Capture of toxic gases in MOFs: SO2, H2S, NH3 and NOx

MOFs are promising candidates for the capture of toxic gases since their adsorption properties can be tuned as a function of the topology and chemical composition of the pores. Although the main drawback of MOFs is their vulnerability to these highly corrosive gases which can compromise their chemical stability, remarkable examples have demonstrated high chemical stability to SO2, H2S, NH3 and NOx. Understanding the role of different chemical functionalities, within the pores of MOFs, is the key for accomplishing superior captures of these toxic gases. Thus, the interactions of such functional groups (coordinatively unsaturated metal sites, μ-OH groups, defective sites and halogen groups) with these toxic molecules, not only determines the capture properties of MOFs, but also can provide a guideline for the desigh of new multi-functionalised MOF materials. Thus, this perspective aims to provide valuable information on the significant progress on this environmental-remediation field, which could inspire more investigators to provide more and novel research on such challenging task.


Introduction
The accelerated growth of our modern society demands huge amounts of energy. Unwittingly, in order to provide these high energetic levels, an indiscriminate combustion of large volumes of fossil fuels occurs, leading to an incommensurable release of toxic pollutants to the atmosphere. Emissions of anthropogenic air pollutants generate a vast range of health complications (e.g., premature death and morbidity). Additionally, these air pollutants are also responsible for the reduction of the biodiversity, crop damages and acidication of soils and waters. 1 As an example of the adverse impacts to humans, the World Health Organization (WHO) recently announced that air pollution was responsible for the premature mortality of approximately 4.2 million people in 2016 alone. 2 In fact, air pollution is now the single largest environmental health risk worldwide since it is responsible for one in eight premature global deaths. 3 For example, PM 2.5 (ne inhalable particles, with diameters that are generally 2.5 micrometres and smaller) are responsible for approximately half of the deaths related to air pollution. 4 Thus, PM 2.5 have been classied as the air pollutant with the highest impact in premature mortality, 5 and a reduction of their emissions is crucial. Such reduction can be achieved in two steps: (i) mitigation of primary PM 2.5 emissions and (ii) extenuation of secondary inorganic aerosols (SIA); which are oxidised from precursor emissions such as sulphur dioxide (SO 2 ), nitrogen oxides (NO x ), ammonia (NH 3 ) and volatile organic compounds (VOCs). 6 Thus, in order to extenuate these emissions, different actions have been taken, such as the shutdown of some coal-red power plants and their replacement by thermoelectric power plants, importing electricity from other countries with strict restrictions on the electricity production and the use of more environmentally friendly energy sources. 7 In addition to these actions, the development of efficient technologies for the capture of toxic gases (e.g., NO x , SO 2 , NH 3 and H 2 S) from static and mobile sources is necessary, in order to achieve a cleaner environment. 8 Porous metal-organic frameworks (MOFs) or porous coordination polymers (PCPs) are amongst the most promising candidates for the capture of these toxic gases since their sorption selectivity is directly tuneable as a function of the topology and chemical composition of the pores. 9 Undoubtedly, for NO x , SO 2 , NH 3 and H 2 S capture, there is a substantial emphasis on optimising the interactions between MOF materials and these toxic molecules, leading to the discovery of new functional porous materials with enhanced gas adsorption properties. 10 Although MOF materials have shown very promising capabilities for the capture of these toxic gases, their main drawback is their vulnerability to these highly corrosive molecules capable of compromising the chemical stability of the MOFs. NO x , SO 2 , NH 3 and H 2 S can disrupt the coordination bonds between the organic ligands (e.g., carboxylate) and the metal centres, occasioning the breakdown of the MOF structure. Therefore, chemical stability of the MOFs is a fundamental requirement for the capture of these toxic gases. The present contribution aims to provide a useful reference in the eld of capturing toxic gases in MOFs, emphasising on their chemical stability and the role of the functional groups to enhance such captures. We hope to encourage more research groups to explore the exciting frontiers of science in environmentalremediation applications for MOFs.

Sulphur dioxide
The anthropogenic release of SO 2 into the atmosphere is mainly due to the combustion of fossil fuels (e.g., electric power generation). One of the rst environmental problems related to SO 2 was observed during the last century in the acidic deposition. 11 To avoid this, since 1990 many European countries have considerably reduced their SO 2 emissions. 12 Nevertheless, the rapid industrialisation of some developing countries has caused a constant presence of SO 2 in the atmosphere, 13 risking the health of millions of people (vide supra). SO 2 is a colourless, non-ammable and corrosive gas with high solubility (120 g l À1 ) in water. Among the technologies for SO 2 capture, wet and dry ue-gas desulfurization (FGD) processes are commonly used, 14 where the cost and recyclability depend on the used technology. Among them, alkaline solutions, activated carbons, zeolites and silica have typically showed low SO 2 adsorption capacities, accompanied by the corrosion of pipelines and high energy regeneration overheads. 15 Thus, the design of new materials is extremely necessary in order to meet all the challenges that the capture of SO 2 signies. MOF materials are an exciting alternative for the capture of SO 2 as they have demonstrated promising results for the SO 2 capture assignment, mainly due to their specic chemical functionality and pore dimensions. 16 The following section focuses on the highlights of remarkable MOF materials stable towards SO 2 , with the emphasis not only on their capture performance, but also on showing the role of different chemical functionalities which are the key for the SO 2 capture.

