Defect-Engineered Metal–Organic Frameworks

Defect engineering in metal–organic frameworks (MOFs) is an exciting concept for tailoring material properties, which opens up novel opportunities not only in sorption and catalysis, but also in controlling more challenging physical characteristics such as band gap as well as magnetic and electrical/conductive properties. It is challenging to structurally characterize the inherent or intentionally created defects of various types, and there have so far been few efforts to comprehensively discuss these issues. Based on selected reports spanning the last decades, this Review closes that gap by providing both a concise overview of defects in MOFs, or more broadly coordination network compounds (CNCs), including their classification and characterization, together with the (potential) applications of defective CNCs/MOFs. Moreover, we will highlight important aspects of “defect-engineering” concepts applied for CNCs, also in comparison with relevant solid materials such as zeolites or COFs. Finally, we discuss the future potential of defect-engineered CNCs.


Introduction
In nature," ideal crystals", with an infinite periodic repetition or ordering of identical groups of atoms in space do not exist. Thes tructure of "real crystals" always deviates from that perfect arrangement and contains ac onsiderable density of structural irregularities or defects. [1] Crystal irregularities could stem from compositional inhomogeneities,and this concept is often used interchangeably with the term "defects". In fact, heterogeneity,s tructural disorders,a nd defects of various nature and length scales are key attributes of solid-state materials and strongly affect their physical and chemical properties.I ns ome cases,i ti sd esirable to have crystals as perfect as possible (e.g. for optoelectronics). [2] However,d efects do not necessarily have adverse effects. Hence,m any important material properties rely as much on characteristic imperfections as on the overall "perfect" nature of the host lattice.F or example,the electronic properties (i.e. electrical conductivity) of important materials such as silicon or Group III/V compound semiconductors are entirely due to trace amounts of chemical impurities and defects. [2,3] Many important classes of materials exhibit some form of structural disorder which gives rise to often counterintuitive and useful material properties and functions,s uch as spin frustration in cooperative paramagnets, [4] thermoelectric, [5] and polar nanodomain formation in relaxor ferroelectrics. [6] Surface defects and interfacial structures of nanocomposites (e.g.m etal/ support interactions) commonly serve as active sites in heterogeneous catalysis for adsorption and reactive transformations. [7][8][9] In all these cases,j udicious control over the defect structure and associated heterogeneity,t hat is," defect engineering", is of paramount importance to manipulate crystal quality and thus the specific properties desired in am aterial. Ac omprehensive analysis of any "real" material structure and performance also requires the identification of the existing defect types,t heir densities,a nd distribution, as well as the roles they play in affecting the materials behavior. Being essentially solid-state materials,i ti si mmediately conceivable that metal-organic frameworks (MOFs) should also carry various kinds of defects and may feature structural complexity as ar esult of heterogeneity. [10][11][12] In this Review, however, we would like to introduce the more fundamental term coordination network compounds (CNCs) for the class of materials that is discussed. Both MOFs and CNCs are based on Werner-type coordination chemistry,w here metal ions and/or clusters of metal ions are linked together by oligotopic organic molecules to yield extended network structures.M OFs are essentially as ubset of crystalline, (potentially) porous CNCs (cp-CNCs). Nevertheless,w e also define dense CNCs (d-CNCs), which are not porous, amorphous CNCs (a-CNCs), as well as flexible CNCs (f-CNCs) and rigid CNCs (r-CNCs). Depending on the particular structural features of CNCs,certain variations such as cf-CNCs, cr-CNCs,and cd-CNCs are important as well.
Investigations of the (intrinsic) defect structure and the intentional design of imperfections and structural heterogeneity,h owever,h ave so far not attracted much attention in MOF,o rm ore broadly CNC,m aterials research. The intensive activities in the MOF field over the past years have mainly been dedicated to reticular synthesis,n ew topologies and structures,a nd potential applications in gas storage/separation, [13] sensing, [14] drug delivery, [15] and catalysis, [13,16] with little explicit focus on defects.O nly recently have the external surface as well as internal imperfections,for example,S chottky-type point defects (such as linker and/or metal node vacancies) been characterized for several "canonical" MOFs. [17][18][19][20][21][22][23] Thea gglomeration or correlation (clustering) of point defects could form (functionalized) mesopores,which in turn could help in overcoming diffusion limitations. [19,[23][24][25][26][27] In addition, the modulation of the electronic situation together with the proximate coordination space at so-called coordinatively unsaturated sites of some MOFs,such as HKUST-1 and UiO-66, has been shown to strongly affect their reactive properties. [21,24,28] It is the intention of this Review to present the emerging research field of defect-engineered MOFs and CNCs in asystematic fashion. We will discuss different types of defects,structural complexity,and heterogeneity as well as the importance for further development of the materials chemistry (and physics) of such molecular network materials. Naturally,the issue of defect-engineered MOFs,inparticular, is likely to attract attention;however,the principles are very general and also other CNCs and molecular network materials in general show interesting properties connected with their defective structure.O fc ourse,i nv iew of the huge current interest in MOFs,t he Review will mainly deal with this subclass of CNC materials.

Definition and Classification of Defects
Va rious structural disorders and heterogeneities that break the periodic arrangement of atoms have recently been reviewed for CNCs. [29][30][31] Goodwin and co-workers defined statically, [32][33][34] topologically,a nd dynamically [35][36][37][38][39][40][41] disorderedC NCs [29,42] on the basis of the type of building blocks,t heir connectivity pattern and periodicity,a sw ell as atom/unit dynamics.Diverse strategies have been outlined for varying the building blocks within the lattice or the guests inside the pores to introduce heterogeneity without losing long-range order.F urthermore,t he integration of distinct types of organic linkers or ac ombination of homologous  linkers bearing different chemical functions,a sw ell as combining various metal-containing secondary building units (SBUs) or more than one kind of metal ion with the same topological role within as ingle MOF structure were discussed as attractive approaches to adjust the properties of materials. [30,31] In the latter cases,s uch manipulations of the parent frameworks often occur in an onperiodic fashion. Similarly,defects could be essentially considered as aspecific form of heterogeneity,i mplying the nonperiodic removal of some structural elements.
We will restrict our discussion on defects exclusively to c-CNCs with much focus on cp-CNCs (i.e.MOFs), dictated by the exceptional and continuous interest in this subclass of CNCs.W edefine defects in CNCs/MOFs as "sites that locally break the regular periodic arrangement of atoms or ions of the static crystalline parent framework because of missing or dislocated atoms or ions" (Figure 1). In this sense,t hermal motion of the constituent atoms or linker rotation is not considered as ad efect. Furthermore,a na lteration of the parent framework that occurs periodically or homogeneously, for example,the quantitative removal of all solvent molecules coordinated to metal sites simply leads to the formation of ad aughter framework featuring coordinatively unsaturated sites rather than leading to ad efective form of the parent structure.
In accordance with their dimensions,a ll structural irregularities in solid materials are conveniently considered under four main divisions:point defects (e.g. vacancies), line defects (e.g. dislocations), planar defects (e.g.b oundaries and stacking faults), and micro-and mesoscale volume defects (e.g. inclusions and voids). Additionally,m acropores,c racks,a nd foreign inclusions that are introduced during production and processing of the solids may also be considered here as macroscale volume defects.These classifications could also be adopted for CNCs.According to their location, defects can be designated as external or surface defects [18,[43][44][45] and internal defects. [19-24, 27, 46] Thel atter could originate from partially missing metal nodes or linkers (either the entire molecule or selected functions when fragmented linkers are integrated) and locally break the framework regularity. [19,20,23,46,47] Such vacancy defects could be considered analogues of Schottky or Frenkel defects in classical solid materials (that is,t he local absence of atoms/ions based on the "ideal" crystal structure accompanied by removal of oppositely charged ions or creation of an interstitial defect at the new location, respectively). In the case of defective CNCs/MOFs,c harge compensation, if required, is realized through either reduc-tion/oxidation of framework components [24,[46][47][48] or inclusion/removal of respective counterions, [49] either during synthesis or postsynthetically.T he formation of linker vacancies may further cause the appearance of modified coordinatively unsaturated metal sites (mCUS) [24,27,28,48] or additional CUSs [21,46,[49][50][51] that can greatly differ from the regular intrinsic CUSs.S uch mCUS could, thus,a lso be defined as point defects.
Taking into account the distribution, size,and the state of the interaction/correlation of defects in the framework, they can be subsequently divided into two groups:a )local defects,t hat is,p oint or isolated defects,and b) large-scale defects (e.g.mesopores;F igure 1). Tw od istinct situations might arise depending mainly on the concentration of defects within ag iven structure and the effective size of the defect field. At low defect concentrations and small defect fields,arandom distribution of isolated or point defects can be found. On the other hand, high concentrations of defects and/or large fields of defects can result in the formation of correlated or large-scale defects through clustering of point defects.C orrelation here means that the probability of forming adefect at acertain location in the crystal lattice depends on the presence of defects in the vicinity of this location. In essence,s uch large-scale defects are always of higher dimensions.C onsequently,c orrelated linker/node vacancies can generate mesopores that: a) might greatly affect mass-transport pathways (important in sorption and catalysis), b) could reduce network rigidity and density, c) bring out unique electronic, magnetic, and optical functionalities and anomalous mechanical properties (e.g. negative thermal expansion, pressure-induced softening, and crystalline-amorphous switching) d) may bring benefits to realize complex catalytically active sites,f or example,r earranged CUSs that can operate in ac ooperative manner, [29,46,48] for targeted catalytic reactions.
All the CNCs/MOFs reported so far which fall into our definitions of defects and defect engineering are summarized in Table 1, and some of these examples will be specifically discussed below.

