
2
binding site with a CO
2
binding energy similar to that of SIFSIX-3-Zn, making it prototypal for physisorbents that exhibit
strong trace CO
2
capture performance under both dry and humid conditions [70].
NG Sweetening or CH
4
Upgrading
CH
4
stream upgradation demands CO
2
concentrations of b50 ppm in the CH
4
effluent stream. Their
close kinetic diameters (Figure 5A: CH
4
=3.8Å;CO
2
= 3.3 Å) and low boiling points (CH
4
= 112 K;
CO
2
= 195 K) mean that traditional sorbents such as activated carbons and zeolites do not exhibit
Figure 5. Target Gases and the Top-Performing Sorbents for Bulk/Trace Separation of Binary Gas Mixtures.
(A) Comparison of molecular size and boiling point of industrially important gases reveals similarity in one or both parameters
for several pairs. (B) The separation performance of gen-1 & gen-2 HUMs versus other sorbents in the context of industrially
relevant binary gas mixtures. Abbreviations: gen, Generation; HUM, hybrid ultramicroporous material; IAST, ideal adsorbed
solution theory; MOF, metal-organic framework; S
AC
, adsorption selectivity of C
2
H
2
over CO2; S
AE
, adsorption selectivity of
C
2
H
2
over ethylene, C2H4; S
CA
, adsorption selectivity of CO
2
over acetylene, C
2
H
2
;S
CM
, adsorption selectivity of CO
2
over
methane, CH4; S
CN
, adsorption selectivity of CO
2
over N
2
. See also [48,64,68,70,71,73,76,77,84–87,94–98].
Trends in Chemistry
Trends in Chemistry, June 2020, Vol. 2, No. 6 513

high enough selectivity for CO
2
over CH
4
(S
CM
N200 is needed). HUMs, however, are selective
enough to reduce CO
2
concentration from 1% or more to b50 ppm, with TIFSIX-3-Ni and
NbOFFIVE-1-Ni affording outlet purities of 25.1 and 52.3 ppm, respectively, in dynamic column
breakthrough (DCB) experiments [73].
The CO
2
capture performance of NbOFFIVE-1-Ni, TIFSIX-3-Ni, and SIFSIX-18-Ni-βis enabled by the
presence of single binding sites that strongly and selectively bind to CO
2
.Single-sitebindingisimpor-
tant since it facilitates consistent binding energy across the full range of gas loading. The binding site
for NbOFFIVE-1-Ni and TIFSIX-3-Ni is essentially that of SIFSIX-3-Zn (Figure 4A). SIFSIX-18-Ni-β,
however, combines CH···O interactions between the partially negative O
δ–
atoms of CO
2
and partially
positive methyl H
δ+
atoms to supplement C
δ+
···F
δ–
binding, thereby affording a hydrophobic but
CO
2
-philic binding pocket mimicking the carbon-fixing enzyme RuBisCO [74].
Ethylene (C
2
H
4
) Purification
C
2
H
4
, the largest volume organic feedstock (~143 Mt per year), requires removal of trace C
2
H
2
and
C
2
H
6
to produce polymer-grade ethylene (N99.95% pure). Acetylene is a trace contaminant (ca. 1%)
that poisons downstream polymerization processes. Specifications require that its concentration
must be reduced to b30 ppm. In 2016, the gen-1 HUM SIFSIX-2-Cu-i offered Norder of magnitude
improvement (S
AE
~44.54 versus 2.08) in performance versus then benchmarks for trace (1:99)
C
2
H
2
/C
2
H
4
selectivity (Figures 4B and 5B) [41]. Figure 4B illustrates the C
2
H
2
binding site in
SIFSIX-2-Cu-i, which strongly discriminates C
2
H
2
versus C
2
H
4
[41]andCO
2
[64]. More importantly,
SIFSIX-2-Cu-i offers insights into the design of gen-2 members of the same platform. SIFSIX-14-Cu-
i, with a pore size of 3.4 Å, is such a material and is the current benchmark material for C
2
H
2
/C
2
H
4
separation (S
AE
~6320) as it functions as an ideal molecular sieve.
C
2
H
6
is also a trace impurity that must be removed from C
2
H
4
.Azolate ultramicroporous
materials, a subclass of azolate MOFs known as metal azolate frameworks (MAFs) [75]with
pores b0.7 nm, such as Zn-atz-ipa [76] and MAF-49 [77], are currently the lead sorbents with
respect to equimolar C
2
H
6
/C
2
H
4
selectivity.