MOFs for SO 2 capture
The chemical nature of the SO 2 molecule and its main binding modes: oxygen, an electron-rich atom that can act as a Lewis base; and sulphur that can be a Lewis acid site, have been exploited to improve its adsorption on the porous surface of the MOFs. Due to the specic functionalisation of these MOFs, interesting SO 2 capture results have been achieved. Such functionalisation can be classied in four main categories: (i) coordinatively unsaturated metal sites (open metal sites), (ii) m-OH groups, (iii) defective sites and (iv) halogen functionalisation.
(i) Coordinatively unsaturated metal sites. Although some research has proven the capture of SO 2 by different MOFs, the majority of these examples have shown to be chemically unstable upon SO 2 exposure. 17 For example, MOF-177 (an open Zn 2+ sites system reported by Janiak and co-workers 18 ) holds the SO 2 capture record (25.7 mmol g À1 at 298 K and 1 bar) but suffers a partial structural degradation, aer the exposure to SO 2 , as demonstrated by PXRD and a signicant BET surface area reduction. 18 Conversely, MFM-170 an open Cu 2+ sites MOF; [Cu 2 (L)] (H 4 L ¼ 4 0 ,4 000 -(pyridine-3,5-diyl)bis([1,1 0 -biphenyl]-3,5dicarboxylic acid)), not only demonstrated to be highly stable to SO 2 (17.5 mmol g À1 at 298 K and 1 bar), but also to humid SO 2 . 19 Yang and Schröder elegantly demonstrated the reversible coordination of SO 2 to open Cu 2+ sites in MFM-170 by in situ synchrotron single-crystal X-ray diffraction, in situ FTIR spectroscopy and in situ inelastic neutron scattering (INS). 19 In brief, all of this comprehensive experimental evidence corroborated that the Cu 2+ site is the thermodynamically strongest SO 2 binding site (see Fig. 1), but is sufficiently weak to be almost entirely desorbed upon reducing the pressure. The material can also be fully regenerated by heating to 400 K without any loss of crystallinity even aer completing y SO 2 adsorptiondesorption cycles at 298 K. 19 Our group reported a partially uorinated version of MIL-101(Cr) named MIL-101(Cr)-4F(1%). 20 The particularity of MIL-101(Cr)-4F(1%) is the different chemical character of the Cr 3+ metal centres, in comparison to MIL-101(Cr). The incorporation of uorine in MIL-101(Cr) promoted a higher acidity of some of the Cr 3+ metal centres (open metal sites), due to the capability of uorine to attract electrons. 20 Such difference afforded a total SO 2 capture of 18.4 mmol g À1 at 298 K and up to 1 bar, chemical stability towards dry and humid SO 2 and an exceptional cycling performance with facile regeneration. 20 This uptake represents the highest SO 2 capture for a structurally stable MOF material. In addition, in situ DRIFT spectroscopy upon the adsorption of CO demonstrated the efficient packing of SO 2 molecules within MIL-101(Cr)-4F(1%). Thus, the SO 2 uptake mechanism takes place in three stages: (i) adsorption at acid (Lewis) Cr 3+ sites with a relatively high heat of adsorption for SO 2  In order to answer this question, their comparison is presented in Table 1, allowing to propose a stability hypothesis for these materials.
The design of stable frameworks for capture, connement and release of corrosive gases uses tools centred in the robustness of the metal-linker bond, the use of inert or higher-valent metals and, the use of poly-nuclear secondary building units (SBUs). 21 In the case of MOF-177, MFM-170 and MIL-101(Cr)-4F(1%), these exhibit at least one of such design tools. These three MOFs are composed of carboxylate ligands and robust SBUs, this multinuclearity gives, in theory, a better thermal and chemical stability. Nonetheless, in the case of MOF-177, 18  and Zn-DMOF. 27 These comparative results suggest that in the case of Zn-based materials, the robustness of the SBU plays a secondary role in chemical stability against SO 2 , and the fact that they are constructed from a divalent metal that forms labile bonds, 28 makes them more susceptible to collapse. Although MFM-170 contains the Cu 2+ paddlewheel SBU, a cluster frequently instable under harsh conditions such as corrosive or acidic gases, 29,30 it is stable even aer 50 adsorption-desorption cycles of SO 2 . 19 This extraordinary chemical stability is mainly due to two important factors: (i) the unusual paddlewheel SBU formed in the framework and (ii) additional intermolecular interactions formed between SO 2 and the porewalls of the framework. The di-cooper paddlewheel SBU shows the particularity of only one of the Cu 2+ centres axially  25 It is signicant that this interaction does not represent the main SO 2 binding site due to the steric hindrance generated by a neighbouring SO 2 molecule. This latter molecule has several interactions: (i) with the C-H portion of the ligand, (ii) with the oxygen atoms in the paddlewheel SBU, and (iii) with other SO 2 molecules. All these interactions promote an efficient packing within the pore surface, maximising the adsorbateadsorbent bonds and presumably avoids the frame destruction. Both materials reviewed above, MOF-170 and MFM-177, are formed by rst-row transition metals Zn 2+ and Cu 2+ , respectively, and follow the stability order of the Irving-Williams series for divalent metals: Mn < Fe < Co < Ni < Cu > Zn. 31 Thus, the high SO 2 capture can be attributed to the large BET surface area and pore size rather than to the local pore environment, while structural stability is associated with the type of metal, and the interactions adsorbate-adsorbent. Additionally, the use of higher-valent metal ions such as Al 3+ , Sc 3+ , Cr 3+ or Zr 4+ allows the formation of stronger M-O bonds with carboxylate ligands compared to divalent ions. 21 Thus, the lability of metal-ligand bonding is higher in Cu 2+ or Zn 2+ -based MOFs that Cr 3+ MOFs. As a consequence, MOFs constructed from divalent metals such Cu 2+ and Zn 2+ , typically show partial or total degradation under harsh conditions. 32 Due to the inertness of the Cr-O bond, Cr 3+ -based MOFs highly resistant to the attack of acid and base guests were reported. 33,34 As described above, the third material, MIL-101(Cr)-4F(1%), has the highest SO 2 adsorption among SO 2 -stable MOFs. 20 This framework poses a robust trinuclear metal cluster with coordinatively unsaturated chromium sites aer the activation process (Table 1). When comparing MIL-101(Cr)-4F(1%) to MFM-170, both materials stable towards SO 2 , we can observe that the Cr 3+ is harder than Cu 2+ ions according to the Pearson acid-base concept 35 and therefore the coordinated bond with carboxylic ligands is stronger for the trivalent cation than for the divalent cation. However, the Cu 2+ paddlewheel increases its stability when the N-containing ligand coordinates to one of the Cu 2+ ions. 36 (ii) m-OH groups. This particular hydroxo-functionalisation in MOF materials is well exemplied by the MFM-300(M) (MFM ¼ Manchester Framework Material, M ¼ Al 3+ , In 3+ , Ga 3+ and Sc 3+ ) family. These outstanding MOF materials are chemically stable towards SO 2 and have been reported by Schröder and Yang. 37 The general chemical formula of this group of MOFs is [M 2 (OH) 2 (L)] (H 4 L ¼ biphenyl-3,3 0 ,5,5 0 -tetracarboxylic acid ¼ C 16 2 ] octahedra connected via the cis-m-OH groups into innite chains, and further coordinated by the tetradentate ligand (L 4À ). This particular organisation produces highly porous materials (see Table 2), well dened by one-dimensional pore channels organized in a "wine rack" display.
The rst MFM-300 material investigated for the capture of SO 2 was MFM-300(Al). 37a Although the SO 2 capture was not remarkably high (see Table 2), the identication of the preferential SO 2 adsorption sites was achieved by sophisticated in situ INS and PXRD experiments, revealing that the m-OH groups bind SO 2 molecules through the formation of O]S]O(d À )/ H(d + )-O hydrogen bonds, reinforced by weak supramolecular interactions with C-H atoms from the aromatic rings of the framework (Fig. 2). 37a In the case of MFM-300(In) (also known as InOF-1 (ref. 38)) the SO 2 uptake was slightly higher than for MFM-300(Al) (see Table 2), and in situ INS and PXRD experiments corroborated the same preferential adsorption site for SO 2 (see Fig. 3) with exceptional chemical stability for SO 2 capture under both dry and humid conditions. 37b In fact, Eddaoudi and Salama fabricated an advanced chemical capacitive sensor, using MFM-300(In), for the detection of very low concentrations of SO 2 (z5 ppb) at room temperature. 39 MFM-300(Sc) showed the highest SO 2 uptake of the MFM-300(M) materials (see Table 2), Grand Canonical Monte Carlo (GCMC) simulations demonstrated the same preferential adsorption sites (m-OH) and by conning small amounts of EtOH (2.6 wt%) the SO 2 capture increased by 40% (see Fig. 4). 37c The only MFM-300 remaining to be investigated in the capture of SO 2 is MFM-300(Ga), 37d where we anticipate, based on the pore volume, performance similar to MFM-300(Sc).