Inherent Defects
Inherent defects are formed during crystal growth as ar esult of stacking faults or dislocations and can occur without targeted engineering.I nherent means that no other manipulations were performed during synthesis besides mixing the normal building blocks of the parent framework under regular synthetic conditions.C NC/MOFs are prone to the formation of inherent defects arising either from misconnections or dislocations during crystallization [44] or from postcrystallization cleavage. [45] An example of such misconnections are screw dislocations,w hich are evidenced by growth spirals on the crystal surface of HKUST-1 ([Cu 3 -  thermal stability dependent on defect concentration;increasing amount of defects with extent of washing [22] . .

Angewandte
Reviews Z. L. Fang et al.

Angewandte
Chemie temperature and starting linker/metal ratio may considerably decrease the concentration of linker vacancies. [55] Thep recipitation rate is also an important factor and its increase could favor vacancy formation. It is clear that (very) fast crystal growth may not ensure enough time for the MOF building blocks to quantitatively adhere to the growing crystal lattice at the right place,and defects cannot be corrected if the coordination bonding is insufficiently reversible.I ndeed, the fast crystallization of IRMOF-3 and MOF-5 leads to the formation of terephthalate (BDC) "holes" with simultaneous decoration of the internal surface with Zn-OH sites,a s revealed by FTIR analysis. [20] This is asimple and convenient method to intentionally generate new catalysts with an acidity and hydrophilicity different from those of the parent MOFs. Some authors have investigated the formation of inherent defects during the growth of membranes and thin films of MOFs;they emphasized the importance of the quality of the self-assembled monolayer substrates (SAMs) used, as their contamination or defective structure can result in ac onsiderably defective surface-mounted MOFs (SURMOFs). [84,85] Other relevant examples will be mentioned below.

Defects Formed during De Novo Synthesis
Theg reat modularity of CNCs/MOFs enables the intentional introduction of various types of defects while retaining the overall structural integrity.

Solid-Solution Approach:C oassembly of Mixtures of Ligands or Ligand Fragments
Thesolid-solution approach, which involves mixing two or more organic linkers directly in ar eaction mixture,i s astraightforward procedure,which has already been successfully used for MOFs.The type of resulting mixed-linker MOF depends on the nature of the incorporated linkers,w hile the framework topology is typically preserved. Fore xample, using isostructural mixed linkers (IMLs) with different secondary functionalities (e.g.s ide groups) but identical linker topology and ligator functionality leads to heterogeneous MOFs ( Figure 3). [86][87][88][89] Alternatively,t he heterostructural mixed linker (HML) strategy uses organic components with different linker topology or structure.Crystalline CNCs with heterostructural linkers may require an ordered distribution of the topologically different linkers in the framework. However,o ther situations are possible.T he kinetics of framework crystallization determine the role of the doping heterostructural linker which may act as ac apping agent, thereby directing the crystallite morphology and surface chemistry, [90,91] or can functionalize the framework interior. [20,27,65,86] Thus,d epending on the connectivity,s ize,a nd secondary functionality of the added linker,the HML strategy can be used in two ways:a )the large mixed linker (LML) approach which utilizes larger linkers that feature ah igher connectivity compared to the parent linker; [65] b) the truncated mixed linker (TML) approach (also called ligand-   [86] . .

Angewandte
Reviews Z. L. Fang et al. fragment coassembly), where molecules with lower connectivity,t hat is,l inker fragments are employed. TheT ML concept is strongly related to the modulation approach, whereby am onocarboxylic acid is added during synthesis to influence crystal growth and morphology.
TheLML strategy is scarcely used, as it often yields only physical mixtures of different phases.T he only example comes from the Matzger research group,w ho prepared defective MOF-5 by doping with 1,3,5-benzenetrisbenzoic acid (H 3 BTB), [65] which triggered ac hange in the crystal morphology from cubic to octahedral. In contrast, the TML approach is more often applied, for example for MOF-5 and Cu 2 paddlewheel based MOFs (HKUST-1, NOTT-101, [27] and NU-125). [59] Choi et al. replaced large amounts of 1,4benzenedicarboxylic acid (H 2 BDC) in the synthesis of MOF-5 by 4-(dodecyloxy)benzoic acid (DBA), which resulted in significant textural changes of the crystals. [66] Curiously,when 50 %ofthe H 2 BDC was replaced, aspongelike morphology was obtained with large meso-and macropores that permeate throughout the entire crystal ( Figure 4).
However,w hen only 30 %o ft he H 2 BDC was substituted, ap omegranate core-shell-type structure was formed, where the central core features the same sponge morphology,but is enclosed by as hell of intact MOF-5. During crystal growth, DBAs eems to coordinate to the growing crystallites and locally impedes crystal growth because of the long alkyl side chains,which eventually leads to the formation of meso-and macropores.H ence,D BA promotes the formation of correlated defects.D BA is removed postsynthetically during the washing steps,t hereby yielding highly accessible mesopores. [66] DBAs erves ad ual purpose as at runcated linker and also as space-filling agent that directs pore formation. Hierarchical dual porous structures with large internal surfaces (mesopores) have also been observed for HKUST-1. [92] Another well-studied MOF for which TML synthesis has proven its merit is the Zr-terephthalate UiO-66. In fact, the use of monocarboxylate modulators (e.g.b enzoic,a cetic,o r trifluoroacetic acid, TFA) in the synthesis of UiO-66 facilitates control not only of the crystallite size [93] but also of the formation of defects.T hus,t he modulator here acts as at runcated linker.I mportantly,t he concentration of defects in the resulting MOFs relies on the modulator concentration in the synthesis mixture,thus allowing acontrolled increase of the solid porosity. [24] Furthermore,V ermoortele et al. showed how the incorporation of TFAo pens up more Lewis-acidic mCUSs on the Zr 6 clusters. [46] TheT FA could be removed from the already defective framework by thermal treatment, thereby resulting in af urther increase in the number of Zr-CUSs.