Acetylene Capture
C
2
H
2
poses an immediate fire and explosive hazard at N2.5% concentrations and features the widest
known flammability range: 2.5%–81%. This underscores the need for energy-efficient C
2
H
2
-capture
sorbents from C
2
H
2
/CO
2
gasmixtures[78]. Trace C
2
H
2
removal is important in the production of
acrylic/vinyl derivatives and acetylenic alcohols [79]. The ultramicroporous sorbents NKMOF-1-Ni
and the gen-2 HUM TIFSIX-2-Cu-i exhibit the required C
2
H
2
/CO
2
selectivities (S
CA
)(Figure 5B),
whereas the ultramicroporous MOF MUF-17 exhibits trace (1:99) C
2
H
2
/CO
2
separation [80].
Xe/Kr Separation
Xe and Kr are volatile noble gases present in ultra-trace concentrations in the atmosphere, 0.09
and 1.1 ppm, respectively [81]. Their boiling point differences (Figure 5A) enables cryogenic dis-
tillation to be used to separate the two gases, albeit with a high energy penalty. To reduce this
energy footprint and to align with safety guidelines to avoid possible ozone formation from radiol-
ysis in liquid air, alternative Xe/Kr separation technologies are needed [81]. Low Xe/Kr selectivity
and small Xe adsorption capacity handicap traditional physisorbents, including the MOF Ni-MOF-
74 (S
Xe/Kr
, 20:80 at 298 K ~4, Figure 5B). In contrast, the mmo topology gen-1 HUM CROFOUR-
1-Ni offers S
Xe/Kr
of ~22 and has potential for selective Xe capture from 20:80 mixtures [71].
CROFOUR-1-Ni and CROFOUR-2-Ni possess distinct micropores: one is decorated by six oxy-
gen atoms from three adjacent CrO
4
2−
linkers (two from each CrO
4
2−
moiety); the other is lined by the
functionalized organic linker (N=N from 4,4′-azobipyridine or C=C from 1,2-bis(4-pyridyl)ethylene). Kr,
Trends in Chemistry
514 Trends in Chemistry, June 2020, Vol. 2, No. 6

despite having similar binding sites to Xe, has lower polarizability and exhibits lower Q
st
(Kr)
versus Q
st
(Xe) (for CROFOUR-1-Ni the values are Xe = 37.2 kJ mol
–1
and Kr = 25 kJ mol
–1
).
C
3
Separations
Propene, C
3
H
6
, is the second highest volume feedstock in the petrochemical industry, with produc-
tion reaching 120 million tons in 2017 [82]. The manufacture of polypropene, also known as polypro-
pylene, accounts for roughly two-thirds of current propene demand. For the other third, propene is
used as a precursor in the synthesis of fine chemicals. The required purity for polymerization is
N99.5%. After C
2
H
4
/C
2
H
6
, the separation of C
3
H
6
from C
3
H
8
is the second most energy-intensive
separation process in the chemical/petrochemical industries. The difference in kinetic diameters is
only 0.2 Å, whereas their boiling points are just 5 K apart (Figure 5A). C
3
H
6
/C
3
H
8
separation typically
involves cryogenic distillation at 243 K and 30 bar, making the process energyintensive and expen-
sive [83]. Traditional physisorbents such as zeolites (4A, 5A,and 13X), titanosilicate (Na-ETS-10),
carbon molecular sieves, and MOFs are not C
3
H
6
selective enough to be used for separation of
these two gases. The gen-2 HUM NbOFFIVE-1-Ni [84] effectively sieves C
3
H
6
from C
3
H
8
. The
resulting high-purity C
3
H
8
in the effluent stream can be used directly as a fuel source.
An alternate approach to produce high-purity C
3
H
6
is to selectively capture C
3
H
4
(propyne) from
a binary mixture of C
3
H
4
/C
3
H
6
[85,86] or to simultaneously capture propyne and propadiene from
a ternary mixture of propyne/propadiene/propene [67]. The MOF BUT-310 was recently reported
to exhibit a propyne/propene selectivity of only 2.73 [87]. The leading sorbents for this separation
are HUMs [85,86] and NKMOF-1-Ni, which is an ultramicroporous MOF. SIFSIX-14-Cu-i is the
lead sorbent for propyne/propene separation thanks to its sieving mechanism to purify C
3
H
6
.
NbOFFIVE-2-Cu-i effectively purifies a ternary propyne/propadiene/propene (0.5:0.5:99) mixture
to release 99.9999% ultra-pure propene as the effluent stream.