Thus, the MFM-300(M) MOF materials have shown the relevance of the hydroxo-functionalisation to the capture and detection of SO 2 with high chemical stability and excellent cyclability involving a remarkably facile regeneration. The key for all of such outstanding properties is the strength of the coordination bonds between the oxygen atoms form the This leads to a higher surface charge density for the Al 3+ metal centre and, therefore, the M-O bond strength decreases in this order: Al 3+ > Sc 3+ > In 3+ . Thus, a higher ligand exchange can occur in the In-based complex. 40b However, MFM-300(In) does not suffer apparent displacement of carboxylate ligands aer SO 2 exposure. In the MFM-300(M) family all the metal cations are nonmagnetic because of all paired electrons, 41 and the comparison between octahedral complexes formed by trivalent cations of group 13 such as Al 3+ , Ga 3+ and In 3+ , with the transition-metal analogue Sc 3+ , has shown that in the group 13 metal complexes the d orbitals are not involved in the metallinker bonding due to their high energy (3d 10 and 4d 10 conguration). In the case of Sc 3+ , a complex with 12e À , d orbitals form part of the M-O bonds (Fig. 5a). On the other hand, these types of complexes are considered as electron-rich hypervalent species with 7-center-12-electron bonding pattern (Fig. 5b). 40c This type of binding confers to the group 13 complexes thermal stability and in this particular case, the kinetic stability is comparable to that of the transition metal analogues. At this point, we can corroborate that frameworks constructed with trivalent cations are more stable than MOFs based on divalent cations (considering the stability of Irving-Williams series discussed above), as well that carboxylate ligands form with M 3+ cations stronger bonds than divalent cations with N-based ligands. 32 (iii) Defective sites. Crystal irregularities (composition inhomogeneities or defects) are fundamental characteristics of some solid-state materials and provide very particular physical and chemical properties. 42 Such defects do not necessarily mean adverse effects. Due to their extraordinary modularity and tunability, MOFs allow the introduction of different kinds of defects while maintaining the overall structure integrity. Typically, "defective MOFs" have showed extraordinary superior catalysis performances, 43 while Navarro and co-workers have taken a step forward and elegantly demonstrated the self-detoxication of defective UiO-66 examples to lter chemical-    warfare agents. 44 In the case of SO 2 , the same research group reported on the increase of the SO 2 adsorption capacities, energies and SO 2 /CO 2 selectivity for defective nickel pyrazolate MOF materials, prepared by introducing extra-framework Ba 2+ ions into the porous structure and ligand functionalisation. 45 Thus, the post-synthetic treatment of the [Ni 8 (OH) 4 (H 2 O) 2 (-BDP_X) 6 ] (H 2 BDP_X ¼ 1,4-bis(pyrazol-4-yl) benzene-4-X with X ¼ H (1), OH (2), NH 2 (3)) systems with ethanolic solutions of potassium hydroxide leads to the formation of defective K [Ni 8 (OH) 3 (EtO) 3 (BDP_X) 5.5 ] (1@KOH, 3@KOH) and K 3 [Ni 8 (-OH) 3 (EtO)(BDP_O) 5 ] (2@KOH). 46 Exposing the 1@KOH-3@KOH materials to aqueous Ba(NO 3 ) 2 solutions incorporates barium cations into the solution, yielding the defective ion exchanged Ba 0.5 [Ni 8 (OH) 3 (EtO) 3 (BDP_X) 5.5 ] (1@Ba(OH) 2 , X ¼ H; 3@Ba(OH) 2 , X ¼ NH 2 ), and Ba 1.5 [Ni 8 (OH) 3 (EtO)(BDP_O) 5 ] (2@Ba(OH) 2 ) systems. 45 Interestingly, density functional theory (DFT) calculations located the extra-framework cations close to the crystal defect sites (Fig. 6a). In addition, DFT results demonstrated that the average 3D structure of the MOF framework is preserved. 45 SO 2 dynamic adsorption experiments conrmed the benecial inuence of the deliberate introduction of defects by a subsequent K + to Ba 2+ ion exchange process on the SO 2 capture. 45 The authors demonstrated that the pre-synthetic introduction of amino and hydroxyl functional groups on the organic linkers and the post-synthetic modications, synergistically increase the SO 2 capture capacity of these materials. In addition, chemisorption of SO 2 was observed in all the systems: aer all available sites for SO 2 chemisorption were occupied, during the rst exposure, reversible adsorption takes place in a steady way evidencing the stability of the materials upon continuous SO 2 exposure. This high affinity of these MOF materials for SO 2 was attributed to the increased basicity of the nickel hydroxide clusters aer the introduction of defects. Lewis acid-base interactions with SO 2 molecules can afford the formation of HSO 3 À and SO 3 2À species according to eqn (1): [Ni 8 (OH) 6 Furthermore, the existence of extra-framework cations located close to defects sites promotes the formation of more stable MSO 3 (M ¼ Ba, 2 K) sulphite species according to eqn (2): These chemisorption mechanisms were supported by the DFT simulations (Fig. 6), showing that 1@Ba(OH)2 establishes stronger interactions with SO 2 molecules than 1@KOH due to the specic interactions with extra-framework barium cations.
Thus, this sophisticated and comprehensive study, by Navarro et al., 45 carefully incorporated essential variables to improve the SO 2 capture performance of these MOF materials. Such variables play together to increase the interaction with SO 2 and can be summarised as (i) enhanced basicity of metal hydroxide clusters as a result of additional hydroxide anions replacing the missing linkers defects; (ii) affinity of extraframework Ba 2+ ions for SO 2 sequestration; (iii) higher pore accessibility of the 3-D structure due to missing linker defects and (iv) the ne-tuning of the pore surface polarity by the benzene functional groups.
The post-synthetic inclusion of Ba 2+ cations to improve the SO 2 adsorption was later validated by Navarro and co-workers 47 SO 2 interactions. The amount of SO 2 adsorbed in 2 increased from 2.0 to 2.5 mmol g À1 in comparison to 1, respectively. This reversibility was conserved over 10 adsorption-desorption cycles. Theoretical calculations demonstrated that the principal binding site of SO 2 molecules is with coordinated water molecules, and it showed moderate adsorption energy of À65.5 kJ mol À1 , otherwise, the direct interaction of SO 2 with Ba 2+ cations would induce a higher adsorption energy (>100 kJ mol À1 ) due to the formation of an irreversible bond. Thus, this research group has elegantly proved that the formation of defective sites and the incorporation of large cations such as Ba 2+ is a good and feasible post-synthetic method for modulating the adsorptive capacities of MOF materials. (iv) Halogen functionalisation. Halogen functionalised materials are developing as promising technological platforms for innovative applications in different research elds such as catalysis, environmental remediation, sensing, and energy transformation. 48 On the other hand, cationic or anionic frameworks have shown potential applications in the capture of gases as the ions present inside the pores of the MOFs can be a benet for the interaction with guest molecules. 49 The presence of halogen counterions in porous materials has demonstrated an enhancement of the SO 2 capture performance due to the involvement of ionic bonding. 50 The following section is devoted to MOFs with halogen ions in their structure that participate in the adsorption of gaseous SO 2 .
Yang and Xing incorporated inorganic hexauorosilicate (SiF 6 2À , SIFSIX) anions (as pillars) into a series of MOF materials , and SIFSIX-3-Ni) and investigated the SO 2 adsorption properties of these materials. 51 The remarkable highly efficient removal of SO 2 from other gases, particularly at a very low SO 2 concentrations, and excellent SO 2 / CO 2 and SO 2 /N 2 selectivities were attributed to the strong electrostatic interactions between the SO 2 molecules and the SiF 6 51 The authors identied the interactions between the SO 2 molecule and SIFSIX materials by modelling studies using rst-principles DFT-D (dispersion-corrected density functional theory) calculations (see Fig. 7). 51 These computational calculations were experimentally corroborated by Rietveld renement of the powder X-ray diffraction patterns of SO 2 -loaded samples to locate the adsorbed positions of the SO 2 molecules in the crystal structure of SIFSIX materials. 51 Thus, this strategy highlighted the relevance on creating the multiple binding sites (anionic and aromatic linkers) which can originate the specic recognition of SO 2 to optimise its capture.
Similarly, Salama and Eddaoudi demonstrated how uorinated MOFs (KAUST-7 and KAUST-8) can be formidable candidates for sensing SO 2 from ue gas and air (250 ppm to 7% of SO 2 ). 52 SCXRD data collected on the SO 2 -loaded KAUST-7 and KAUST-8 identied the preferential adsorption sites for SO 2 revealing, comparable to SIFSIX materials, strong electrostatic interactions between the SO 2 molecules and the uorinated pillars (S d+ /F dÀ ), supported by dipole-dipole interactions with the pyrazine ligand (O dÀ /H d+ ), see Fig. 8. 52 These interactions were also corroborated by DFT calculations.
Another remarkable halogen functionalisation was presented by Janiak and co-workers, 53 where MOF-801 was modi-ed by reacting zirconium halides (ZrX 4 ; X ¼ Cl, Br, I) in water with acetylenedicarboxylic acid. The HX addition and material construction occurred in a one-pot reaction yielding three microporous HHU-2-X MOFs (X ¼ Cl, Br, I). The material HHU-2-Cl showed in increased SO 2 uptake (by 21%) in comparison to the nonhalogenated MOF-801. 53 Therefore, as previously demonstrated by the SIFSIX materials, KAUST-7, and KAUST-8, the electrostatic interactions between the SO 2 molecules and the Cl (S d+ /Cl dÀ ), might be responsible for such SO 2 capture enhancement.