Metal Node Vacancies
Metal ion/node vacancies can occur in asimilar way to,or possibly even simultaneously with, linker vacancies.However, reports of such defects are scarce.O ne example concerns copper vacancies in some MOFs featuring paddlewheel clusters. [80] Similarly,w hen using isophthalates as linker fragments in the synthesis of HKUST-1, up to 8% of the Cu dimers were reported missing in the defective material. [59] Moreover,C liffe et al. showed that in UiO-66(Hf), besides linker vacancies,entire Hf 6 clusters can be missing. [23]

Defect Formation by Postsynthetic Treatment
Acid/base postsynthetic treatment has been demonstrated to be an efficient strategy for introducing defects into preformed MOFs.F or example,w hen [Fe 3 O(BTC) 6 (OH)-(H 2 O) 2 ]( MIL-100(Fe)) was treated with TFAo rH ClO 4 , reprotonation of one of the BTC linkers at the Fe trimers occurred with the concomitant formation of additional Lewisacidic CUSs ( Figure 5). However, the loss of one negative charge of the linker necessitates the incorporation of counteranions in the pores of acid-treated MIL-100(Fe);these could decrease the porosity of the defective material. [49] Defect sites can also be formed during solvent exchange and subsequent material evacuation/activation. Forexample, Shearer et al. observed an increase in the concentration of linker vacancies in UiO-66 with increased washing,p ossibly through hydrolysis. [22] Similarly,t he immersion of [Zn 4 O- Figure 4. Pore structures of MOF-5, spng-MOF-5, and pmg-MOF-5 revealed by SEM observationo ft heir crystal surface and interior (scale bars, green 50 mm, red 1 mm, blue 500 nm). [66] Figure 5. Proposed acid-activation mechanism of the Fe 3 Ocluster by aB r ø nsted acid (HX). Anew mCUS is opened up on Fe (yellow squares). [49] Defect-Engineered CNCs Angewandte Chemie (PyC) 3 ]( PyC = 4-pyrazolecarboxylate) in water leads to the formation of vacancies.Interestingly,over the course of 24 h, up to half of the linkers and 25 %ofthe Zn ions are leached from the framework, thereby resulting in an ew lower symmetry phase with ordered or correlated metal and linker vacancies. [55] Arecent report revealed that thermal treatment of MOF-5 at temperatures below the decomposition point of the framework but far above the conventional evacuation temperatures induces the in situ decarboxylation of BDC, effectively generating linker fragments postsynthetically. [50,69]

Defect Engineering in CNCs/MOFsv ersus Related Inorganic and Organic Materials
CNCs are molecular network materials.T hey can be regarded as hybrid materials with properties between those of inorganic networks,s uch as zeolites,a nd purely organic frameworks,s uch as activated carbons or porous organic polymers.W hile conventional (porous) inorganic solids consist mainly of Al, Si, O, and P, the almost infinite combinations of metal ions and organic linkers results in CNCs/MOFs featuring ahigher degree of chemical and structural diversity. Ther elatively low bond energies of coordination bonds (15-50 kcal mol À1 )l ead to ac ertain lability and in some cases kinetic reversibility of the coordination bonds. [94] Therefore,it is expected that defect engineering in CNCs/MOFs allows more degrees of freedom for rational design and as ac onsequence amuch better physicochemical understanding of the defects,their formation, and implications could be gained. For zeolites,the most common "defect" is the substitution of Si 4+ by Al 3+ . [95,96] As ac onsequence of charge neutrality,s uch isomorphous substitution is accompanied by the formation of hydroxy groups,which is the main reason for the existence of Brønsted acidity in zeolites.T his is in analogy to the mixedmetal CNC/MOF solid solutions which display similar phenomena. Local metal exchange can be described as substitutional point defects or impurities.O nt he other hand, in inorganic aluminosilicates,s elective removal of Al atoms (dealumination) or Si atoms (desilication) could be achieved by steaming or acid/base leaching. [95][96][97] Likewise, ap ostsynthetic treatment with H 2 O 2 is commonly employed to remove titanium. [97] Such network modifications,a si n certain cases of isomorphous metal substitution, give rise to additional internal OH groups.T he removal of even more significant fractions of framework cations leads to the formation of am esoporous matrix throughout the zeolite crystal, thereby increasing the pore volume and surface area of the zeolite.I naway,t hese processes are similar to the formation of linker vacancies in MOFs.Apeculiar observation in terms of the formation of these network materials was reported by Meza et al. They showed that surface nucleation and terrace spreading during silicalite synthesis can be switched on/off through careful control over the silicate supersaturation. Insufficient rates of terrace spreading relative to surface nucleation rates lead to the incorporation of defects in the framework, which opens up the possibility to control the defect density and intergrowths through as ubtle control of the synthetic conditions. [98] Likewise,f ast precip-itation and modulation approaches have been used to control the nucleation, growth, and crystal defectiveness of CNCs/ MOFs. [20,23] Defect engineering in covalent organic frameworks (i.e. COFs,often crystalline and porous solids) and related organic materials is in its infancy. Interestingly,the studies by Ourdjini et al. on self-assembled surface-confined COFs illustrate that defect density crucially depends on the targeted dimensionality.T hus,2 D-ordered COFs always contain topological defects,w hile lower dimensional chains/ribbons are almost always more or less perfect. [99] Furthermore,i na nalogy to MOFs,t he TML or fragmented-linker concept afforded defect incorporation in COF-102. By co-condensing atrigonal trisboronic acid with the parent tetrahedral monomer,aCOF-102 with over 30 %i ncorporation of the defective monomer was synthesized. [100] Moreover,afast precipitation synthesis was likewise useful to strongly increase the microporosity of porous organic cages. [101] Liang et al. recently reported on the chemical modification of electrical defects in ap rototypical organic semiconductor,r egioregular poly(3-hexylthiophene) (P 3 HT). [102] Tr eatment of P 3 HT,e ither with LiAlH 4 or Me 2 SO 4 ,r esulted in elimination of some of the p/n-type defects,a nalogous to p/n-type doping.Asimultaneous treatment with both compounds further strongly decreased the defect density,thus leading to an improvement in the material performance,e specially in terms of stability against photodegradation. Similar defect engineering and defect correction could be envisioned also for COFs when aiming at fine-tuning semiconductor properties.

Experimental Studies on Analyzing Defects
Thed iscussion on defective CNCs/MOFs has so far primarily focused on their synthesis and properties.A s ac onsequence of the lack of suitable characterization methods and developed analytical procedures,amolecularlevel understanding of the electronic and steric properties at the defective sites is limited. Another major challenge is to establish fundamental correlations between the defects and properties of the resulting defective materials.Inthis section we will highlight some of the most useful techniques for the physicochemical analysis of defective CNCs/MOFs. Long-range defects can be directly imaged using AFM, SEM, and TEM. Forexample,AFM snapshots can reveal the formation of cracks through desolvation, [45] or AFM can be used to image the formation of growth spirals associated with screw dislocations,a sf or HKUST-1 (see Figure 2), [Cu(1,3bis(4-pyridyl)propane) 3 Cl 2 ]·2 H 2 O, [82] and MOF-5. [60] Furthermore,f orce modulation microscopy (FMM) is used extensively to image surface defects and compositional changes in composite materials.C onfocal fluorescence microscopy (CFM), which has been widely applied in biological imaging, has found application also in the imaging of porous materials to visualize defects that are confined to the crystal interior. Fore xample,i th as been used to study the initial catalytic activity of individual zeolite H-ZSM-5 crystals by monitoring . .