The C
3
binding sites in the C
3
H
4
selective HUM ZJUT-1a reveal that immobilization of dual func-
tionalities (–NH
2
and SiF
6
2−
) into the ultramicropores can induce a multibinding environment to in-
crease C
3
H
4
selectivity over C
3
H
6
. This performance reduces when amines no longer line the
pores, as in SIFSIX-3-Ni [85]. SIFSIX-2-Cu-i has a pore size of ~3.4 Å that matches well with
the dimensions of C
3
H
4
but not C
3
H
6
, which is sieved out [86]. The key unmet need in C
3
sepa-
rations is a sorbent that preferentially adsorbs and removes propane from C
3
H
6
.
Concluding Remarks
Finding the Right Haystacks to Enable Different Things That We Can Do
The potential utility of ultramicroporous materials, especially HUMs, for gas purification is evident
thanks to exploitation of crystal engineering [3,4,88,89], in situ characterization, and modeling
[90,91]. Specifically, systematic control over pore-size and pore-chemistry through crystal engi-
neering has granted access to platforms of ultramicroporous porous materials with ultra-high
selectivity. Elucidation of the structure–function relationships that define the thermodynamics
and kinetics of sorption come from modeling and structural studies on respective binding sites
[76]. Overall, HUMs have taught us that physisorbents with binding energies on the order of
45–60 kJ mol
–1
lie in the ‘sweet spot’: for (i) capturing trace impurities at ambient conditions
(Figure 4), and (ii) releasing the captured impurity through mild temperature, vacuum, or pressure
swings. HUMs and related ultramicroporous materials have thereby taught us an important
message: a high density of tight binding sites (enabled by ultramicroporosity) is seemingly a
prerequisite for ultra-selective interactions with C
1
,C
2
,C
3
,andnoblegases.
The thermodynamic and kinetic requirements for efficient separation processes using
physisorbents [62] are now in hand, but the full spectrum of performance parameters [92]has
Outstanding Questions
How does one find the right needle in
the right haystack? There are N75
000 CNs (including MOFs and HUMs)
and the number of potential porous
CNs is effectively infinite. The situation
is beyond the proverbial ‘finding a
needle in a haystack’,ratheritisto
first find the right haystack (platform)
before screening; predictive modeling
can address this challenge.
Can we mitigatethe effects of humidity?
Humidity is ubiquitous and water
molecules have both a smaller kinetic
diameter and, for most sorbents, a
stronger binding affinity than other
trace impurities. Hydrophobic coatings
or changing pore size/chemistry could
solve the humidity problem but at the
expense of reducing selectivity for the
desired impurity.
Can crystal engineering address the
high cost of new sorbents? The bulk
cost and commercial availability of
substrates should be criteria applied at
the design stage of sorbent research.
Limiting which metals and linker ligands
are used still leaves an enormous
number of porous CNs that are yet to
be discovered.
Can crystal engineering be compatible
with process engineering? Process
engineering challenges include control
of particle size, shape, and defects
during sorbent manufacture and
formulation as pellets. The emergence
of mechanochemistry offers potential
for continuous manufacture of the final
product in pellet form, especially if the
substrates selected are amenable to
self-assembly and formulation.
Trends in Chemistry
Trends in Chemistry, June 2020, Vol. 2, No. 6 515

not yet been addressed. The impact of humidity upon sorption performance is one such param-
eter and progress has been made: water sorption can be controlled by alkyl group-induced hy-
drophobicity [70]; certain metal nodes [e.g., Zr(IV), Nb(V)] can address hydrolytic stability [93].
Nevertheless, it is a big leap from promising sorbents in the laboratory to new industrial-scale pro-
cesses. Further, most research has thus far focused upon binary mixtures of gases, even though
the most important applications (e.g., biogas production, syngas production, DAC of CO
2
, and
NG sweetening) involve multicomponent gas mixtures of varying composition. Very recently,
the ultra-high selectivity exhibited by HUMs enabled proof-of-principle for synergistic sorbent
separation technology (SSST) [76] through the use of bespoke physisorbents. SSST enables
one-step production of polymer-grade (N99.9%) ethylene from four-component gas mixtures
through the use of three ultramicroporous physisorbents, one for each impurity (CO
2
,C
2
H
2
,
and C
2
H
6
;Figure 6). Whereas purification of such multicomponent gas mixtures was hitherto un-
realistic with low selectivity sorbents, the recent trends suggest that Feynman’s vison, bottom-up
design approaches to materials that enable ‘different things that we can do’,couldfinally be
within reach. Nevertheless, there remain a spectrum of challenges that mustbe overcome before
implementation can occur (see Outstanding Questions).