Hydrogen sulphide
H 2 S is released to the environment via natural events such as volcanic eruptions, gas streams, hot springs, decomposition of organic matter and bacterial reduction. 54 Additionally, H 2 S is emitted by some chemical industries, e.g., oil desulphurisation processes at oil reneries, burning fossil fuels and mass transportation. 55 H 2 S is a main air pollutant which negative impacts the environmental as it is one of the main sources of acid rain, 55 and it is highly toxic to humans (concentrations above 700 ppm in the air can cause death). 56  H 2 S is a colourless, ammable gas with a characteristic rotten egg odour. Typical strategies to capture H 2 S comprise alkanolamines and ionic liquids, adsorption in solid materials (i.e., zeolites, activated carbons and metal oxides), membrane separation and cryogenic distillation. 57 However, these techniques have major drawbacks, such as low H 2 S capture, corrosion of pipelines and large cost of expenditure and recovery. 58 Thus, the development of new technologies for a dry adsorption process (avoiding solvents/water consumption is essential to reduce waste generation), for removal and capture of H 2 S signies a technological challenge which could allow not only the capture of H 2 S from the main source, but also its use as feedstock, similarly to the CO 2 technologies (CCS). 59 MOFs have been visualised for the capture of H 2 S; however, most of them have shown poor chemical stability in its presence. 60 Nonetheless, chemically stable MOF materials have demonstrated promising results in the reversible capture of H 2 S. 61 This section highlights outstanding MOF materials that have shown high H 2 S capture performances and in situ H 2 S transformations, as well as biomedical applications.

MOFs for H 2 S capture
Distinctive investigations reported by De Weireld, 62 Zou 63 and Eddaoudi 64 reveal that the majority of MOF materials experience structural decomposition upon adsorption of H 2 S or the desorption of it resulted complicated, due to relatively strong host-guest binding in the pores (strong physisorption or chemisorption). In such cases, the reactivation of the materials involves a large energy penalty or is not feasible. Thus, the identication of new MOFs capable of capturing H 2 S under industrially practical pressure-swing desorption conditions, 65 represents a very important challenge to solve. This perspective focused on (i) high and reversible H 2 S capture in MOFs and (ii) chemical transformation of H 2 S within the pores of MOFs: formation of polysulphides.
(i) High and reversible H 2 S capture in MOFs. The capture of H 2 S by MOFs has conrmed some crucial difficulties such as the formation of strong bonds, typically irreversible, (e.g., a metal-sulphur bond), which can compromise the chemical stability and cyclability of MOFs. Therefore, it is required to modulate host-guest interactions between the MOFs and H 2 S to avoid structure collapse and afford a feasible cyclability. In order to achieve these goals, such interactions should arise through noncovalent bonding between the functionalisation of MOFs and H 2 S guest molecules. For example, hydrogen bonding has been postulated, by density functional theory (DFT) methods and grand canonical Monte Carlo (GCMC) simulations, 66 as a promising moderate interaction between MOFs and H 2 S molecules. Experimentally, a remarkable work by Hamon and co-workers showed the high chemical stability and cyclability of MIL-47 (M ¼ V 4+ ) and MIL-53(M) (M ¼ Al 3+ , Cr 3+ ) when exposed to H 2 S. 67 These MOF materials are constructed with m-OH functional groups which presumably established hydrogen bonds with H 2 S.
Humphrey and Maurin showed the H 2 S capture on Mg-CUK-1. 68 This MOF is assembled from a 2,4-pyridinecarboxylate ligand and Mg 2+ octahedral centres, connected into innite chains of [Mg 3 (m 3 -OH)] 5+ clusters by m 3 -OH groups. Although the total H 2 S capture of Mg-CUK-1 was relatively low (3.1 mmol g À1 ), this material demonstrated to be chemically stable to H 2 S and a remarkably easy regeneration aer the H 2 S capture using temperature swing re-activation (TSR). Most importantly, GCMC simulations showed how H 2 S molecules interact with the hydroxo functional groups (by moderate hydrogen bonds m 3 -O-H/SH 2 ), see Fig. 9. 68 Such moderate H 2 S/Mg-CUK-1 interaction was found to be consistent with easy regeneration of the material aer H 2 S exposure.
This work demonstrated experimentally and in good correlation with advanced computational calculations, the chemical structure integrity of a hydroxo-functionalised MOF material to H 2 S, and the critical role of a moderate interaction (hydrogen bonding) to facilitate its cyclability. The study of more m-OH functionalised MOFs for the efficient and cyclable capture of H 2 S initiated.
Maurin and Gutiérrez-Alejandre investigated another m-OH functionalised MOF (MIL-53(Al)-TDC) for the capture of H 2 S. 69 This material, an Al 3+ -based constructed with a carboxylate ligand (TDC ¼ 2,5-thiophenedicarboxylate) which contains [AlO 4 -trans-(m-OH) 2 ] octahedra where the Al 3+ centre coordinates to two m-OH groups and six oxygen atoms from the TDC ligands, has established the highest H 2 S adsorption (18.1 mmol g À1 at room temperature) reported for a microporous material. 69 Structural stability of MIL-53(Al)-TDC, aer the H 2 S capture experiment was corroborated by PXRD and SEM analyses. TGA experiments demonstrated the complete desorption of H 2 S molecules at 65 C. H 2 S sorption-desorption cycles (ve cycles with a value of 18.5 AE 0.7 mmol g À1 ) demonstrated high H 2 S regeneration capacity. This cyclability conrmed the full regeneration of the material by only increasing the temperature to 65 C, showing the low energy requirement to fully desorb H 2 S. This suitable cyclability indicated weak interactions between H 2 S molecules and the pores of the material. In situ DRIFT experiments investigated the interactions between H 2 S and MIL-53(Al)-TDC, showing (i) the formation of hydrogen bonds between H 2 S molecules themselves conned in the pores of MIL-53(Al)-TDC; (ii) relatively weak interaction between H 2 S and the m-OH groups and (iii) H 2 S molecules interact with the other functionality of the MOF: the thiophene ring from the TDC ligand. 69 To corroborate these interpretations from the DRIFT experiments, Monte Carlo simulations were carried out for different loadings corresponding to the experimental discoveries. These calculations corroborated that at low loading, H 2 S interacts via its S-atom with the H-atom of the m-OH group (see Fig. 10), representing the preferential adsorption site. In addition, it was shown that H 2 S also interacts with the thiophene ligand, and, upon increasing the H 2 S loading, hydrogen bonds between the H 2 S molecules were also identied ( Fig. 10). 69 These theoretical ndings revealed the fundamental role of the thiophene ligand. Experimentally speaking, this M-OH hydroxo group containing functional material shows an extraordinarily facile re-activation leading to an easy H 2 S cyclability, which should not be anticipated due to the dominant interaction between the m-OH group and H 2 S reenforced further by the interaction of the H 2 S molecule with the thiophene unit as demonstrated by in situ DRIFT measurements. However, computational calculations postulate that the interaction of H 2 S with the thiophene ring weakens the interaction with the preferential adsorption site (m-OH group), favouring the easy displacement of the H 2 S molecule when the material is re-activated.
As previously discussed, (vide supra) for the MFM-300(M) family, the extraordinary chemical stability arises from the strength of the coordination bonds between the oxygen atoms form the carboxylic ligand and the M 3+ metal centres. Thus, MIL-53(Al)-TDC is very similar to MFM-300(Al) since it is also constructed with Al 3+ . Therefore, the chemical stability of MIL-53(Al)-TDC towards H 2 S is envisaged based on the same arguments used for MFM-300(Al).
(ii) Chemical transformation of H 2 S within the pores of MOFs: formation of polysulphides. As described in the previous section, the study of more hydroxo-functionalised MOF materials attracted high interest since the promising results on the effective and efficient cyclable capture of H 2 S. Thus, a couple of materials from the MFM-300(M) family were investigated by Maurin and Gutiérrez-Alejandre, 70 for the capture of H 2 S: MFM-300(Sc) and MFM-300(In).
First, MFM-300(Sc) exhibited a H 2 S uptake of 16.5 mmol g À1 (at 25 C), which is comparable to the H 2 S uptake of MIL-53(Al)-TDC (18.1 mmol g À1 ). 69 Upon an inspection of the structural integrity of the material aer the H 2 S experiment, PXRD experiments revealed the retention of the crystalline structure. Interestingly, when investigating the porosity of MFM-300(Sc) aer the H 2 S uptake experiment, the pore volume was reduced by 34% from 0.56 to 0.37 cm 3 g À1 indicating that some remaining species are still present in the pores of the material. Aer conrming the retention of the crystallinity and a reduction of the intrinsic porosity of MFM-300(Sc) aer the rst H 2 S cycle, additional H 2 S cycling experiments were performed. Over the second cycle, the H 2 S adsorption capacity decreased by 39% to 10.08 mmol g À1 , which is in a good agreement with the reduction of the pore volume by about 34%. The average H 2 S capture in cycles 2 to 5 was 10.22 mmol g À1 . Aer the h cycle, PXRD experiments corroborated the retention of the crystalline structure of MFM-300(Sc), while a N 2 adsorption experiment showed that the pore retained its reduced volume of 0.37 cm 3 g À1 . In an attempt to remove the remaining sulphur species formed inside the pores of MFM-300(Sc), the sample aer the rst H 2 S cycle was activated at 250 C. Aer this thermal treatment, the pore volume did not change (0.37 cm 3 g À1 ) conrming that the sulphur species were irreversibly adsorbed within the MFM-300(Sc) framework. 70 The identication of these sulphur species was rst approached by TGA and diffuse-reectance infrared Fouriertransform spectroscopy (DRIFT) experiments which corroborated strong interactions between the guest species and the pore-walls of MFM-300(Sc). Raman spectroscopy, complemented with elemental analysis (EDX) and conventional elemental analysis, postulated the rst indication of the possible nature of the remaining (irreversibly adsorbed) sulphur species within the pores, i.e., polysulphides (see Fig. 11). 70 Even though the redox properties of polysulphides are highly complex, electrochemical experiments were the key to fully identify them. An electrochemical cell (MFM-300(Sc))-CSP/ PVDF/1 M LiTFSI in triglyme/Li0 3 Me) was examined (for 20 h) through open-circuit potential (OCP) measurements and analysed by cyclic voltammetry. Then, this electrochemical cell showed an initial potential of 2.29 V (vs. Li 0 /Li + ) for MFM-300(Sc) and aer reaching equilibrium, this potential was equal to 2.15 V (vs. Li 0 /Li + ).
Later, a H 2 S exposed sample of MFM-300(Sc) (H 2 S@MFM-300(Sc)) was used to construct a different electrochemical cell showing a different behaviour on the variation of the potentials (the potential increased from 1.90 to 2.33 V (vs. Li 0 /Li + )). According to Mikhaylik and Akridge, 71 electrochemical potentials lower than 2.10 V correspond to low-order polysulphides (S n 2À ), (n ¼ 2). Thus, the initial potential for H 2 S@MFM-300(Sc) of 1.90 V suggested the formation of such low-order S 2 2À polysulphides. The formation of these low-order species arises from the strong mutual H 2 S/H 2 S hydrogen bond interactions with a characteristic H(H 2 S)/S(H 2 S) distance of 2.91Å, as demonstrated in the corresponding RDF plot from the Monte Carlo simulations. In the case of MFM-300(In) by constructing the equivalent electrochemical cell (MFM-300(In))-CSP/PVDF/1 M LiTFSI in triglyme/Li 0 and same operation conditions (OCP measurements and analysed by cyclic voltammetry for 20 h), the electrochemical cell exhibited an initial potential of 2.65 V (vs. Li 0 / Li + ) for MFM-300(In) and aer reaching equilibrium, this potential was equal to 2.15 V (vs. Li 0 /Li + ). The difference of 0.36 V, from the initial potentials for MFM-300(Sc) and MFM-300(In), was attributed to the difference in the electronegativity of the metal centres (Sc 3+ and In 3+ ) in both materials. Aer the adsorption of H 2 S, H 2 S@MFM-300(In), MFM-300(In) showed a different trend than MFM-300(Sc): the potential decreased from 2.50 to 2.15 V (vs. Li 0 /Li + ). Such a trend difference for both stabilisation potentials and the potential values indicated that the chemical composition of the sulphur species in both materials (MFM-300(Sc) and MFM-300(In)) were different. Again, Mikhaylik and Akridge indicate, 71 that electrochemical potentials higher than 2.10 V correspond to highorder polysulphides (S n 2À ), (n ¼ 6). Such large polysulphides block completely the pores of MFM-300(In), as corroborated by the loss of the pore volume in H 2 S@MFM-300(In) (0.02 cm 3 g À1 ) and Raman spectroscopy. Interestingly, although these two materials are isostructural, a small difference in the pore size (8.1 and 7.5Å for MFM-300(Sc) and MFM-300(In), respectively) contributes along with the different electrochemical potential of both materials (different electron densities) to the formation of distinct polysulphide species.
Although there is a hypothesis that justies the polysulphide formation in both materials (based on the "disproportionation" type reaction, 72 were the protons of H 2 S play the role of an oxidant and the sulphide plays the role of a reducing agent), there is a great opportunity to deeply investigate this phenomenon. For example, can the polysulphide formation be a consequence of reversible metal-ligand bonding upon the adsorption of H 2 S? Recently Brozek and co-workers 73 elegantly demonstrated metal-ligand dynamics for carboxylate benchmark MOFs, via variable-temperature diffuse reectance infrared Fourier transform spectroscopy (VT-DRIFTS) coupled with ab initio plane wave density functional theory. Thus, new exciting horizons can be discovered by taking a different approach when H 2 S is adsorbed in MFM-300(M), which could explain such fascinating polysulphide formation.
The last MOF material, to date, that demonstrated the chemical transformation of H 2 S to polysulphides is SU-101. 74 This bioinspired material was synthesised (by Inge et al. 74 ), using ellagic acid, a common natural antioxidant, and bismuth (Bi 2 O(H 2 O) 2 (C 14 H 2 O 8 )$nH 2 O), see Fig. 12. Then, the resultant breakthrough H 2 S experiment led to a gas uptake of 15.95 mmol g À1 , representing one of the highest H 2 S uptakes reported and even more interesting since the BET surface area of SU-101 (412 m 2 g À1 ) is considerably lower than top H 2 S capture materials (e.g., MIL-53(Al)-TDC; BET ¼ 1150 m 2 g À1 ). On a second H 2 S adsorption cycle, the initially observed capacity and surface area were reduced to only 0.2 mmol g À1 and 15 m 2 g À1 , respectively, even though the crystallinity was retained as conrmed by PXRD. Raman spectra of SU-101 (before and aer H 2 S adsorption) conrmed the presence of low-order polysulphides (n ¼ 2), S 4