Angewandte
Reviews Z. L. Fang et al.
the intracrystalline location of fluorescent products obtained by the acid-catalyzed self-condensation of furfuryl alcohol inside such crystals. [103] Ameloot et al. recently used CFM to image defects in HKUST-1 crystals,r elying on the same furfuryl alcohol self-condensation, catalyzed by acidic framework defects. [17] 3D CFM images were constructed from as eries of 2D micrographs of cross-sections through the crystals.These 3D images help to visualize the penetration of surface defects into the crystal interior.Inthis way,the main crystallographic directions along which plane dislocations result in free,p endant carboxylic acid groups could be identified ( Figure 6). CFM was also auseful tool for imaging postsynthetic defect formation and assessing the catalytic activity in [Ba(BTC)(NO 3 )]. [50] Thus,C FM can be used to judge the relative merits of different routes for the preparation of catalysts and to evaluate MOF stability associated with defect formation.
Thermogravimetric analysis (TGA) is one of the most accessible methods that can provide first-hand information to assess the presence of defects.Arepresentative example of its use is found in the study of UiO-66, built up from [Zr 6 O 4 (OH) 4 ]c lusters interconnected by am aximum of 12 BDC linkers.Onthe basis of TGA data it was estimated that about 1t o3out of 12 linkers is inherently missing at each cluster ( Figure 7). [54] Apart from an estimation of the stoichiometry,T GA, particularly when coupled with MS, provides reliable information on the type and quantity of incorporated guest molecules,which commonly increases for the same structure upon the creation of defects.T his was shown, for example,f or Prussian Blue structures with lattice defects in the form of missing [Fe II (CN) 6 ] 4À moieties.Solvated water, which fills up such vacancies,s trongly favors Cs + adsorption. [81] Finally,several studies reported lower thermal stabilities for defect-rich MOFs in comparison with their "nondefective" analogues. [22] In certain cases,f or example,w hen mixed or fragmented linkers are used, the introduction of defects might easily be confirmed by routine spectroscopic techniques,f or example, FTIR, UV/Vis,diffuse reflectance (DR), or Raman spectroscopy.F or example,L illerud and co-workers found that the Raman spectrum of defective UiO-66 features splitting of several bands associated with carboxylate groups and ac oncomitant weakening of the fingerprint vibrations below 500 cm À1 . [22] Furthermore,N MR spectroscopy and HPLC analyses of digested samples are also used to qualitatively and quantitatively assess the inclusion of distinct linkers,w hile CP-MAS NMR spectroscopy may provide information on the chemical state of the linkers and linker fragments. [20,46] Ty pically,d efective derivatives retain the long-range order and topology of the parent framework, but suffer from losses in short-range order because of the presence of randomly located defects.T herefore,t heir powder X-ray diffraction (PXRD) patterns should provide similar symmetry and lattice information as the patterns of the respective parent frameworks.T hus,a lthough being the most common technique for structure determination/identification, PXRD generally provides little information on defects.H owever, high-resolution neutron scattering studies can provide direct structural evidence for linker vacancies,a sh as been recently demonstrated for UiO-66. [19] As the X-ray scattering crosssection of each element is proportional to the square of its atomic number,the XRD pattern of UiO-66 is dominated by the heavy Zr atoms,w hile being rather insensitive to the lighter elements of the linker molecules.O nt he other hand, organic linkers and metal centers are equally sensitive to neutron diffraction as ac onsequence of the similar neutronscattering cross-sections of Zr, O, C, and D. Consequently, through structural refinement, the refined linker occupancies derived from data recorded at 4K and 300 Kw ere determined to be 91.7 and 89.0 %, respectively,which corresponds to 1o ut of 12 linkers missing, in good agreement with the TGA results.F inally,s ingle-crystal XRD could provide extremely valuable insight into the defect structures of systems available as sufficiently large single crystals (ca. 5- Figure 6. Top: 3D CFM image of an HKUST-1 single crystal obtained after an extended crystallization time. [17] Bright planes represent planes of COOH dislocations. Bottom:2 Dsections through the 3D reconstruction. Figure 7. TGA curves of UiO-66 samples from different synthesis batches, indicating the presence of linker vacancies. The theoretical weight loss for af ully coordinated UiO-66 is indicated by the bold vertical arrow between the two horizontallines. [54] Defect-Engineered CNCs Angewandte 100 mm). [42] Forexample,the Lillerud group was able to refine the UiO-66 structure and reveal an approximate 73 % occupancyofthe BDC linker based on synchrotron measurements (l = 0.760 ) on single UiO-66 crystals. [70] Ad eeper understanding of the structural and electronic changes at the defect sites requires methods that are more sensitive to local chemical environments.T his toolbox includes electron paramagnetic resonance (EPR), extended X-ray absorption fine structure (EXAFS), and X-ray absorption near-edge structure (XANES) analyses.F or example,by using EXAFS and XANES measurements,B aiker et al. proved the general decrease of the coordination number of Cu ions upon doping of HKUST-1 with the modified/ fragmented linker 2,5-pyridinedicarboxylate (PyDC), thus proving its framework incorporation. [28] Moreover,EXAFS in conjunction with X-ray photoelectron spectroscopy (XPS) and energy-dispersive X-ray spectroscopy (EDX) data enabled Yang and co-workers to confirm the presence of Cu vacancies in MOF-505 and to determine their concentration. [80] Remarkably,t he EXAFS spectrum of MOF-505 measured at the Cu Ke dge lacks the Cu-Cu coordination peak evident in Cu acetate,t hus,s trongly indicating the presence of Cu vacancies.V alvekens et al. employed EPR measurements to study the low-coordinated Ba sites in their postsynthetically modified defective material [Ba(BTC)-(NO 3 )] (Scheme 1). [50] Exposure to molecular oxygen results in the formation of superoxide ions on the Ba sites;t he superoxides are readily identified in the resulting EPR spectra. Finally,c ertain combinations of the aforementioned analyses can shed light on the distribution of defects.F or example,when diamagnetic [Al(OH)L] n MOFs (L = BDC or NDC) were doped with paramagnetic V 4+ centers,M AS-NMR and EPR spectroscopy could be used to prove the random distribution of these defects. [78] One the most powerful methods to characterize the properties of CUSs in MOFs is FTIR spectroscopy with various probe molecules.Ithas found widespread application in the characterization of the acid-base properties of oxides and porous materials. [104][105][106] Several molecules including CO, CO 2 ,a nd CD 3 CN appear to be the most suitable probes for studies on MOFs.F or example,V imont et al. used CO to assess the Lewis acidity of MIL-100(Cr). [107] Following this successful approach, similar investigations have also been conducted on the parent HKUST-1 [108] as well as on defectengineered MIL-100(Fe). [49] Furthermore,COchemisorption and thermal desorption monitored by ultrahigh-vacuum (UHV) FTIR [109,110] has recently been used by Fischer and co-workers to characterize the local environments of mixedvalence Cu 2+/+ paddlewheel nodes in defective HKUST-1. [24] Thed ata, in conjunction with accurate density functional theory (DFT) calculations afforded insight into the structural and electronic properties of the formed mCUSs.T he optimized Cu 2 paddlewheel structures were consistent with one CO molecule coordinating to aregular Cu II 2 (BTC) 4 unit and two CO molecules coordinating to the defective Cu II Cu I -(BTC) 3 units with linker vacancies.A nalogous experimental and theoretical studies were also conducted on HKUST-1 thin films,where two dominant CO FTIR signals at high and low frequency for the Cu 2+ and Cu + ions,r espectively,w ere observed, thus indicating the presence of defective units. Thus,high-resolution UHV-FTIR measurements can serve as aquality-control indicator for defects. [47] CD 3 CN was used by several groups to study the number and strength of Lewisacidic defects in materials such as the Bi-BTB MOF CAU-7 [79] or in Zr-based MOFs,for example,MIL-140 [53] and the UiO-66 series. [46] While anegligible influence on acid strength was observed in the defect-engineered UiO-66, the amount of Lewis-acidic CUSs on Zr 4+ increased from 0.72 mmol g À1 in the nonmodulated material to 1.1 mmol g À1 in the defective MOF when TFAw as added to the synthesis mixture.T he latter corresponds to two Zr-CUSs per cluster. [46] Finally, pyridine was applied as aprobe by Ravon et al. to investigate defects introduced in MOF-5 and IRMOF-3. [20] Thea bovementioned methods are particularly helpful to detect and characterize local defects and to probe their chemical and physical environment. However, they do not distinguish between isolated and correlated defects.O ne indication for defect clustering is the characteristic type IV isotherm for N 2 adsorption. In fact, by identifying this sorption behavior, several groups could prove that the clustering of numerous local defects results in larger-scale mesoporosity for HKUST-1, [24] UiO-66, and PCN-125. [19,27,46] Calculation of the Brunauer-Emmett-Teller (BET) surface area also revealed ar elationship between the defect concentration and porosity. [19,22,24,27] However,t here is no stringent connection between the change in surface area or pore volume and the employed method of defect engineering. When the concentration of defects is sufficient to create mesopores,small-angle X-ray scattering can also be useful to study the morphology and approximate size of these voids in MOFs. [25,62] Ordered defect correlation is different from the random defect clustering that forms mesopores in some materials. Nanoregions with an ordered (correlated) structure of missing nodes have been found in defect-engineered UiO-66(Hf) (Figure 8). [23] In fact, symmetry-forbidden PXRD reflections Scheme 1. Schematic representation of mCUS and superoxide formation upon activation of [Ba 2 (BTC)(NO 3 )];M=Ba. [50] . .