Acknowledgments
M.J.Z. gratefully acknowledges the Science Foundation Ireland (13/RP/B2549 and 16/IA/4624).
Supplemental Information
Supplemental information associated with this article can be found online at https://doi.org/10.1016/j.trechm.2020.02.013.
References
1. Feynman, R.P. (1999) There’s plenty of room at the bottom.
In ThePleasureofFindingThingsOut,TheBestShort
Works of Richard P. Feynman (Robbins, J., ed.), Perseus
Publishing
2. Bhattacharya, S. et al. (2018) The role of hydrogen bonding in
co-crystals. In Co-crystals: Preparation, Characteri zation and
Applications (Aakeroy, C.B. and Sinha, A.S., eds), pp. 3379,
The Royal Society of Chemistry
3. Pepinsky, R. (1955) Crystal engineering: a new concept in
crystallography. Phys. Rev. 100, 971
4. Schmidt, G.M.J. (1971) Photodimerization in the solid state.
Pure Appl. Chem. 27, 647–678
5. Lehn, J.-M. (1988) Supramolecular chemistry—scope and
perspectives molecules, supermolecules, and molecular
devices (Nobel lecture). Angew. Chem. Int. Ed. Engl. 27,
89–112
Trends
Trends
in
in
Chemistry
Chemistry
Figure 6. The Synergistic Sorbent Separation Technology Approach and Its Proof-of-Principle. SIFSIX-3-Ni
(trace and bulk CO
2
), TIFSIX-2-Cu-i (trace C
2
H
2
), and Zn-atz-ipa (trace and bulk C
2
H
6
) enable one-step C
2
H
4
purification
from 1/1/1/1 or 1/33/33/33 (C
2
H
2
/CO
2
/C
2
H
4
/C
2
H
6
) gas mixtures [76].
Trends in Chemistry
516 Trends in Chemistry, June 2020, Vol. 2, No. 6

6. Groom, C.R. et al. (2016) The Cambridge structural database.
Acta Crystallogr. B 72, 171–179
7. Ermer, O. (1988) Five-fold diamond structure of adamantane-
1,3,5,7-tetracarboxylic acid. J. Am. Chem. Soc. 110,
3747–3754
8. Desiraju, G.R. andSarma, J.A.R.P.(1983) Crystalengineering via
donor–acceptor interactions. X-ray crystal structure and solid
state reactivity of the 1:1 complex, 3,4-dimethoxycinnamic
acid–2,4-dinitrocinnamic acid. J. Chem. Soc. Chem. Commun.
1, 45–46
9. Etter, M.C. et al. (1989) Solid-state nucleophilic aromatic sub-
stitution reaction of a carboxylic acid cocrystal. Tetrahedron
Lett. 30, 3617–3620
10. Walsh, R.D.B. et al. (2003) Crystal engineering of the
composition of pharmaceutical phases. Chem. Commun. 9,
186–187
11. Huang, N. et al. (2016) Covalent organic frameworks: a mate-
rials platform for structural and functional designs. Nat. Rev.
Mater. 1, 16068
12. Vishweshwar, P. et al. (2006) Pharmaceutical co-crystals.
J. Pharm. Sci. 95, 499–516
13. Almarsson, Ö. et al. (2012) The A to Z of pharmaceutical
cocrystals: a decade of fast-moving new science and patents.
Pharm. Pat. Anal. 1, 313–327
14. Kavanagh, O.N. et al. (2019) Pharmaceutical cocrystals: from
serendipity to design to application. Drug Discov. Toda y 24,
796–804
15. Batten, S.R. et al. (2013) Terminology of metal–organic frame-
works and coordination polymers (IUPAC Recommendations
2013). Pure Appl. Chem. 85, 1715–1724
16. Hoskins, B.F. and Robson, R. (1990) Design and construction
of a new class of scaffolding-like materials comprising infinite
polymeric frameworks of 3D-linked molecular rods. A reap-
praisal of the zinc cyanide and cadmium cyanide structures
and the synthesis and structure of the diamond-related
frameworks [N(CH
3
)
4
][Cu
I
Zn
II
(CN)
4
]andCu
I
[4,4′,4′′,4′′′-
tetracyanotetraphenylmethane]BF
4
.xC
6
H
5
NO
2
.J. Am. Chem.