2À
. Thus, the electrochemical potential of SU-101 should be similar to the one for MFM-300(Sc) (2.29 V), 70 considering that S 4 2À species were only found for MFM-300(Sc). Future investigations are anticipated to continue in order to verify the electrochemical potential of SU-101 and any possible reversible metal-ligand bonding upon the adsorption of H 2 S. A variety of MOFs, as well as classic materials such as zeolites, activated carbon and metal oxides, have been studied for the adsorption of these toxic gases, where some of these are summarized in Table 3. Noteworthy, the comparison of materials for H 2 S capture is sometimes challenging due to the difference in the experimental conditions (e.g., ue gas concentration).

Bio-compatible MOFs for H 2 S detection and controlled delivery
Although H 2 S is catalogued as highly toxic, paradoxically, it can also be crucial as an endogenous biological mediator. 82  Endogenous pathways for H 2 S have been found to have wide extending activities in vivo, such as in the cardiovascular system. 83 For example, H 2 S executes a crucial activity in the regulation of blood pressure, neurotransmission, anti-inammatory mechanism, anti-oxidation, angiogenesis and apoptosis. 84 H 2 S can be biosynthesised by enzymatic reactions via a sequence of endogenous processes, 85 and in the biological medium, the H 2 S concentration can uctuate from nano-to micromolar levels. 86 However, at higher concentrations of H 2 S in the bloodstream can cause severe physiological disorders such as diabetes, Alzheimer's disease, cirrhosis, different types of cancer and Down's syndrome. 87 Therefore, in order to recognise the specic role of H 2 S in these processes, the detection (spatial and temporal information) of H 2 S levels in living cells and organisms is crucial. However, due to the high reactivity, volatility and diffusible properties of H 2 S, traditional detection technologies (chromatography, high performance liquid chromatography, electrochemical analysis and colorimetric) 88 are unsuitable for the identication of H 2 S in living cells because they result in destruction of the cells and tissues and difficulties associated to real-time detection. 89 Fluorometric detection techniques have emerged as highly signicant alternatives since their nondestructive characteristics, fast response, high selectivity and sensitivity, fast response, high spatial sampling capability, realtime monitoring and easy sample preparation. 90 Turn-on type uorescence probes for H 2 S detection have been recently investigated and these are commonly based on the reactivity characteristics of H 2 S (e.g., copper sulphide precipitation, dual nucleophilic addition, H 2 S-mediated hydroxyl amide and nitro/ azide reduction). 91 Such characteristic reactivity of H 2 S can efficiently distinguish it from other biological species; however, many of these detectors do not full basic requirements such as rapid detection and selectivity. Thus, the investigation of new H 2 S probes is still very attractive due of the complexity of different molecular events involved in signalling transduction, and other complications related to the probes. Therefore, biocompatible MOFs have been postulated as promising probes for the detection of H 2 S.
The rst concept of a MOF material in in vitro H 2 S delivery was demonstrated by Morris and co-workers who used CPO-27 for this purpose. 92 H 2 S can be released by exposing the material to H 2 O since CPO-27 is an open metal site system thus, the water molecule replaces the hydrogen sulphide molecule at the metal site. The authors analysed the H 2 S delivery in two isostructural materials (Zn-CPO-27 and Ni-CPO-27) nding that the Ni 2+ material exhibited higher delivery capacity than the Zn 2+ material: aer 30 min 1.8 mmol g À1 of H 2 S was released from Ni-CPO and only 0.5 mmol g À1 from the Zn-CPO. Aer 1 h, the H 2 S release for both materials was completed. The amount of H 2 S released by each material corresponds to approximately 30% of the chemisorbed gas. Later, the release of H 2 S (stored in Zn-CPO-27 and Ni-CPO-27) was investigated under physiological conditions (endothelium-intact ring of pig coronary artery), nding a substantial vasodilatory action (relaxation) with only a short induction period of approximately 5 min. 92 These results represented a very signicant progress in the eld of MOFs and biological applications, since such H 2 S release mechanisms offered new protocols to better understand the effects of H 2 S on this particular vascular system.
Later, the rst uorescence probe example based on a MOF for the detection of H 2 S was presented by Wang and co- workers. 93 Through a post-synthetic modication of ZIF-90 with malononitrile (N^C-CH 2 -C^N), they obtained MN-ZIF-90 which undergoes a specic reaction with H 2 S, achieving an enhancement of photoluminescence, constituting the base for the detection of H 2 S (see Fig. 13). Then, the malononitrile moieties conjugated to the host (ZIF-90) through double bonds aided as quenchers of the host uorescence through intramolecular photoinduced electron transfer. The detection mechanism was based on the a,b-unsaturated bond of the malononitrile which is susceptible to thiol compounds, leading to the breaking of the double bond and thus, the uorescence of MN-ZIF-90 was recovered once H 2 S was introduced into the system. In addition, the probe exhibited favourable biocompatibility. 93 Taking a different approach but keeping the fundamental idea of introducing a reactive site for H 2 S, Tang et al., 94 demonstrated the use of a porphyrin-based MOF for the uorescent detection of H 2 S. By introducing reactive Cu 2+ metal centres into a Al 3+ MOF material {CuL[AlOH] 2 } n (H 6 L ¼ mesotetrakis(4-carboxylphenyl) porphyrin) the detection of H 2 S was selectively followed by uorescence under physiological pH (see Fig. 14). Additionally, they successfully demonstrated the capability of the probe to detect exogenous H 2 S in living cells, while the probe showed low toxicity and high biocompatibility. 94 Later, Biswas and co-workers demonstrated, on a dinitrofunctionalised UiO-66 (UiO-66-(NO 2 ) 2 ), uorescence turn-on behaviour towards H 2 S in simulated biological medium (HEPES buffer, pH ¼ 7.4). 95 Remarkably, UiO-66-(NO 2 ) 2 exhibited highly sensitive uorometric H 2 S sensing while also showing a visually detectable colorimetric change to H 2 S in daylight (see Fig. 15). In addition, the high selectivity of this functionalised MOF material to H 2 S was preserved even when several other biological species were present in the detecting medium. Finally, uorescence microscopy studies on J774A.1 cells demonstrated the effectiveness of UiO-66-(NO 2 ) 2 for H 2 S imaging in living cells, detection of H 2 S in human blood plasma (HBP) and monitoring of the sulphide concentration in real water samples. 95 These results emphasised the biocompatibility of UiO-66-(NO 2 ) 2 . The reaction mechanism of UiO-66-(NO 2 ) 2 and H 2 S is caused by the reduction of the nitro groups to electron donating amine groups causing an increase in the uorescence intensity that can be registered.
Eu 3+ /Ag + @UiO-66-(COOH) 2 (EAUC) composites were reported by Li and Qian as biomarkers for the for potential diagnosis of asthma. 96 The decrease of the H 2 S production in the lung has been identied as an early detection biomarker for asthma. Thus, uorescent experiments showed that EAUC exhibited high selectivity and sensitivity with a limit of a real-time H 2 S detection of 23.53 mM. The detection of H 2 S by EAUC was based on the introduction of active metal centre (Ag + , H 2 S-responding site) to the Eu 3+ @UiO-66-(COOH) 2 as the lanthanide-luminescence sensitiser. 96 MTT assay and cell viability analysis in PC12 cells demonstrated relatively low cytotoxicity for EAUC and biocompatibility. Finally, the determination of H 2 S was tested in diluted fetal bovine and human serum samples demonstrating that EAUC can detect H 2 S in real biological samples. Zhang and coworkers 97 covalently modied (via click chemistry) PCN-58 with target-responsive two-photon (TP) organic moieties to uorescently detect H 2 S. These TP-MOF probes showed good photostability, high selectivity, minor cytotoxicity, and excellent H 2 S sensing performance in live cells. These modied PCN-58 materials also showed intracellular sensing and depth imaging capabilities (penetration depths up to 130 mm). 97   Up to this point we have presented remarkable examples of MOF materials for the detection of H 2 S (biomarkers). All of these have been constructed with the fundamental principle of incorporating into these MOFs "the right chemical functionality" which can react with H 2 S. Remarkably, Wang and Xie 98 took a step forward and very recently showed how an endogenous MOF-biomarker-triggered "turn-on" strategy was capable to produce therapeutic agents in situ, as a promising example in nanomedicine for the precise treatment of colon cancer. HKUST-1 was used as an endogenous H 2 S-activated copper MOF which demonstrated to synergistically mediate H 2 S-activated nearinfrared photothermal therapy and chemodynamic therapy. As a proof of principle, the protocol worked as follows: in normal tissues, the photoactivity of as-synthesised HKUST-1 nanoparticles was in the "OFF" state, and no obvious adsorption within the NIR region was observed. On the other hand, when HKUST-1 nanoparticles with high overexpression of H 2 S reached the colon tumour tissues, these nanoparticles were activated to the "ON" state by reacting with endogenous H 2 S to produce in situ photoactive copper sulphide with stronger NIR absorption, which is viable for the subsequent photothermal therapy and corresponding thermal imaging (see Fig. 16). In addition, besides the in situ formation of the photothermal agent, HKUST-1 nanoparticles also showed a horseradish peroxidase (HRP)-mimicking activity to efficiently convert overexpressed H 2 O 2 within cancer cells into more toxic cOH radicals for chemodynamic therapy. Thus, this promising H 2 S-triggered "turn-on" strategy based on the endogenous biomarker from the tumour microenvironment can provide a precise diagnosis, marginal invasion, and possibly clinical translation. 98

Ammonia
NH 3 is one of the most important chemicals in the world since it is an irreplaceable feedstock for global agriculture and industry (e.g., fertilizers, drug production, heat pumps and fuel cells). 99 Although fundamental to our diet supply and economy, NH 3 is a highly toxic (even in small concentrations) and corrosive gas (i.e., difficult to handle and store). Thus, considerably large emissions of NH 3 from livestock breeding, industrial production, fertiliser application and public transportation can have serious consequences on the environment (highly unfavourable to air quality and aquatic life), and human health. 100 In addition, NH 3 reacts with NO x and SO 2 to form PM 2.5 , seriously contributing to the increase of air pollution. NH 3 is a colourless gas with a characteristic pungent smell which is typically liqueed and conned in metal tanks in order to be efficiently stored and transported. 101 This implies high pressures (approximately 18 bar) and constant corrosion of pipelines and containers. Thus, new sorbent technologies capable of efficiently removing or storing NH 3 are highly desirable for air remediation and separation of NH 3 from N 2 and H 2 since NH 3 is considered an interesting energy intermediate. 102 It is clear that capture and recycling of NH 3 implies a dual connotation in the elds of environment and energy. Classic materials such as activated carbons, zeolites, silicas and even organic polymers have been investigated for the capture and separation of NH 3 showing low uptakes, poor selectivity and, in some cases, irreversible storages. 103,104 MOFs have been postulated as a promising option for NH 3 capture, due to the access to a wide range of chemical functionality (e.g., Lewis or Brønsted acid sites which provide higher affinity to the basic NH 3 molecule). 104 However, the main problem that most MOF materials exhibit poor chemical stability towards NH 3 . 105 The following section describes chemically stable MOFs with interesting NH 3 capture performances.