Angewandte
Reviews Z. L. Fang et al.
were observed for such UiO-66 samples and its functionalized analogues. [111,112] These reflections could be indexed in the same cubic cell by lowering the symmetry from face-centered to primitive;h owever, their origin remained unclear. Furthermore,t hese diffuse scattering peaks were revealed to be strongly dependent on the synthetic conditions.R ecently, Goodwin and co-workers used ac ombination of theoretical and experimental techniques to rigorously attribute the presence of the superlattice reflections to correlated defect nanoregions.T he authors concluded that primitive nanodomains arise within UiO-66(Hf) through correlated linker and node vacancies.T hese domains adopt the reo topology with eight-connected clusters as opposed to the fcu topology of the parent lattice which contains twelve-connected nodes, while retaining an almost similar cell parameter.Anatomistic model that accounts successfully for both the sharp Bragg peaks and the broad diffuse scattering component in PXRD pattern was constructed ( Figure 8). As ac onsequence of the simultaneously missing linker and nodes,t he rather complex framework composition cannot be deduced by TGA or neutron diffraction alone. [23]

Theoretical Methods To Model Defects in CNCs/MOFs
Molecular modeling is ap owerful and essential complementary tool for understanding the nature and structure of defective sites and their distribution through the framework matrix ( Figure 8);i ti se qually important for elucidating the origin of the enhanced or changed adsorption or reactive properties of defect-engineered CNCs/MOFs.
Dislocations and stacking faults create as et of extended microscale crystal defects which can radically alter material properties.H owever,i tr emains am ajor computational challenge to determine the influence of such defects on the atomic level. By applying an ewly developed simulation technique,W alker et al. have successfully modeled the structure of screw dislocations in zeolite A. [43] Thep ore system is perturbed at the location of the screw dislocation, thereby causing local blocking of the pores that connect perpendicular to the channel containing the dislocation line, while the neighboring channels are instead only deformed from acircular to an oval cross-section. Thepredicted channel structure,and generated local chiral sites,should considerably affect the transport of molecules from the surface to the interior,w hile retarding transport parallel to the surface. Related studies on HKUST-1 revealed similar results. [56] In this latter case,t he modeled screw dislocations run through the Cu 2 dimers,which are no longer fully coordinated by the BTC linker,but instead by monomeric ions (e.g. -OH);inthis way,c harge neutrality of the framework is preserved.
Quantum mechanical/molecular mechanical (QM/MM) calculations were applied to investigate the local defects at CUSs in HKUST-1 doped with PyDC ( Figure 9). [24] The studies confirmed asignificant modification of the proximate coordination space and as imultaneous change in the electronic properties of the MOFs upon incorporation of PyDC. In accordance with the experimental studies,t he incorporation of PyDC facilitates the formation of mixed-valence Cu 2+ / + paddlewheel nodes compared to the parent framework which contains only as mall fraction of inherent Cu + sites (< 5%). Importantly,a saconsequence of the generation of   3 (PyDC) and energetically feasible binding modes for one to three adsorbed CO molecules (only the QM system is shown for clarity;Cub rown;C black;Ored;Nblue;Hwhite) together with the computed (scaled) CO stretching normal-mode frequencies(cm À1 ). e) The defect (QM system) embedded in the MM environment. [24] Defect-Engineered CNCs Angewandte Chemie additional coordination space around the metal sites,m ore CO probe molecules could be adsorbed at the mCUSs,w ith higher calculated binding energies relative to the native CUSs.
Similar calculations were performed on UiO-66 to study its catalytic activity in the cyclization of citronellal. Vermoortele et al. [21] reported that realistic transition states for this reaction could only be found for defect cluster models,that is, lacking at least one linker. Subsequent studies succeeded in controlling the number of defects by incorporation of TFAas alinker fragment. [46] Periodic DFT calculations revealed that the free energy needed to remove TFAf rom the Zr 6 cluster equals 24.7 kJ mol À1 ,w hich indicates that high-temperature activation of such modulated UiO-66 materials results in af urther increase in the defect site number,p erfectly in line with the experimental data. From the construction of freeenergy diagrams,i tb ecame apparent that the formation of defects in UiO-66 is not thermodynamically favored. However,t his barrier is significantly reduced in the presence of TFAd ue to entropic effects.T hus,w hen there is ah igh concentration of TFAinthe synthesis,the formation of linker vacancies becomes very likely despite the higher free energy associated with this defect type. [78] As imilar approach was applied to defects generated in [Ba(BTC)(NO 3 )]·DMF,where DFT-based simulations were used to calculate proton affinity values for Ba-CUSs. [50] These calculations showed the defect sites on Ba to be more basic than in the native material.
Lastly,g rand canonical Monte Carlo simulations were employed by Snurr and co-workers to simulate the water adsorption behavior of defective UiO-66(Zr). [71] By creating au nit cell in which one missing linker was replaced by four hydroxy groups,t hey simulated that water uptake already starts at 0.1-0.2 p/p 0 ,w hile for the nondefective UiO-66 this only occurs at 0.8 p/p 0 and with alower saturation loading for the parent MOF.T he shape of the measured isotherms more closely fits to simulated isotherms of defective materials, albeit with the adsorption onset at higher p/p 0 and lower saturation values,thus indicating that the real samples contain fewer vacancies than the modeled ones.Inthis case,the rather hydrophobic UiO-66 is thus imbued with hydrophilicity because of OH-filled vacancies,t he location of which influences the onset of water adsorption and saturation values.

