Analysis of a Cu‐Doped Metal–Organic Framework, MFM‐520(Zn1‐xCux), for NO2 Adsorption

Abstract MFM‐520(Zn) confines dimers of NO2 with a high adsorption of 4.52 mmol g−1 at 1 bar at 298 K. The synthesis and the incommensurate structure of Cu‐doped MFM‐520(Zn) are reported. The introduction of paramagnetic Cu2+ sites allows investigation of the electronic and geometric structure of metal site by in situ electron paramagnetic resonance (EPR) spectroscopy upon adsorption of NO2. By combining continuous wave and electron‐nuclear double resonance spectroscopy, an unusual reverse Berry distorted coordination geometry of the Cu2+ centers is observed. Interestingly, Cu‐doped MFM‐520(Zn0.95Cu0.05) shows enhanced adsorption of NO2 of 5.02 mmol g−1 at 1 bar at 298 K. Whereas MFM‐520(Zn) confines adsorbed NO2 as N2O4, the presence of monomeric NO2 at low temperature suggests that doping with Cu2+ centers into the framework plays an important role in tuning the dimerization of NO2 molecules in the pore via the formation of specific host‐guest interactions.


Introduction
Air pollution by nitrogen dioxide (NO 2 ) is associated with serious environmental problems and health risks. [1]Conventional porous materials based upon zeolites and activated carbons generally suffer from low uptake and/or severe structural degradation on adsorption of NO 2 owing to the highly corrosive nature of this substrate. [2]Recently, robust metal-organic framework (MOF) materials have been confirmed to act as efficient spectroscopy (HYSCORE), can overcome this limitation of CW EPR. [12]he ionic radii of Cu 2+ (0.65 Å, 5-coordinate) and Zn 2+ (0.68 Å, 5-coordinate) are similar, [13] and thus partial replacement of Zn 2+ centers in MFM-520(Zn) by Cu 2+ at low concentration is a viable route to afford well-isolated Cu 2+ centers.Here, we describe the synthesis, crystal structure, and in situ EPR studies of a series of Cu 2+ -doped MFM-520(Cu x Z 1-x ) (x = 0.005, 0.01, 0.05) materials.The unusual form of the CW EPR spectrum of MFM-520(Zn 0.995 Cu 0.005 ) can be rationalized by invoking an unusual reverse Berry distorted coordination geometry at Cu 2+ , consistent with the known geometry of Zn 2+ in this framework.This is also confirmed by ENDOR measurements on ligand nuclei ( 14 N, 1 H).Notably, a 10% enhancement of NO 2 adsorption upon inclusion of Cu 2+ sites is observed compared with the parent MFM-520(Zn) (5.02 and 4.52 mmol g −1 , respectively, at 298 K and 1 bar).Modulation of the incommensurate structure is observed upon doping with Cu 2+ ions as well as via the packing of NO 2 molecules that show an elongated intermolecular O 2 N•••NO 2 distance [1.65(5)−1.87(4)Å] compared with MFM-520(Zn) [1.46 (7)  Å].We also report an in situ CW and pulsed EPR study of MFM-520(Zn 0.995 Cu 0.005 ) as a function of NO 2 adsorption focussing on Cu 2+ -NO 2 interactions.This provides a rigorous analysis of the geometry of the Cu 2+ site and affords key insights into the role of the metal site on NO 2 adsorption.

Synthesis, Characterization, and Structure Determination
Cu 2+ -doped MFM-520 materials {[Zn 2-2x Cu 2x (L)]•4H 2 O} ∞ (x = 0.005, 0.01, and 0.05) were synthesized via hydrothermal reactions of ZnCl 2 , CuCl 2 and H 4 L at 130°C for 6 days.X = 0.05 (or 5% dilution) is the highest ratio of doping that can be achieved under these conditions; higher levels of doping yield unknown phases by powder X-ray diffraction (PXRD).The mixed-metal MOFs were isolated as yellow-green microcrystalline powders, and the presence of Cu 2+ confirmed by solid state UV-vis spectra (Figure S3, Supporting Information).Scanning electron microscopy (SEM) confirms the block morphology of the materials with crystal size distribution of 2-10 μm, and energy dispersive X-ray spectroscopy (EDS) analysis confirms the homogenous distribution of both Cu and Zn in all samples (Figure S6, Supporting Information).Analysis of the ratio of Zn/Cu by inductively coupled plasma optical emission (ICP-OES) gives good agreement with the stoichiometries used in the synthetic procedures (Table S1, Supporting Information).The PXRD patterns of all mixed-metal MOFs are consistent with that of the parent MFM-520(Zn) material, confirming retention of the framework structure and the 4 4 6 6 topology [14] (Figure S1, Supporting Information).Attempts to introduce Cu 2+ sites via post-synthetic modification by immersing the pristine MFM-520(Zn) in an aqueous solution of CuCl 2 failed, instead leading to the formation of a new [Cu 2 (L)] material showing a layered structure, which will be reported elsewhere.The retention of vibrational features of MFM-520(Zn) in the mixed-metal MOFs is confirmed by infrared (IR) spectroscopy (Figure S2, Supporting Information).Thus, these results confirm the successful preparation of MFM-520(Zn 1-x Cu x ) (x = 0.005, 0.01, and 0.05) materials with homogenous dilution of Cu 2+ sites within the framework.
The presence of micropores in these MOFs has been confirmed by N 2 adsorption isotherms at 77 K showing the typical Type-I profile (Figure S5, Supporting Information).The Brunauer-Emmett-Teller (BET) surface areas of MFM-520(Zn 1-x Cu x ) (x = 0.005, 0.01, and 0.05) complexes were calculated from the isotherms to be 307, 303, and 309 m 2 g −1 , respectively, similar to that of MFM-520(Zn) (313 m 2 g −1 ).The retention of microporosity rules out the presence of clusters of copper oxide within the pores.Deolvated MFM-520(Zn) has been reported to have an incommensurately modulated structure, being aperiodic in 3D space but periodic in (3 + 2)D space owing to the distortion of [ZnO 4 N] moieties. [15]Synchrotron X-ray single-crystal diffraction at 150 K confirms that MFM-520(Zn 0.95 Cu 0.05 ) crystallises in triclinic system with a 4 4