MOFs for NH 3 capture
The NH 3 molecule is a Lewis and Brønsted base which is the key to its adsorption on the porous surface of chemically stable MOFs. Thus, the incorporation of acidic functional groups within the MOFs is required to achieve relevant NH 3 capture results. These functionalisations can be organised in three main types: (i) coordinatively unsaturated metal sites (open metal sites), (ii) m-OH groups and (iii) defective sites.
(i) Coordinatively unsaturated metal sites. The majority of MOF materials evaluated for the capture of NH 3 have demonstrated to be chemically unstable. 106 However, really interesting examples have been reported by Dincȃ and co-workers. 107 They reported a series of new mesoporous MOFs constructed from extended bisbenzenetriazolate ligands and coordinatively unsaturated metal sites (Mn 2+ , Co 2+ , and Ni 2+ ), which showed high and reversible NH 3 uptakes (15.47, 12.00, and 12.02 mmol g À1 , respectively). 107 During desorption at different temperatures, all three materials showed pronounced hysteresis which emphasised the strong interaction between the open metal sites and NH 3 . Such bound NH 3 molecules can be fully removed upon heating the materials up to 200 C under dynamic vacuum. Interestingly, none of the three materials showed a decrease in the NH 3 uptake capacity upon cycling. Thus, the high chemical stability of azolate MOFs with open metal sites, towards NH 3 postulate them as promising alternatives in environmental applications such as the capture of corrosive gases from power plant ue gas streams. This chemical stability can be attributed to the strength of the N atom coordinated to divalent metal centres as described previously. 21 In this case, the use of linear bistriazolate ligand, in combination with late transition metals such as Ni 2+ , increases the heterolytic metalligand bond strength due to its higher basicity (compared to a carboxylate ligand) which allows better MOF stability to acidic gas exposure. This trend is also explained by Pearson's acid and base principle due to Mn 2+ , Cu 2+ and Ni 2+ are considered so cations due to their charge/radius ratio, and nitrogen-based linkers are soer than carboxylate linkers. 35 Later, the Dincȃ's group reported another series of microporous triazolate MOFs (containing open metal sites; Cu 2+ , Co 2+ , and Ni 2+ , see Fig. 17), that exhibited remarkable static and dynamic NH 3 capacities (up to 19.79 mmol g À1 , at 1 bar and 298 K). 108 These isoreticular analogues to the bisbenzenetriazolate examples, 107 are constructed from smaller triazolate ligands and therefore, they exhibited smaller pore windows (see Fig. 17). Interestingly, these microporous MOFs showed higher NH 3 captures than their mesoporous analogues. 107 This remarkable property was not only accounted by the increase in the density of open metal sites, in addition cooperative proximity effects resulted very important. NH 3 breakthrough experiments on these microporous materials showed the potential applicability for both personal protection and gas separations (NH 3 capacities of 8.56 mmol g À1 ). Finally, once again, the superior chemical stability of these triazolate MOFs arises, as previously described, from the coordination of the N atom to metal centres (vide supra).
Very recently, Hong and co-workers recently presented the NH 3 adsorption properties of M 2 (dobpdc) MOFs (M ¼ Mg 2+ , Mn 2+ , Co 2+ , Ni 2+ , and Zn 2+ ; dobpdc 4À ¼ 4,4-dioxidobiphenyl-3,3-dicarboxylate), which aer activation exhibited open metal sites (see Fig. 18). 109 The NH 3 uptake of Mg 2 (dobpdc) at 298 K and 1 bar was 23.9 mmol g À1 at 1 bar and 8.25 mmol g À1 at 570 ppm, representing the NH 3 record high capacities at both pressures among existing porous adsorbents. 109   constant over ve breakthrough wet cycles. 109 Thus, Mg 2 (dobpdc) is an exceptional MOF material that can be easily synthesised (i.e., microwave-assisted), highly recyclable (even aer exposure to a humid NH 3 ) without structural degradation or NH 3 capacity loss and holds the record for the highest NH 3 uptake at room temperature and atmospheric pressure. All of these relevant properties are possible due to the exceptional chemical stability of this Mg 2+ -based MOF material. But why is it so stable towards NH 3 ? The authors proposed that such chemical stability was due to the higher affinity of Mg 2+ to oxygen atoms than nitrogen atoms, as conrmed by van der Waals (vdW)-corrected density functional theory (DFT) calculations. They further investigated the origin of such remarkable chemical stability nding that the oxygen adjacent to the carboxylate participates in the coordination of the ligand to the Mg 2+ cation acting as a tetratopic linker, similar to a chelate effect, increasing the stability of the cluster. This strategy is widely used to prevent pore collapse due to the activation energy barrier to ligand reorganisation or removal, which is increased by the higher linker connectivity. 21 (ii) m-OH groups. The MFM-300(M) family (see the section of SO 2 ) demonstrated remarkable NH 3 adsorption properties with high uptakes and attractive cyclabilities. First, MFM-300(Al), reported by Yang and Schröder, showed a total NH 3 capture of 15.7 mmol g À1 at 273 K and 1.0 bar. 110 Although the NH 3 capture was not particularly high (even at a lower temperature than typically 298 K), the identication of the preferential NH 3 adsorption sites was achieved by sophisticated in situ neutron powder diffraction (NPD) and synchrotron FTIR microspectroscopy. Thus, a structural analysis via Rietveld renement of NPD data for 1.5 ND 3 (deuterated ammonia)/Al-loaded MFM-300(Al) identied three distinct binding sites (I, II and III) in [Al 2 (OH) 2 (L)](ND 3 ) 3 , (H 4 L ¼ 1,1 0 -biphenyl-3,3 0 ,5,5 0 -tetracarboxylic acid ¼ C 16 O 8 H 6 ) (see Fig. 19). This cooperative network of ND 3 molecules spreads down the length of the 1D channel, xed in place by site I. Bond distances for sites II/III and a slightly lengthened site I/II are similar to a characteristic intermolecular bond between ND 3 molecules in the solid state at very low temperature (i.e., 2 K (N/D ¼ 2.357(2)Å)), while the bond between the framework m 2 -OH and site I is considerably shorter. Upon increasing the ND 3 loading (from 0.5 ND 3 /Al to 1.5 ND 3 /Al), a general shortening of the framework m 2 -OH/site I and sites I/II and an increase in the site II/III was observed.
Renement of the NPD data for ND 3 -loaded MFM-300(Al) demonstrated that on increasing the loading and thus the ND 3 /m 2 -OH ratio, the hydrogen atoms of the hydroxo functional groups experienced a reversible site exchange with the deuterium from the guest ND 3 molecules residing at site I in the pore to resulting in the formation of m 2 -OD moieties. 110 Interestingly, this H-D reversible exchange did not lead to any detectable structural degradation of the long-range order of the MOF material. Thus, the adsorption of ND 3  In the case of MFM-300(In), a signicant loss of NH 3 capture capacity over repeated NH 3 cycles was shown which corroborated its chemical instability. On the other hand, MFM-300(V 4+ ) exhibited the highest NH 3 uptake (17.3 mmol g À1 ) among the MFM-300(M) family. Interestingly, the NH 3 desorption phase for MFM-300(V 4+ ), with pore dimensions of approximately 6.7 Â 6.7Å 2 , showed a hysteresis loop. Although this could indicate a characteristic capillary NH 3 condensation (e.g., in mesopores and/or due to a broad distribution of pore size and shape), the authors demonstrated, taking into account the pore dimensions of this MOF, a specic and potentially strong host-guest charge transfer upon the adsorption of NH 3 . In addition, MFM-300(V 4+ ) showed an increase of both NH 3 capacity and residue within the rst 18 cycles, which was not observed for the rest of the MFM-300(M) materials. Such residual amount of NH 3 le within MFM-300(V 4+ ) upon regeneration (pressure-swing) gradually increased from 8 to 20% along these cycles, indicating an increase of strongly bound NH-derived species in MFM-300(V 4+ ). This NH-derived residue, which was not desorbed by only reducing the pressure, was completely removed by increasing the temperature under dynamic vacuum, although some structural degradation of MFM-300(V 4+ ) was observed. The chemical stability of these materials to NH 3 under humid conditions was investigated by PXRD, conrming the retention of their crystallinity for both MFM-300(M) (M ¼ Cr 3+ , V 3+ ), while MFM-300(M) (M ¼ Fe 3+ , V 4+ ) showed some structural degradation. 111 Neutron powder diffraction data for ND 3 -loaded MFM-300(M) combined with Rietveld renements, showed the preferential binding sites for ND 3 (Fig. 20). Two binding sites for ND 3 were clearly identied for MFM-300(M) (M ¼ In 3+ , V 3+ ), while MFM-300(Fe) has an additional binding site for ND 3 . Site I showed the highest occupancy, with hydrogen bonding between the hydroxo functional group and the ND 3 molecule (m 2 -OH/ ND 3 ) (1.411-1.978Å), complemented by additional hydrogen bonding from the organic ligand (H aromatic /ND 3 ¼ 2.738-3.174 A; ND/O ligand ¼ 3.078-3.179Å) and electrostatic interactions (ND 3 /aromatic rings ¼ 2.946-3.132Å) (Fig. 20). Similar to MFM-300(Al), 110 hydrogen/deuterium site exchange was also observed between the adsorbed ND 3 and the m 2 -OH group on for MFM-300(M) (M ¼ In 3+ , Fe 3+ , V 3+ ). Site II is located toward the centre of the pore (xed by hydrogen-bonding interactions ND 3 /O ligand ¼ 2.284-3.065Å). Site III in MFM-300(Fe) is constructed by electrostatic interactions (ND 3 /aromatic rings ¼ 3.146Å). Additionally, in MFM-300(Fe), intermolecular hydrogen bonds between ND 3 molecules (2.327Å) were identi-ed, propagating along the 1D channel to form a cooperative {ND 3 } N network.
The preferential ND 3 binding sites for MFM-300(V 4+ ) were identied and these were different to the previous examples. MFM-300(V 4+ ) is not constructed with m 2 -OH groups. It shows bridging oxo centres to balance the charge of the oxidised V centre. Thus, ND 3 molecules are located in the centre of the pore. This particular arrangement implicates a very interesting situation: chemical reaction between the ND 3 molecules: a N 2 D 4 molecule was formed and located at site II with a ND 3 molecule at site I (being partially protonated to ND 4 + ). Then, both sites are stabilised via hydrogen bonding (ND 3 /O ligand ¼ 2.529-3.092Å), and the amount of N 2 D 4 at site II was estimated as 0.5 N 2 H 4 per V centre. The formation of N 2 H 4 was explained in terms of the redox activity of MFM-300(V 4+ ) which promotes a host-guest charge transfer between the material and the adsorbed NH 3 molecules, promoting the reduction of the V 4+ centres and oxidation of NH 3 to N 2 H 4 . 111 These host-guest charge transfer results were experimentally determined by EPR and corroborated by bond valence sum (BVS) calculations, nding as well that the charge transfer between adsorbed NH 3 molecules and the V 4+ centre can only occur when the loading of NH 3 is sufficiently high so that a predominant occupancy of the N site, which is located close to the metal chain, is reached to initiate the redox reaction. Finally, such mechanisms could also be investigated by the approach of metal-ligand dynamics as recently proposed by Brozek and co-workers, 73 to fully understand the interaction of NH 3 and the hydroxo functional group in the MFM-300(M) family.
(iii) Defective sites. As previously described in the SO 2 capture section (vide supra), crystal irregularities can provide very particular physical and chemical properties. Then, defects in MOF materials also demonstrated remarkable NH 3 adsorption properties. Very recently, Wu and Tsang presented a comprehensive study on the responsive adsorption behaviours of defect-rich Zrbased MOFs upon the progressive incorporation of NH 3 . 112 They investigated UiO-67 and UiO-bpydc containing 4,4 0 -biphenyl dicarboxylate and 4,4'-(2,2 0 -bipyridine) dicarboxylate ligands, that despite their structural similarities demonstrated a drastic difference in the NH 3 adsorption properties when the biphenyl groups in the organic ligand were replaced by the bipyridine moieties. Such replacement can confer exibility to the framework in the context of mainly ligand "ipping" but without signicant pore volume alteration.
Then, defective UiO-67 (non-monodisperse pore structure created by missing ligand defects) was synthesised thanks to the high connectivity of the Zr 6 oxoclusters, which help to retain the overall crystal structures even with some ligands are missing. UiO-67 involves uniform trigonal windows with a diameter of 11 A that lead to interconnected tetrahedral and octahedral pores inside the structures. In the presence of missing ligand defects, the trigonal windows surrounding the defects in UiO-bpydc are fused into lozenge windows with a dynamic size larger than 14Å. NH 3 adsorption-desorption isotherms at 298 K and up to 1 bar were performed for both materials: UiO-67 and UiO-bpydc. In the case of Ui-O-67, the adsorption isotherm showed "step-like shape" with two events (see Fig. 21, bottom). This characteristic isotherm shape is caused in MOFs by a gate-opening phenomenon due to the interaction between guest NH 3 molecules and pore walls, or a pore lling process. 113 Thus, at the beginning of the adsorption, the rapid and sudden increase in NH 3 adsorption at approximately 30 mbar (1.70 mmol g À1 ) (position I) suggests the presence of strong adsorption sites inside the framework, e.g., strong interaction of the NH 3 molecules with the hydroxo (m 3 -OH) functional groups. Later, the NH 3 uptake increased from 2.40 to 4.40 mmol g À1 at 250 mbar (position II) and from 5.60 to 8.40 mmol g À1 at 650 mbar (position III, Fig. 21). The desorption phase exhibited large, opened hysteresis, indicating that the NH 3 molecules strongly interact with the pore walls of the material. 112 These NH 3 molecules were fully desorbed by heating up to 423 K for 1 h under dynamic vacuum. Later, the chemical stability of UiO-67 towards NH 3 was investigated by conducting three NH 3 adsorption-desorption cycles and nally exposing the material to NH 3 vapour for 1 week. PXRD experiments corroborated the stability of UiO-67 to dry and humid NH 3 . Then, in order to identify the preferential NH 3 adsorption sites in UiO-67, in situ high-resolution neutron powder diffraction (NPD) and synchrotron powder X-ray diffraction (SPXRD) experiments were carried out. The rened structures of UiO-67 at different ND 3 pressures, reviled ve such adsorption preferential sites (Fig. 21). 112 At site I, the ND 3 was found close to the m 3 -OH with a OH/N I distance of 1.96(1)Å, and O/N I distance of 2.80(1)Å (see Fig. 21). This distance suggested the formation of a relatively strong Hbonding interaction between the m 3 -OH and ND 3 , similar to the one reported by Yang and Schröder in the MFM-300(M) family. 110,111 ND 3 molecules at site II were situated close to the walls of the trigonal windows suggesting interactions with the MOF organic ligands (H ligand /N II ¼ 2.58(1) and 2.68(1)Å), Fig. 21. On increasing the amount of ND 3 , these molecules were located near to the tetragonal pore (sites III and IV). At site III, the ND 3 molecule lled the shallow pore positions with N II /N III bond distances among these ND 3 sites in the range of 2.33(1) and 2.57(1)Å, creating a H-bonded network of these ND 3 molecules (Fig. 21).
In the case of UiO-bpydc, the NH 3 adsorption isotherm exhibited an uptake of 8.4 mmol g À1 at 298 K and 1 bar (see Fig. 22). Conversely to UiO-67, UiO-bpydc showed only one large and sharper transition step (position II 0 ) for the NH 3 adsorption phase (Fig. 22). Then, in situ NPD experiments and followed by Rietveld renements corroborated the existence of missing ligand defects. Similarly to UiO-67, the preferential NH 3 adsorption sites within UiO-bpydc were also identied by in situ NPD experiments (replacing NH 3 by ND 3 ). Thus, two independent preferential sites of ND 3 were found close to the m 3 -OH functional group (site I 0 ) and to the organic ligands (site II 0 , bipyridine ligand), Fig. 22. This strongly suggested that the adsorption of ND 3 mainly takes place at site II 0 . For site I 0 , the OH/N I bond distance was estimated to be 2.10(2)Å, which is similar to that found in the isostructural UiO-67. Interestingly, the ND 3 -bipyridine ligand interaction was estimated to be stronger than the ND 3 -biphenyl ligand interaction with longer distances (H ligand /N II ¼ 2.68(1), 3.19(1), and 3.20(1)Å, respectively) (see Fig. 22). Increasing the dosing of ND 3 molecules, increased the H-bonding between the ND 3 molecules giving rise to a network of six ND 3 molecules with the three bipyridine ligand located around the trigonal window. This particular situation demonstrated the gate-opening/closing behaviour via ligand ipping of UiO-bpydc upon ND 3 adsorption. 112 This phenomenon was further investigated by DFT calculations to understand the role of the bipyridine ligands of UiObpydc in forming the H-bonding network with NH 3  molecules, in comparison with the biphenyl ligands of UiO-67. Thus, the DFT-optimised structures demonstrated the progressive distortion of the aligned bipyridine ligands with g changed from 20.34 (in good agreement with the NPD rened desolvated structure of UiO-bpydc) to 19.47 , 10.83 , 9.49 , and 6.33 by increasing the number of NH3 molecules at the trigonal window. Finally, these remarkable results demonstrated that the different pore openings (windows) induced by missing ligands can introduce stepped NH 3 sorption with a strong hysteresis into the UiO type MOFs (i.e., biphenyl ligands are replaced by pyridine ligands). 112