Impact of Defects on CNCs/MOFsF unctions and Properties
MOFs (cp-CNCs) are defined by their regularly repeating crystal structures in which all the pores have exactly the same size,s hape,a nd functionality.T his greatly facilitates the establishment of structure-property relationships,w hich in turn allows us to tailor MOFs for certain targeted applications,s uch as shape-and size-selective adsorption, catalysis, and sensing.T he introduction of defects disrupts the regular porous interior of MOFs and so the behavior for the intended applications can be drastically altered compared to that of the parent MOFs.F or example: 1) linker and/or metal vacancies could affect mass-transport pathways within the pores which is important for adsorption and separation processes; 2) metal vacancies might give rise to electronic coupling states,w hich in turn influence electronic,m agnetic,a nd optical functionalities; 3) defect engineering may prearrange CUSs in acooperative manner so that complex active sites can be achieved for targeted catalysis within MOFs.

Defect-Engineered CNCs/MOFsinC atalysis
In the past decade,m uch research has been focused on introducing catalytic species into MOFs,w hile the role of defects in catalytic processes and the engineering of defectbased active sites in MOFs have only relatively recently gained attention. Almost without exception, the targeted defects for catalysis are coordinative mismatches between the linker and metal ions,which results in Brønsted or Lewis acid sites,r espectively.F or example,F arrusseng and co-workers synthesized defective MOF-5 and observed that linker vacancies partly occupied by OH groups catalyze the alkylation of biphenyl with tert-butyl chloride.W hile the conversion of biphenyl over fast-precipitated MOF-5 was comparable to that obtained with zeolite H-BEA (28 % versus 30 %), MOF-5 was 100 % para-selective,v ersus only 55 %s electivity for H-BEA. [20,63] Similar observations were made by Llabres iXamena et al.,who found acidity in MOF-5asaresult of Zn-OH species and free carboxylic acids;this resulted in an active catalyst for the Knoevenagel condensation of benzaldehyde and ethyl cyanoacetate.T hey further reported on IRMOF-3, which features similar acidic defects, thus making it abifunctional catalyst by virtue of its structural NH 2 groups and OH groups of defective origin. [64] Ameloot et al. exploited the Brønsted acid catalyzed selfcondensation of furfuryl alcohol (see Section 5.1). Importantly,t he catalytic activity of HKUST-1, MOF-5, and MIL-53(Ga) was found to be directly related to the number of defects. [17] Furthermore,inherent Lewis-acidic defects on Bi 3+ observed in CAU-7 were shown to be selective catalytic sites in the hydroxymethylation of 2-methylfuran and possessed the right acidity to avoid the acid-induced formation of furan oligomers. [79] Moreover,Z IF-8 was found to catalyze the transesterification of vegetable oils with several aliphatic alcohols by virtue of defect sites at its external surface but not in the bulk micropores.T hrough combined DFT calculations and CO chemisorption, Chizallet et al. demonstrated that the ratios of acidic Zn 2+ and NH groups as well as of basic N À and OH groups depend on the operating pressure and temperature of the catalyst. [74] Nevertheless,i ti sq uite likely that other reactions catalyzed by the small-pore ZIFs occur at the external surface and involve reactive defects. [75] Leus et al. in situ in MIL-47 by the removal of linkers or by partial linker decoordination. Alonger induction time for the reaction in ndecane than for the reaction in water supports this hypothesis and could point to apartial hydrolysis.However,the presence of inherent V-CUS defects in the as-synthesized materials should not be excluded. [77] Catalytically active acid sites were also introduced in MIL-100(Fe) by apostsynthetic treatment with aprotic acid, which gives rise to an increase in both Lewis and Brønsted acidity because of the protonation and decoordination of acarboxylate group from the Fe 3 Otrimers.Consequently,this MOF was first examined as ac atalyst in the ene-type cyclization of citronellal to isopulegols. [49] Thec yclization of each citronellal isomer results in four isopulegol enantiomers, with the selectivity profile being an indication of the Lewis or Brønsted acidity of the catalyst. Thed efective MIL-100(Fe) showed am arked increase in selectivity to (À)-isopulegol in comparison to the native material, which is atypical behavior of aL ewis acid. While at first this seems counterintuitive,a s the number of Brønsted sites also increases,t he authors propose that ac ooperative effect between the open Fe site and the COOH group in the immediate vicinity results in afacilitated proton migration, thus favoring formation of the industrially relevant (À)-isopulegol. Furthermore,t he acidtreated MIL-100(Fe) was applied as acatalyst in Diels-Alder reactions between 1,3-cyclohexadiene and several dienophiles.I nterestingly,i td isplayed increased activity and selectivity in the reactions with oxygenated dienophiles such as dimethyl fumarate and methyl acrylate,which is attributed to the enhanced activation of the dienophiles at the modified active sites. [49] Another interesting example,concerning postsynthetically generated Lewis-basic sites is the partial removal of nitrate anions as NO x species at high temperatures from the Ba-nitrate chains in [Ba(BTC)(NO 3 )] (see Scheme 1). [50] Thef ormed Ba-O 2À sites provided ah igh activity in several Knoevenagel condensation and Michael addition reactions.
Apart from using inherent or engineered Lewis and Brønsted acid sites for catalysis,d efect sites can be used to introduce other catalytically active species.F or example, Matzger and co-workers showed that up to 2.3 wt %Pdcould be grafted on dangling COOH groups in adefect-engineered MOF-5 derivative doped with H 3 BTB. [65] Ther esulting material appeared to be ag ood catalyst in the Heck-type phenylation of naphthalene with diphenyliodonium tetrafluoroborate.
TheT ML strategy has been used in several materials to modulate catalytic activity.F or example,M arx et al. demonstrated as trong increase in the activity of PyDC-doped variants of HKUST-1 compared to the parent MOF in the oxidation of toluene with hydrogen peroxide.R emarkably, the formation of mainly benzaldehyde and p-methylbenzoquinone was observed, while the parent HKUST-1 is reasonably selective towards ortho-a nd para-cresol. [28] Defect engineering by PyDC doping has also been recently applied to the Ru 2+/3+ -analogue of HKUST-1. [48] In fact, PyDC doping led to materials which feature more strongly reduced Ru x+ species (x < 2) as ar esult of partial reduction of the metal node upon increasing the carboxylate ligator vacancies.
Interestingly,t hose samples demonstrated up to four times higher activity in the hydrogenation of 1-octene,w hich was attributed to the creation of additional modified Ru-CUSs. Furthermore,DeV os and co-workers attributed the catalytic activity of the UiO-66(Zr) mostly to inherent linker vacancies,which are needed to form realistic transition states. [21] To increase the concentration of linker vacancies in UiO-66, monocarboxylates were used as linker fragments in afollowup study.Astriking example of the efficiencyofthis approach was the Meerwein-Ponndorf-Verley reduction of 4-tertbutylcyclohexanone (TBCH) with isopropanol. While the parent UiO-66 showed almost no activity in this reaction (5 % conversion), the defective MOF was able to convert more than 60 %o fTBCH after 24 h (Figure 10). [46] 6.2. Defective CNCs/MOFsinA dsorption, Separation, and Storage In terms of gas adsorption in MOFs,two main properties that can be influenced by defect engineering are the density of the CUS as well as the pore-size distribution and specific surface area. Theintroduction of vacancy defects in sufficient quantities might result in the formation of mesopores,a s explained before.D efect engineering could also be at ool to turn CNCs from ad ense,n onporous material (cd-CNC) to aporous form (cp-CNC), as in the case of aUiO-66 derivative with the small squaric acid (C 4 O 4 H 2 )a slinker. [72] An interesting example of the impact of defects on gas adsorption is NOTT-202, built up from [In(COO) 4 ]u nits linked through biphenyl-3,3',5,5'-tetra(phenyl-4-carboxylate) linkers and featuring ap artially double interpenetrated structure. [36] Thef irst net (A) is fully present whereas the second net (B) is only 75 %o ccupied and is,m oreover, disordered over two positions (B1 and B2) due to symmetry relationships.T he resulting framework thus consists of adominant net Aand independent domains of the secondary nets B1 and B2, the occupancyo ft he latter two each being 37.4 %. Thes econdary nets B1 and B2, however,c annot be connected to each other because of steric constraints and linker overlap.T his induces defect slits with aw idth defined as the distance between opposite dangling ligands in B1 and  [46] Defect-Engineered CNCs Angewandte B2. Then etwork fragmentation and defects allow the desolvated phase (NOTT-202a) to achieve ah igh specific surface area (2220 m 2 g À1 )a nd pore volume,d espite the interpenetration. TheCO 2 isotherm recorded for this material at 195 Ks hows three steps in the adsorption profile,t hus indicating astepwise filling of the pores.T his behavior is not observed for temperatures above the triple point of CO 2 , which indicates non-ordering of CO 2 within the pores.T hus, NOTT-202a has alarge affinity for CO 2 and has the potential for its trapping from gas mixtures,asother gases do not show this type of behavior.
Wu et al. found that linker vacancies lead to adramatically enhanced porosity of UiO-66. Thep ore volume and BET surface area of the samples doped with most linker fragments were found to be 150 %a nd 60 %, respectively,h igher than the theoretical values of the parent material. [19] Furthermore, increased heats of CO 2 adsorption and mesopore formation were achieved in PCN-125 derivatives containing functionalized linker fragments ( Figure 11). [27] MOF-5 synthesized with DBAalso features meso-and macropores with ahigher CO 2 adsorption capacity compared to parent MOF-5. [66] Moreover, thermally annealed MOF-5 samples with in situ generated benzoate fragments show higher CO 2 uptake capacities because of the presence of Zn mCUSs. [69] Modulation of the porosity by defect engineering was also reported for HKUST-1 and NU-125. [59] Isophthalate-doped derivatives feature increased pore volumes and surface areas, while doping with the more bulky 5-perfluorobutylisophtha-late had an opposite effect. Thed oping with isophthalate derivatives was suggested to cause partial removal of Cu 2 paddlewheels (ca. 8%)i nH KUST-1. Consequently,b ecause of these metal node vacancies,the total uptake of H 2 and CH 4 is reduced despite the higher porosity in some cases.This may be due to the reduced number of adsorption sites involving CUSs.I nc ontrast, an enhancement of MOF performance in sorption-related processes could be reached with defectgenerated mCUSs. [48] Tsao et al. studied the correlations between H 2 uptake by bridge spillover in MOF-5 and its pore network, specific surface area, and lattice defects.T hey suggested that aggregated defects,m esopores with ar adius ! 32 , are responsible for variations in H 2 uptake.Similarly,the occurrence of defect mesopores counteracts pore blocking in IRMOF-8 and increases the uptake capacity for H 2 up to 4.7 wt %. [62] Such au nique mesopore network combined with other structural defects (i.e.Z nO species in the nanopores and lattice interpenetration as am inor phase) plays an important role in the uptake of H 2 at room temperature by ab ridged spillover mechanism.