Soc. 112, 1546–1554
17. Yaghi, O.M. et al. (2003) Reticular synthesis and the design of
new materials. Nature 423, 705–714
18. International Council for Harmonisation of Technical
Requirements for Pharmaceuticals for Human Use (ICH)
Guideline (2003) ICH Q1A (R 2) Stability Testing of New Drug
Substances and Drug Products, Pharmaceutical Guidance
19. Bailar Jr., J. (1964) Coordination Polymers, Interscience,
pp. 1–25
20. Keggin, J.F. and Miles, F.D. (1936) Structures and formulae of
the Prussian blues and related compounds. Nature 137,
577–578
21. Powell, H.M. and Rayner, J.H. (1949) Clathrate compound
formed by benzene with an ammonia–nickel cyanide complex.
Nature 163, 566–567
22. Gable, R.W. et al.(1990) A new type of interpenetration involving
enmeshed independent square grid sheets. The structure of
diaquabis-(4,4′-bipyridine)zinc hexafluorosilicate. J. Chem. Soc.
Chem. Commun. 23, 1677–1678
23. Zaworotko, M.J. (1994) Crystal engineering of diamondoid
networks. Chem. Soc. Rev. 23, 283–288
24. Fujita, M. et al. (1994) Preparation, clathration ability, and catal-
ysis of a two-dimensional square network material composed
of cadmium(II) and 4,4′-bipyridine. J. Am. Chem. Soc. 116,
1151–1152
25. Subramanian, S. and Zaworotko, M.J. (1995) Porous solids by
design: [Zn(4,4′-bpy)
2
(SiF
6
)]
n
·xDMF, a single framework octa-
hedral coordination polymer with large square channels.
Angew. Chem. Int. Ed. Engl. 34, 2127–2129
26. Kitagawa, S. et al. (1992) Synthesis of the novel infinite-sheet
and -chain copper(I) complex polymers {[Cu(C
4
H
4
N
2
)
3/2
(CH
3
CN)](PF)
6
.cntdot.0.5C
3
H
6
O}.infin. and {[Cu
2
(C
8
H
12
N
2
)
3
]
(ClO
4
)
2
}.infin. and their x-ray crystal structures. Inorg. Chem.
31, 1714–1717
27. Gardner, G.B. et al. (1995) Spontaneous assembly of a hinged
coordination network. Nature 374, 792–795
28. Moghadam, P.Z. et al. (2017) Development of a Cambridge
structural database subset: a collection of metal–organic
frameworks for past, present, and future. Chem. Mater. 29,
2618–2625
29. Batten, S.R. and Robson, R. (1998) Interpenetrating nets: or-
dered, periodic entanglement. Angew. Chem. Int. Ed. 37,
1460–1494
30. Batten, S.R. (2001) Topology of interpenetration. CrystEngComm
3, 67–72
31. Zhang, J. et al. (2009) Temperature and concentration control
over interpenetration in a metal−organic material. J. Am.
Chem. Soc. 131, 17040–17041
32. Alexandrov, E.V. et al. (2017) How 2-periodic coordination
networks are interweaved: entanglement isomerism and
polymorphism. CrystEngComm 19, 1993–2006
33. Kondo, M. et al. (1997) Three-dimensional framework with
channeling cavities for small molecules: {[M
2
(4,4′-bpy)
3
(NO
3
)
4
]·xH
2
O}n (M = Co, Ni, Zn). Angew. Chem. Int. Ed. Engl. 36,
1725–1727
34. Li, H. et al. (1998) Establishing microporosity in open metal−organic
frameworks: gas sorption isotherms for Zn(BDC) (BDC = 1,4-
benzenedicarboxylate). J. Am. Chem. Soc. 120, 8571–8572
35. Ma, S. and Zhou, H.-C. (2006) A metal−organic framework with
entatic metal centers exhibiting high gas adsorption affinity.
J. Am. Chem. Soc. 128, 11734–11735
36. Chui, S.S.-Y. et al. (1999) A chemically functionalizable nanoporous
material [Cu
3
(TMA)
2
(H
2
O)
3
]
n
.Science 283, 1148–1150
37. Li, H. et al. (1999) Designand synthesis of an exceptionally stable
and highly porous metal-organic framework. Nature 402,
276–279
38. Hönicke, I.M. et al. (2018) Balancing mechanical stability and
ultrahigh porosity in crystalline framework materials. Angew.
Chem. Int. Ed. 57, 13780–13783
39. Farha, O.K. et al. (2012) Metal–organic framewor k materials
with ultrahigh surface areas: is the sky the limit? J. Am.