NO x
Nitrogen oxides (NO x ¼ N 2 O, NO, N 2 O 3 , NO 2 N 2 O 4 and N 2 O 5 ) are considered one of the major air pollutants generated by anthropogenic activities, particularly those that involve fuel combustion from stationary and mobile sources such as thermal power plants and vehicles. 114 It has been estimated that fossil fuel combustion generates around 5% of NO 2 , and 95% NO. 114 Nevertheless, once NO is released into the atmosphere, it rapidly reacts with O 2 to form NO 2 . The latter is a poisonous redorange gas responsible for the reddish-brown colour of the smog, although at lower temperatures it can dimerize into a colourless N 2 O 4 dimer. Due to the highly reactive nature of NO x , the uncontrolled emissions of such gases into the atmosphere are associated with a series of health and environmental issues. [114][115][116] For instance, the presence of high concentrations of NO 2 has been directly related to the formation of tropospheric ozone O 3 , 109,116 which not only is considered an important greenhouse gas; but is also associated with pulmonary and chronic respiratory diseases. 115,116 The adverse effects caused by atmospheric NO 2 have motivated an intense research for the development of NO 2 abatement technologies. 114,117 One of the most studied methods to mitigate the atmospheric NO 2 is the development of porous materials for the selective capture of this pollutant. The traditional adsorbent materials used for NO 2 removal include zeolites, 118 calciumbased adsorbents, 119 and activated carbon. 120 However, such materials are typically affected by the reactive oxidative nature of NO 2 , which hampers the fully reversible desorption of the guest molecules and limits the material regeneration. 117

MOFs for NO x capture
Current investigations in NO x sorption have pointed out that MOFs represent an attractive alternative for storage, selective separation, and/or catalytic transformation of NO 2 . 121 Such materials not only display a large surface area, but they also exhibit tuneable pore functionalities 122 allowing for the stabilisation of guest molecules through the formation of supramolecular host-guest interactions. For instance, a study reported by Peterson et al. 123 points out the key role of the organic linker for the capture of NO 2 . In this work, the authors conducted a comparative study bout the NO 2 sorption performance of UiO-66 and UiO-66-NH 2 .
The gas sorption experiments were performed under dry and controlled humid conditions. The results obtained from the microbreakthrough experiments reveal that UiO-66-NH 2 exhibits a higher capacity for the NO 2 sorption than UiO-66 (20.3 mmol g À1 vs. 8.8 mmol g À1 ; respectively). Under humid conditions (80% RH), UiO-66-NH 2 exhibits higher uptake of NO 2 than under dry conditions (31.2 mmol g À1 vs. 20.3 mmol g À1 ; respectively), and it produces a signicantly lower amount of NO as a by-product than its analogous UiO-66 (4.5%, 9.5%; respectively).
Such differences were explained in terms of the higher capability of UiO-66-NH 2 to adsorb H 2 O vapours. As the H 2 O coadsorbed within the pore network might enhance the stabilization of NO 2 molecules through the formation of supramolecular interactions and it facilitates the preferential formation of nitrous acid as a by-product. The complementary characterization of UiO-66-NH 2 upon NO 2 adsorption revealed that although the crystallinity of UiO-66-NH 2 remains intact, the high reactivity of the adsorbate leads to a series of reactions with the organic linker, which ends up in the post functionalization of the phenyl group (Fig. 23). 123 The main transformation that suffered the organic linker upon NO 2 uptake were the nitration of the aromatic ring and the formation of the diazonium ion at the amino group, while in the inorganic SBU the terminal -OH group is replaced by the NO 3À ion. More recently, a study conducted by Schröder and Yang,124 demonstrated that the synergistic effect between the organic ligand and the inorganic building block in MFM-300(Al) allows for the stabilisation of highly reactive NO 2 species within the pore network (Fig. 24a-c). The authors reported that under ambient conditions MFM-300(Al) exhibits a fully reversible NO 2 isotherm uptake of 14.1 mmol g À1 at 298 K. Moreover, the host material retains its crystallinity and sorption capacity aer ve cycles of NO 2 adsorption/desorption. Remarkably, MFM-300(Al) displays outstanding performance for the selective removal of low-concentration of NO 2 (5000 to <1 ppm) from gas mixtures (Fig. 24d). The optimal uptake and selectivity of MFM-300(Al) for NO 2 was attributed to the existence of both host-guest and guest-guest interactions. The former involves mainly the hydrogen bonding interaction between NO x and the -OH pendant group of the inorganic node, while the latter refers to the supramolecular interactions between the guest molecules in their monomeric (NO 2 ) and dimeric (N 2 O 4 ) form. Such interactions give rise to a one-dimensional helical chain arrangement comprised of alternating monomer-dimer molecules (NO 2 $N 2 O 4 ) N running along the channel of MFM-300(Al) (Fig. 24d).
The cooperative supramolecular interactions between NO 2 and N 2 O 4 , conned within the pore network, allows for the stabilization of the highly reactive NO 2 molecules and inhibits the guest-host electron transfer. This affords an unusual, fully reversible desorption of NO 2 without altering the framework structure of the host material. The structural versatility of MOFs not only permits the stabilization of highly reactive NO x species within the pore channels but also opens the possibility for the catalytic conversion of NO x into more valuable and/or less abrasive compounds. In this regard, Schröder and Yang reported the adsorption and catalytic transformation of NO 2 by using a redox-active MOF, 125 termed MFM-300(V) [V 2 (OH) 2 (C 16 H 6 O 8 )]. This system not only exhibits high adsorption capacity for NO 2 (13 mmol g À1 at 298 K and 1.0 bar), but it also allows for the catalytic reduction of NO 2 into NO. The structural analysis reveals that upon gas sorption, NO 2 molecules get primarily anchored to the pore walls through a hydrogen-bonding interaction with the bridging hydroxyl groups of the inorganic nodes. Further stabilization is reached by forming of 8-fold supramolecular interactions with the aromatic ligand (Fig. 25). Then, those interactions give rise to the host-guest charge transfer process and the formation of NO (through the oxidation of the metal centre from V 3+ to V 4+ ), and water, which is produced by the deprotonation of the bridging -OH groups anchored to the SBU.
These ndings were explained in terms of the differences in the tetrahedral coordination environment around the metal centre N 3 M-Cl (M ¼ Zn 2+ , Cu 2+ ). In the rst case, the saturated coordination environment around Zn 2+ prevents the formation of stronger metal-NO interactions, whereas in the second system, the N 3 Cu-Cl geometry is sufficiently distorted to allow the NO molecules to approach closer to the metal centre leading to stronger Cu-NO interactions. The formation of Cu 2+ -nitrosyl species upon NO adsorption was corroborated by diffuse-reectance infrared Fourier-transform spectroscopy (DRIFT) study. Finally, the third system (Cu + -MFU-4l) was obtained by the reduction of Cu 2+ to Cu + accompanied by the concomitant loss of chloride. This system displays a signicant improvement in the low-pressure NO uptake (1.24 mmol g À1 below 1.4 torr). This value is three times the amount of NO adsorbed by Cu 2+ -MFU-4l (0.0008 mmol g À1 ) and Zn 2+ -MFU-4l (0.0002 mmol g À1 ) under the same conditions. The DRIFT analysis suggests that the metal centre in Cu 1+ -MFU-4l allows for the formation of Cu + -NO complex even when the material is exposed to low concentrations of NO in Ar (10 ppm). This interaction favours the NO disproportionation 3 NO / NO 2 + N 2 O and the oxidation of the metal centre to Cu 2+ which strongly binds to the NO 2 product (Fig. 26). Then, the Cu 2+ -nitrite bond can be cleaved by exposing Cu 2+ -MFU-4l-(NO 2 ) to high temperatures, releasing the anchored NO 2 and thereby regenerating the adsorbent Cu + -MFU-4l. According to the authors, this catalytic system represents a potentially attractive scheme for cold-start NO capture. The design of porous materials for storage and controlled release of NO have gained relevance in biomedicine, as it has been shown that the exposure to controlled doses of NO induces wound healing and prevents the formation of blood clots. Moreover, NO is a potent antimicrobial agent; therefore, the development of materials for the local release of NO could be highly advantageous to reduce the risk of infections.
One of the rst reports about the use of MOFs for the ondemand delivery of NO was published in 2007 by Morris and co-workers. 127 In this work, the authors used HKUST-1 as a reservoir for NO. The gravimetric adsorption experiments performed at 196 K and 298 K display signicant isotherm hysteresis with an adsorption capacity of ca. 9 mmol g À1 and 3 mmol g À1 at one bar, respectively. Upon gas desorption processes, both isotherms exhibit a remaining amount of NO trapped within the pore network (2 mmol g À1 ). The IR analysis of NO-loaded material reveals that the presence of open copper sites in the walls of HKUST-1 framework allows for the formation of NO/Cu 2+ coordination adduct; thereby leading to the irreversible adsorption of NO. However, NO molecules trapped within the pore network can be released on demand upon exposure of the NO-loaded HKUST-1 to water vapours, as H 2 O acts as a nucleophile replacing the NO molecules coordinated to the open copper sites.
More recently, the same research group reported the development of MOF-based composites for the controlled delivery of NO. 128 The selected host material was CPO-27-Ni, as this MOF affords highly efficient adsorption, storage, and release cyclable prole for NO. 129 Moreover, this material exhibits high stability towards adsorbed water, which allows for the controlled release of NO when exposed to moisture, without altering the structural arrangement of the host material.
The optimal MOF lms were obtained by varying the amount of CPO-27-Ni (wt%) integrated within a polyurethane matrix. The composite lms exhibited homogeneous distribution of the MOF within the polymer matrix. The release of NO was triggered by exposing the composite lms to 11% or relative humidity. Such conditions allow for the controlled release of the NO adsorbed without affecting the polymer properties. Finally, the antimicrobial properties of the resultant MOF-based composite demonstrated that the controlled release of biologically active levels of NO provides a bactericidal effect against Escherichia coli (E. coli) and Staphylococcus aureus (S. aureus).