Luminescence, Magnetic, Thermal, and Mechanical Properties
As mentioned in the introduction, the defect structure plays ak ey role in determining the physical properties of solids.M OFs are no exception;h owever, the influence of defect engineering on their physicochemical characteristics has only been investigated sporadically.N onetheless,i th as been shown that the photoluminescence and magnetic properties of defective CNCs/MOFs differ from their native counterparts.Anumber of MOFs have received attention because of their photoluminescence properties and have been recently reviewed. [113][114][115] It is well known that defects play ac onsiderable role in the optical properties of diverse solidstate materials. [114] Forexample,defect-induced luminescence has been found in Co 2+ -doped anatase,Cd-rich CdS and ZnS nanoparticles,a nd Eu 2+ -doped Sr 2 MgSi 2 O 7 .D efects may act either as energy-storage centers,w hich introduces luminescence,o ra sq uenching centers in materials.F or example, defects in Sr 2 MgSi 2 O 7 act as electron traps close to the materials conduction band. These sites store absorbed energy and contribute to an efficient persistent luminescence. [116] Larsen and co-workers reported that [Ru(2,2'-bipyridine) 3 ] 2+ , encapsulated in volume defects in RWLC-2 (built from H 3 BTB and Zn 2+ ions), features reduced emission lifetimes compared to in the solution state because the co-confinement with aq uenching molecule in defect voids activates altered decay channels. [83] Defects could also allow modification of the electronic band structure of CNCs.I nf act, larger band gaps ranging from 3.31 to 3.87 eV were found in TML-modified HKUST-1, while avalue of 2.87 eV was reported for the hydrated parent HKUST-1. [24] This is associated with both the doped amount of linker fragments and the concomitant partial reduction of Cu 2+ to Cu + .  [27] . .

Angewandte
Reviews Z. L. Fang et al.
Structural defects might also alter phonon transport, thereby affecting the thermal conductivity of solids.F or example,this was shown for b-SiC,inwhich phonon scattering by defects strongly reduces the thermal conductivity. [117] Kaviany and co-workers reported the dependence of the thermal conductivity of MOF-5 on the mode of phonon scattering.T he phonon mean free path length in MOF-5 reaches am inimum below 13 K, which is typically attributed to point defects (either inherent or formed through thermal stresses). As revealed, thermal conductivity increases with increasing temperature,and scattering by lattice defects is an important phonon scattering mechanism below 35 K. [67,68] Thed eliberate introduction of metal ion defects,f or example,forming mixed-metal MOFs,could open apromising door to modifying the magnetic properties of CNCs.F or example,W illiams and co-workers observed af erromagnetic coupling between Cu 2 paddlewheels in HKUST-1, with ad ramatic rise in the magnetic susceptibility below 70 K. [118] Motivated by vacancy-induced magnetism in many materials, Shen et al. further proposed the ferromagnetism in aseries of Cu 2 paddlewheel based MOFs to be induced by vacancy defects. [80] Their DFT results excluded effects of other possible defects (such as atomic Ca nd Ov acancies) on the induction of any magnetic moment because of the delocalization of their defect wave functions.B oth theoretical and experimental data indicated that the observed ferromagnetism is due to point defects,t hat is,C u 2 + vacancies.Aclear correlation between magnetic properties and the doping level of PyDC was found in TML-engineered HKUST-1. [24] Finally,the impact of defects on the mechanical properties of MOFs was reported by Goodwin and co-workers,who used ab initio calculations to predict the mechanical stiffness of UiO-66(Hf) containing the correlated reo defect domains. Their calculations revealed as trong decrease in stiffness (Youngs modulus for parent UiO-66 equals 46.8 GPa, for the defective reo regions 23.3 GPa), which potentially could lead to pressure-induced amorphization. [23] Defect-induced anomalous mechanical properties might be ap romising field that could draw future interest. [119]

Conclusions and Further Directions
Defect structure characterization and defect engineering have been mainly connected with catalysis and gas sorption. This area could mature in the coming years,e specially benefiting from deliberately induced linker and metal ion node vacancies.I nt his sense,m any MOFs whose parent structure does not allow much diversity (e.g.s teric and electronic restrictions at the framework metal ion sites) can be considered as "sleeping beauties", in which active sites can be unlocked by careful engineering of defects.W eexpect that the lower coordination number of defect metal ion sites will greatly affect the selective sorption properties of CNC materials with respect to gas separation and storage and also their catalytic activity because of the modified steric and electronic environments in the material. Defect engineering will lead to materials with amultivariate nature,which could be interesting for the implementation of multiple function-alities in these materials.I ng eneral, we propose that the investigation of the (external) surface defect structure of essentially nonporous c-CNCs may also be important for the application of such materials in heterogeneous catalysis.I ti s also clear that using MOFs (cp-CNCs) for catalysis and also for biomedical applications (e.g.drug carriers,drug targeting, cell function control) does actually require some scaling down of the crystallites to the nanosize regime and/or the implementation of hierarchical porosity (macro-meso-micro), because diffusion limitations need to be taken into account. Thes ynthetic methods of downscaling themselves,h owever, are intimately connected with the formation of various kinds of defects and structural heterogeneity,a sw eh ave pointed out. Thus,the connection between all these issues needs to be considered in the future.