Chem. Soc. 134, 15016–15021
40. Nugent, P. et al. (2013) Porous materials with optimal adsorp-
tion thermodynamics and kinetics for CO
2
separation. Nature
495, 80–84
41. Cui, X. et al. (2016) Pore chemistry and size control in hybrid
porous materials for acetylene capture from ethylene. Science
353, 141–144
42. O’Nolan, D. et al. (2017) Water vapor sorption in hybrid pillared
square grid materials. J. Am. Chem. Soc. 139, 8508–8513
43. Burtch, N.C. et al. (2014) Water stability and adsorption in
metal–organic frameworks. Chem. Rev. 114, 10575–10612
44. Lin, R.-B. et al. (2019) Microporous metal-organic framework
materials for gas separation. Chem. 6, 337–363
45. Chen, B. et al. (2001) Interwoven metal-organic framework on
a periodic minimal surface with extra-large pores. Science 291,
1021–1023
46. Wilson, S.T. et al. (1982) Aluminophosphate molecular sieves:
a new class of microporous crystalline inorganic solids.
J. Am. Chem. Soc. 104, 1146–1147
47. Buser, H.J. et al. (1977) The crystal structure of Prussian blue:
Fe
4
[Fe(CN)
6
]
3
.xH
2
O. Inorg. Chem. 16, 2704–2710
48. Roque, J. et al. (2007) Porous hexacyanocobaltates(III): r ole of the
metal on the framework properties. Microporous Mesoporous
Mater. 103, 57–71
49. Dybtsev, D.N. et al. (2004) Rigid and flexible: a highly porous
metal–organic framework with unusual guest-dependent dy-
namic behavior. Angew. Chem. Int. Ed. 43, 5033–5036
50. Mohamed, M.H. et al. (2012) Highly selective CO
2
uptake
in uninodal 6-connected “mmo”nets based upon MO
4
2–
(M = Cr, Mo) pillars. J. Am. Chem. Soc. 134, 19556–19559
51. Noro, S.-i. et al. (2002) Framework engineering by anions and
porous functionalities of Cu(II)/4,4′-bpy coordination polymers.
J. Am. Chem. Soc. 124, 2568–2583
52. Noro, S.-i. et al. (2000) A new, methane adsorbent, porous co-
ordination polymer [{CuSiF
6
(4,4′-bipyridine)
2
}
n
]. Angew. Chem.
Int. Ed. 39, 2081–2084
53. Wang, S. et al. (2011) Anion-tuned sorption and catalytic
properties of a soft metal–organic solid with polycatenated
frameworks. J. Mater. Chem. 21, 7098–7104
54. Cadiau, A. et al. (2017) Hydrolytically stable fluorinated metal-
organic frameworks for energy-efficient dehydration. Science
356, 731–735
Trends in Chemistry
Trends in Chemistry, June 2020, Vol. 2, No. 6 517

55. Kopf, A.L. et al. (2005) Poly[nickel(II)-di-μ-4,4′-bipyridyl-κ-4N:N′-
μ-dichromato-κ-2O:O′] and poly[copper(II)-di-μ-4,4′-bipyridyl-
κ-4N:N′-μ-dichromato-κ-2O:O′]. Acta Crystallogr. C 61,
m165–m168
56. Lin, M.-J. et al. (2011) Molecular tectonics: control of interpen-
etration in cuboid 3-D coordination networks. CrystEngComm
13, 776–778
57. Halasyamani, P. et al. (1996) Syntheses and structures of two
new Cu/Nb/pyrazine complexes: three dimensional CuNb
(pyz)
2
OF
5
·(pyz)(H
2
O) and two dimensional [Cu(pyz)
2.5
]
+
[NbF
6
]
−
· (pyz). Z. Anorg. Allg. Chem. 622, 479–485
58. The International Council of Chemical Associations (2019)
Chemical Industry Contributes $5.7 Trillion to Global GDP
and Supports 120 Million Jobs, New Report Shows, ICCA
59. Kitagawa, S. (2015) Porous materials and the age of gas.
Angew. Chem. Int. Ed. 54, 10686–10687
60. Sholl, D.S. and Lively, R.P. (2016) Seven chemical separations
to change the world. Nature 532, 435–437
61. Yang, R.T. (1997) Gas Separati on by Adsorption Processes,
Elsevier
62. Oschatz, M. and Antonietti, M. (2018) A search for selectivity to
enable CO
2
capture with porous adsorbents. Energy Environ.