Selectivity of trace gases
One important aspect of evaluating MOF materials is their selectivity of adsorbing one gas preferentially over other molecules that are present in the industrial ue gas streams, as these are mainly composed of CO 2 (10-15% [v] gases by MOFs should be accompanied by a selectivity study in the presence of other gases in order to have a complete prole of their potential applications at industrial levels. Table 4 lists the most promising MOF materials for the separation of SO 2 and H 2 S gases, with a good selectivity performance.

Stability and reusability of MOFs
As previously shown, there is a wide variety of MOFs with high structural stability towards corrosive gases (SO 2 , H 2 S, NH 3 and NO x ). The stability, as well as the reusability of the MOFs, is required for tangible industrial applications on the adsorption eld. Considering that the CSD currently contains near to 70 000 MOF structures, 143 a selection criterion for MOFs that could be potentially useful in the adsorption of toxic gases is essential, as are the guidelines for generating new materials focused on this target. 144 Although the nature of MOFs is very diverse, MOFs stable in the presence of corrosive gases feature some common characteristics: (i) the strength of the ligandmetal bond; (ii) oxidation state of the metal centre; (iii) robustness of the metallic cluster; and (iv) the formation of supramolecular host-guest interactions. The presence of at least one of these characteristics is responsible for the stabilization of the framework under such harsh conditions. For example, when the strongest binding site corresponds to a free metal site, the metal-ligand bond should be strong enough to avoid linker displacement and subsequent structure degradation. The combination of ligands with strong donor groups/ atoms and late transition metals, or the use of metals with a higher oxidation state together with carboxylate ligands are good synthetic strategies to improve the strength of the metalligand bonds and, therefore, increase the overall stability of the framework. 21,40a,145 For example, MOFs formed by triazolate ligands such as the M 2 Cl 2 BTDD family (M ¼ Ni 2+ , Co 2+ , Mn 2+ , and Cu 2+ ) 107 with uncoordinated metal sites are stable even aer electrostatic interactions between the SO 2 molecules and the SiF 6 2À anions (S d+ /F dÀ ), assisted by dipole-dipole interactions with the ligand (O dÀ /H d+ ). The identication of MOFs capable of capturing H 2 S under industrially practical pressure-swing desorption conditions, symbolises a very promising application which is based on the chemical stability of these materials. For example, stable MOF materials constructed with m-OH functional groups (e.g., MIL-53(Al)-TDC) exhibited high and fully reversible H 2 S capture where the formation of hydrogen bonds between H 2 S and the m-OH functional group (a relatively weak interaction) facilitates the H 2 S cyclability.
The chemical transformation of H 2 S within the pores of MOFs to produce, in situ, polysulphides is a new and exciting discovery in MFM-300(M) and SU-101 MOF materials. In addition to the promising application as novel MOF-lithium/ sulphur batteries, the chemical investigation behind them, provided addition information on the electrochemical potential of MOFs which could guide new interesting properties to be discovered. Then, although the polysulphide formation was explained in terms of a "disproportionation" type reaction (protons of H 2 S play the role of an oxidant and the sulphide plays the role of a reducing agent), other hypotheses need to be explored. What if the polysulphide formation is a consequence of reversible metal-ligand bonding upon the adsorption of H 2 S? We anticipate new perspectives will be shortly investigated.
Paradoxically, the role of H 2 S as an endogenous biological mediator in the human body is crucial. For example, H 2 S executes a vital activity in the regulation of blood pressure, neurotransmission, anti-inammatory mechanism, antioxidation, angiogenesis and apoptosis. Thus, the delivery and detection of H 2 S levels in living cells and organisms is an outstanding application for MOF materials. For example, Ni-CPO-27 showed a promising release of H 2 S (under physiological conditions) with a short induction period of approximately 5 min. In the eld of H 2 S detection, for example, a postsynthetic modication of ZIF-90 with malononitrile (N^C-CH 2 -C^N), exhibited a specic reaction with H 2 S completing an enhancement of photoluminescence, which constitutes the base for the detection of H 2 S. Thus, the development of this research is expanding by the fundamental principle of introducing a reactive site for H 2 S within MOF material (monitored by uorescence), which can work under physiological pH, to detect exogenous H 2 S in living cells, while the MOF-probe shows low toxicity and high biocompatibility.
Although most MOF materials have exhibited poor chemical stability towards NH 3 , chemically stable MOFs constructed with different chemical functionalities have shown interesting NH 3 capture performances.
Coordinatively unsaturated metal sites incorporated, for example, in bisbenzenetriazolate and triazolate MOFs, showed high and reversible NH 3 uptakes with strong interactions between the open metal sites and NH 3 . The chemical stability of these MOF materials was attributed to the strength of the N atom coordinated to divalent metal centres. Mg 2 (dobpdc) demonstrated high NH 3 capture and chemical stability to not only dry NH 3 , but also to wet NH 3 . Interestingly, it was experimentally and computationally demonstrated that NH 3 was preferentially adsorbed, by coordination bonds, to the coordinatively unsaturated metal sites of Mg 2 (dobpdc). The exceptional chemical stability of this Mg 2+ -based MOF material credited to the higher affinity of Mg 2+ to oxygen atoms than nitrogen atoms, as conrmed by van der Waals (vdW)-corrected density functional theory (DFT) calculations.
The m-OH functionalised MFM-300(M) family demonstrated remarkable NH 3 adsorption properties, high chemical stability, high uptakes and attractive cyclabilities. Not surprisingly, as in the case of SO 2 and H 2 S, the preferential adsorption sites of the MFM-300(M) materials were found at the hydroxo functional group. For example, in the case of MFM-300(Al) the adsorption of NH 3 showed an atypical adsorption mechanism where adsorbent and adsorbate experienced a rapid site-exchange via reversible formation and cleavage of O-H and O-H (from NH 3 ) chemical bonds, in other words a pseudo-chemisorption binding mechanism.
As previously described in the SO 2 capture section, crystal irregularities can provide very particular physical and chemical properties. Then, defects in MOF materials also demonstrated remarkable NH 3 adsorption properties. For example, UiO-67 and UiO-bpydc containing 4,4 0 -biphenyl dicarboxylate and 4,4 0 -(2,2 0 -bipyridine) dicarboxylate ligands, despite their structural similarities demonstrated a drastic difference in the NH 3 adsorption properties when the biphenyl groups in the organic ligand were replaced by the bipyridine moieties. Such replacement can confer exibility to the framework in the context of mainly ligand "ipping" but without signicant pore volume alteration.
Current investigations in NO x capture have pointed out that chemical stable MOFs represent an attractive alternative for storage, selective separation, and/or catalytic transformation of NO 2 . For example, the synergistic effect between the organic ligand and the inorganic building block in MFM-300(Al) showed the stabilisation of highly reactive NO 2 species within the micropores. The optimal uptake and selectivity of MFM-300(Al) for NO 2 was attributed to the existence of both host-guest and guest-guest interactions. The former involves mainly the hydrogen bonding interaction between NO x and the hydroxo functional group, while the latter refers to the supramolecular interactions between the guest molecules in their monomeric (NO 2 ) and dimeric (N 2 O 4 ) form.
Interestingly, the design of MOF materials for storage and controlled release of NO have gained relevance in biomedicine, as it has been shown that the exposure to controlled doses of NO induces wound healing and prevents the formation of blood clots. For example, CPO-27-Ni exhibited highly efficient adsorption, storage, and release cyclable prole for NO with a high stability to water, allowing a controlled release of NO when exposed to moisture.
Despite the signicant progress in the capture of NO x , SO 2 , NH 3 and H 2 S pollutants by MOF-based technology, further research in this eld is required to overcome some of the main challenges: chemical stability, reusability, and suitable functionalisation. Considering the large number of MOF materials synthesised to date (over 70 000 structures reported in the CCD), it is crucial to understand what makes the above-reported MOFs so chemically stable. Thus, this perspective summarises their most important characteristics (metal centre, strength of the metal-ligand bond and functional groups) that should be further investigated in order to be able to design tailor-made MOFs for the capture of corrosive gases. Additionally, an important aspect to consider is the study of breakthrough experiments and selectivity calculations, where more realistic conditions for industrial applications are investigated, providing a direct performance comparison of MOFs to classic materials such as activated carbon, zeolites and metal oxides. On the other hand, the synthesis scalability of MOFs is one of the weak points to contemplate for corrosive gas adsorption applications, as only a few materials can be synthesised on an industrial scale. We desire that this perspective can provide useful information on the signicant progress of this eld and inspire new investigations to be carried out.

Conflicts of interest
There are no conicts to declare.