Synthesis, Characterization, and Modeling
Thei ntentional implementation of defects in CNCs/ MOFs on various length scales,t hereby causing structural and compositional complexity,i sb eginning to emerge as ap art of the research efforts to develop new classes of functional molecular network materials.W hile the role of defects in other solid-state materials such as oxides has been well-established, the pronounced effects of defects on the properties of CNCs/MOFs deserve much more attention. The very nature of coordination network compounds,i ng eneral, being crystalline scaffolds of inorganic metal ion nodes connected by organic linkers,o ffers great opportunities for defect engineering.T he relative ease by which multiple components such as fragmented linkers and redox active metal ions can be introduced into metal-organic frameworks is aprime example of this.Nevertheless,there is aclear need to investigate the full range of possible defect types and new methods by which they can be introduced into CNCs/MOFs with some control. Fore xample,t he formation of defects by irradiation with high-energy ions,w hich has been used to create mesopore defects in zeolites, [120] could be equally efficient in CNCs/MOFs.A nother important aspect is the application and the development of characterization techniques to unambiguously establish the nature and distribution and/or correlation of defects,a so nly from this information will it be possible to rationally deduce engineering strategies. We highlighted some of the currently employed experimental characterization methods,including chemisorption combined with spectroscopic techniques,a si st he case for FTIR with aCOprobe,gas adsorption techniques to detect changes and anomalies in the nature of the pore system, and several imaging techniques such as SEM and AFM to study surface defects.More advanced techniques can provide alot of useful information concerning defects,f or example the use of pair distribution function (PDF) analysis of X-ray or neutron diffraction data, or CFM to directly image dislocations within am aterial. Scanning tunneling microscopy (STM) as ah ighresolution technique is ideally suited to study structural properties of surface-confined 0-, 1-, and 2D molecular arrangements in real space not only of pristine metal and semiconductor surfaces,b ut also of adsorbate-covered surfa-Defect-Engineered CNCs Angewandte ces. [121] This technique might be suitable for the study of various surface defects of electrically conductive MOF/CNC thin films.Such samples should of course allow acurrent flow between the probe tip and the sample.
Standard techniques for structure determination such as diffraction methods are insensitive to local defects,w hereas spectroscopic investigations do not provide direct structural information. Thus,f or an in-depth understanding of defect engineering,c learly,t heoretical methods are needed as an additional tool to predict the relative stability of potential defective structures and to support the interpretation of spectroscopic results.U pt on ow,o nly af ew research groups have combined comprehensive experimental methods and calculations for this purpose.F uture directions of research will take into account that the approximation of periodic boundary conditions cannot easily be maintained for the modeling of dilute defects embedded in ar egular crystalline environment. To accurately capture this situation within ah igh-level and full periodic quantum mechanical DFT calculation, al arge unit cell must be used, thus leading to av ery large,i fn ot unmanageable numerical effort. On the other hand, because of the molecular nature of the investigated systems,e lectronic interactions (correlations,a lso of defects) are very local. Thus,s pecific multiscale simulation techniques for defective CNCs/MOFs need to be developed. Based on anovel genetic algorithm optimization strategy,an accurate force-field parametrization can be derived in as ystematic way from first principles reference data of nonperiodic model systems.I tw as shown that complex dynamic effects such as negative thermal expansion could be quantitatively predicted by such am ethod. [122] Furthermore,t he controlled introduction of defects during the synthesis of CNCs/MOFs is directly connected to the growth mechanism of the crystalline materials.E ssentially,t his refers to the chemistry of the (external) solid/liquid interface.O nly recently has computational modeling of the external surface structure of CNCs/MOFs (surface termination) been done, again with HKUST-1 as the prototypical case. [123] This study is immediately connected with the coordination modulation concept and, therefore,p oints towards the mechanisms of fragmented linker incorporation as well. Fore xample,t he missing-node defect type created by the TML approach can be regarded as creating an internal surface termination. At this point we want to connect the defect topic with the solvent-assisted linker exchange (SALE) concept. [124][125][126][127][128][129] SALE has already been demonstrated as an important strategy for the generation of av ariety of multifunctional MOF materials previously unobtainable by direct synthesis methods.I tw as recently reviewed by Hupp,F arha, and coworkers. [129] Herein, we propose that defects with missing or weaker coordination bonds in the framework will facilitate SALE. Furthermore,t he mechanism of SALE may well be connected to the defect structure and/or dynamic defects of the parent frameworks.C onsequently,t he engineering of defects and its theoretical modeling may well be relevant to ab etter understanding of the mechanism of SALE.

Defect Structure Related Properties and Functions
In general, the physical properties of CNCs/MOFs and novel functions related to the "physics" of the materials need to be investigated with much more emphasis than in the past. If so,t hen understanding the defect structure will become even more important. Fore xample,m etal ions and counterions could be grafted at defects to control the ionic conductivity.T he resulting solid electrolytes will be potentially useful for enhancing the operation of next-generation lithium batteries. [130] Starting from the pioneering work of Shen et al. [80] on metal-ion vacancies inducing al ong-range ferromagnetic ordering, it will be promising to fabricate new ferromagnetic materials by designing CNCs/MOFs with deliberate metal-ion vacancies.F urthermore,s uch defects could act as energy-storage sites that promote persistent luminescence.T he incorporation of defects further offers the possibility to fine-tune band gaps for optical applications.A breakthrough in CNC/MOF materials research would certainly be the realization of the combination of electrical conductivity with tunable porosity and the design of the internal coordination space.S uch kinds of materials would combine features known from organic semiconductors and "organic synthetic metals" with the reticular chemistry of MOFs.I fs uch materials are developed it is immediately evident that understanding and controlling the respective defect structure for band-gap engineering,the charge-carrier concentration, and mobility will be as important as for other electronically conductive materials.W hile ah uge amount of knowledge exists on conductive (nonporous) coordination polymers,a ne xplicit and systematic investigation of the control of the conductive properties of defect structures has only been done within the context of doping with redox-active components,which is somewhat different from the viewpoint of our discussion.
We want to draw attention to two recent and important results which point towards the more realistic possibility to achieve materials for "MOFtronics". [131] HKUST-1 was turned into an Ohmic conductor by loading with TCNQ (TCNQ = tetracyanochinodimethane). [132] Coordination of TCNQ at the native Cu-CUSs led to ad rastic increase in the conductivity to reach optimum values of 710 À2 S·cm À1 , however, simultaneously compromising the porosity (drop from the parent 1840 to 210 m 2 g À1 for the loaded material). A more detailed investigation of the local, short-range,and the longer-range defect structure (grain boundaries) would certainly be important to better understand the origin and details of charge-transport mechanisms.T ailoring the porosity by the TML approach, similar to the work of Barin et al. [59] mentioned above,would add another dimension of engineering to the material properties.F urthermore,w ew ant to mention the new semiconductive,g raphene-like porous material [Ni 3 (HITP) 2 ] n (HITP = 2,3,6,7,10,11-hexa-iminotriphenylene) with athin-film conductivity of 40 S·cm À1 . [133] The strictly regular, crystalline nature of the parent material frameworks,i np rinciple,a llows more rigorous theoretical investigations and searching for the compositional and structural conditions needed for the development and identification of promising candidates for "MOFtronics". [131] . .