Sci. 11, 57–70
63. Sing, K.S.W. (1982) Reporting physisorption data for gas/solid
systems. Pure Appl. Chem. 54, 2201–2218
64. Chen, K.-J. et al. (2016) Benchmark C
2
H
2
/CO
2
and CO
2
/C
2
H
2
separation by two closely related hybrid ultramicroporous
materials. Chem 1, 753–765
65. Bhatt, P.M. et al. (2016) A fine-tuned fluorinated MOF ad-
dresses the needs for trace CO
2
removal and air capture
using physisorption. J. Am. Chem. Soc. 138, 9301–9307
66. Li, B. et al. (2017) An ideal molecular sieve for acetylene re-
moval from ethylene with record selectivity and productivity.
Adv. Mater. 29, 1704210
67. Yang, L. et al. (2018) An asymmetric anion-pillared metal–
organic framework as a multisite adsorbent enables simulta-
neous removal of propyne and propadiene from propylene.
Angew. Chem. Int. Ed. 57, 13145–13149
68. Madden, D.G. et al. (2017) Flue-gas and direct-air capture of
CO
2
by porous metal-organic materials. Philos. Trans. R.
Soc. A Math. Phys. Eng. Sci. 375, 20160025
69. Kumar, A. et al. (2015) Direct air capture of CO
2
by physisorbent
materials.Angew. Chem. Int. Ed. 54, 14372–14377
70. Mukherjee, S. et al. (2019) Trace CO
2
capture by an
ultramicroporous physisorbent with low water affinity. Sci.
Adv. 5, eaax9171
71. Mohamed, M.H. et al. (2016) Hybrid ultra-microporous mate-
rials for selective xenon adsorption and separation. Angew.
Chem. Int. Ed. 55, 8285–8289
72. Kumar, A. et al. (2017) Hybrid ultramicroporous materials
(HUMs) with enhanced stability and trace carbon capture per-
formance. Chem. Commun. 53, 5946–5949
73. Madden, D.G. et al. (2019) Highly selective CO
2
removal for
one-step liquefied natural gas processing by physisorbents.
Chem. Commun. 55, 3219–3222
74. van Lun, M. et al. (2014) CO
2
and O
2
distribution in Rubisco
suggests the small subunit functions as a CO
2
reservoir.
J. Am. Chem. Soc. 136, 3165–3171
75. Zhang, J.-P.et al. (2012) Metal azolate frameworks: from crystal
engineer ing to functio nal materials. Chem. Rev. 112, 1001–1033
76. Chen, K.-J. et al. (2019) Synergistic sorbent separation for
one-step ethylene purification from a four-component mixture.
Science 366, 241–246
77. Liao, P.-Q. et al. (2015) Efficient purification of ethene by an ethane-
trapping metal-organic framework. Nat. Commun. 6, 8697
78. Molero, H. et al. (1999) The hydrogenation of acetylene cata-
lyzed by palladium: hydrogen pressure dependence. J. Catal.
181, 49–56
79. Pässler, P. et al. (2011) Ullmann's Encyclop edia of Industrial
Chemistry, Wiley
80. Qazvini, O.T. et al. (2019) Multipurpose metal–organic frame-
work for the adsorption of acetylene: ethylene purification and
carbon dioxide removal. Chem. Mater. 31, 4919–4926
81. Banerjee, D. et al. (2015) Potential of metal–organic frameworks for
separation of xenon and krypton. Acc. Chem. Res. 48, 211–219
82. Buckl, K. and Meiswinkel, A. (2002) Ullmann’s Encyclopedia of
Industrial Chemistry, Wiley, pp. 337–340
83. Eldridge, R.B. (1993) Olefin/paraffin separation technology: a
review. Ind. Eng. Chem. Res. 32, 2208–2212
84. Cadiau, A. et al. (2016) A metal-organic framework–based
splitter for separating propylene from propane. Science 353,
137–140
85. Wen, H.-M. et al. (2018) Fine-tuning of nano-traps in a stable
metal–organic framework for highly efficient removal of
propyne from propylene. J. Mater. Chem. A 6, 6931–6937
86. Li, L. et al. (2018) A metal–organic framework with suitable pore
size and specific functionalsites for the removal of trace propyne
from propylene. Angew. Chem. Int. Ed. 57, 15183–15188
87. Wu, X.-Q. et al. (2019) Integrating multiple adsorption sitesand
tortuous diffusion paths into a metal–organic framework for
C
3
H
4
/C
3
H
6
separation. J. Mater. Chem. A 7, 25254–25257
88. Desiraju, G.R. and Parshall, G.W. (1989) Crystal engineering:
the design of organic solids. Mater. Sci. Monogr. 54, 312
89. Moulton, B. and Zaworotko, M.J. (2001) From molecules to
crystal engineering: supramolecular isomerism and polymor-
phism in network solids. Chem. Rev. 101, 1629–1658
90. Rabone, J. et al. (2010) An adaptable peptide-based porous
material. Science 329, 1053–1057
91. Hazra, A. et al. (2019) CO
2
-induced single-crystal to single-
crystal transformations of an interpenetrated flexible MOF ex-
plained by in situ crystallographic analysis and molecular
modeling. Chem. Sci. 10, 10018–10024
92. Mukherjee, S. et al. (2019) Metal-organic framework based
carbon capture and purification technologies for clean environ-
ment. In Metal-Organic Frameworks (MOFs) for Environmental
Applications (Ghosh, S.K., ed.), pp. 5–61, Elsevier
93. Wang, Q. et al. (2019) Separation of Xe from Kr with record se-
lectivity and productivity in anion-pillared ultramicroporous ma-
terials by inverse size-sieving effect. Angew. Chem. Int. Ed. 59,
3423–3428
94. Li, L. et al. (2019) Inverse adsorption separation of CO
2
/C
2
H
2
mixture in cyclodextrin-based metal–organic frameworks.
ACS Appl. Mater. Interfaces 11, 2543–2550
95. Peng, Y.-L. et al. (2018) Robust ultramicroporous metal–
organic frameworks with benchmark affinity for acetylene.
Angew. Chem. Int. Ed. 57, 10971–10975
96. Lin, R.-B. et al. (2018) Molecular sieving of ethylene from
ethane using a rigid metal–organic framework. Nat. M ater.
17, 1128–1133
97. Lin, R.-B. et al. (2018) Boosting ethane/ethylene separation
within isoreticular ultramicroporous metal–organic frameworks.
J. Am. Chem. Soc. 140, 12940–12946
98. Li, L. et al. (2018) Ethane/ethylene separation in a metal-organic
framework withiron-peroxo sites. Science 362, 443–446
99. Yoshihito, H. et al. (2001) Three-dimensional polymers with metal–
oxygen and metal–ligand chains: [Mn(II)(4,4′-bipy)2(Cr
2
O
7
)(H
2
O)
2
]
and [Co(II)(4,4′-bipy)
2
(Cr
2
O
7
)]. Chem. Lett. 30, 562–563
100. Rajendran, A. et al. (2008) Correction procedures for extra-
column effects in dynamic column breakthrough experiments.
Chem. Eng. Sci. 63, 2696–2706
Trends in Chemistry
518 Trends in Chemistry, June 2020, Vol. 2, No. 6

Supplementary Information
Crystal engineering of hybrid coordination networks: from form to function
Soumya Mukherjee and Michael J. Zaworotko*
Bernal Institute, Department of Chemical Sciences, University of Limerick, Limerick V94
T9PX, Republic of Ireland.
* Correspondence: xtal@ul.ie (M.J. Zaworotko)

Figure S1. A) Linker ligands used for FOXY (M'OxF6-x2-) pcu nets [M (L)2(M′OF5)]n (L =
1, 2) (left) and M'FFIVE pcu net [M(L')2(M'F5)]n (L' = 1) (right); B) linker ligands 1-20
form pcu nets of formula [M(L)2(M'F6)]n; C) linker ligands used for mmo nets [M
(L)2(M′O4)]n (L = 1‒2) (left) and DICRO pcu nets [M(L')2(Cr2O7)]n (L' = 1‒8) (right). This
scheme of ligand codes forms the basis of nomenclature (inorganic anion-linker ligand code-
metal ion) in HCNs that belong to distinct platforms or families.

Figure S2. Prototypal HCNs for some of the most widely studied pillar anions and
topologies: A) SIFSIX-1-Zn (non-interpenetrated pcu net based upon fluorometallate); B)
SIFSIX-2-Cu-i (2-fold interpenetrated pcu net based upon fluorometallate, F atoms are
disordered); C) NbOFFIVE-1-Ni (non-interpenetrated pcu net based upon
fluorooxymetallate or FOXY anion); D) DICRO-3-Ni-i (interpenetrated pcu net based upon
linear oxometallate); E) MOOFOUR-1-Ni (self-penetrated mmo net based upon angular
oxometallate) (colour codes: grey = C, blue = N, red = O, green = F, orange = Si, purple =
Nb, yellow = Cr, lilac = molybdenum, saffron = metal cation, interpenetrated nets in cyan).