Continuous Wave EPR Study
CW EPR spectroscopy of MFM-520(Zn 1-x Cu x ) (x = 0.005, 0.01 and 0.05) shows broad bands at room temperature, with only partial resolution of the anisotropic g-values (Figure 1b).On cooling, the spectra sharpens and a shift of the low-field (high g) features to lower field is observed.Below ca. 100 K, well-resolved spectra are observed (Figure 1b) with resolution of the 63,65 Cu hyperfine interaction (nuclear spin I = 3/2, combined 100% natural abundance).At X-band, we observe hyperfine coupling to a single 14 N nucleus (ca.30 MHz; also observed in ENDOR), consistent with the Cu 2+ dopant being incorporated into the Zn 2+ site in rather than being "free" as solvated ions in the pores (Figure 1a).MFM-520(Zn 1-x Cu x ) (x = 0.005 and 0.05) give identical spectra with the linewidth increasing with increasing Cu 2+ content due to long range Cu 2+ •••Cu 2+ interactions.Thus, the spectra confirm dilution of Cu 2+ ions within the Zn 2+ -based lattice (Figure 1c; Figure S13, Supporting Information).
The spectra at low temperature are near to axial, even at Qband resolution.Simulation of the spectra [16] gives g 1, = 2.098, g 2, = 2.112, g 3 = 2.424 with 63 Cu hyperfine coupling constants |A 1 | = 140, |A 2 | = 125, |A 3 | = 206 MHz.These are unusual parameters for five-coordinate Cu 2+ complexes: they are very nearly axial despite the highly distorted geometry, g 3 is very large, and |A 3 | is very small (with |A 1,2 | being unusually large).These observations can be rationalized in the context of the unusual coordination geometry at the metal site with C 2 point symmetry based upon the averaged structure, but it approximates closely to C 2v (Figure 1d). [14]The metal ion is bound by a pyridyl donor on the C 2 axis and four carboxylates, two from the bound pyridyl (O2A/B), and two from the orthogonal layer (O1A/B).There are The two parent geometries for five-coordinate Cu 2+ are elongated square-pyramidal (SP) and compressed trigonal bipyramidal (TBP), with the former much more common. [17]The distortion between these extremes is often quantified via the index  5 = (−)/60 o where  >  are the two largest L•••M•••L angles. [18]The relationship between EPR g-values and TBP-SP distortions is well understood: regular TBP ( 5 = 1) gives a d z 2 ground state with characteristic g x,y > g z ≈ g e (g e is the free-electron value, 2.0023), whilst regular SP ( 5 = 0) gives a d x 2 −y 2 ground state with characteristic g z > g x,y > g e (defining z as the axial direction in either case).Intermediate geometries (0 <  5 < 1) have rhombic g-values due to the mixing of d x 2 −y 2 and d z 2 . [19]This analysis assumes that the distortion lies on the conventional Berry pseudorotation pathway between ideal TBP (D 3h ) and SP (C 4v ): this involves bending the two axial ligands (z, in D 3h ) away from one of the equatorial ligands (say, the x-axis), whilst opening the angle between the other two, on a C 2v pathway (Figure 1d).Parameters for the metal site in MFM-520 give  5 = 0.37, indicating a significant distortion from either ideal geometry.This seems incompatible with the almost axial g-values.However, the geometry lies on a reverse Berry distortion pathway (Figure 1d) with the two "axial" ligands (O2A/B) bent toward one of the equatorial ligands (N), and the angle between the other two equatorial ligands greatly decreased from 120 o (O1A•••Zn•••O1B 99 o ).This is a very rare geometry for Cu 2+ , and has only been observed in a few systems, for example [Cu(terpy)X 2 ] (X = NCS − , Br − ; terpy = 2,2′,6′,6′'terpyridine). [20,21]In these complexes, the central pyridyl defines the C 2 direction, and the distal pyridyls form the "axial" direction.This is partly due to the constraints imposed by the tridentate ligand, but also has a subtle variation with X.Other complexes [Cu(terpy)X 2 ] form the more common distorted SP with an {N 3 X} basal plane. [20,21]The vast majority of small molecules of the [CuLX 2 ] type (L = planar tridentate ligand), including those closely related to the metal site in MFM-520, show distorted SP geometries with L defining the basal plane. [22]In MFM-520, the geometry at Cu 2+ is constrained by the tridentate ligand, but the two additional ligands, carboxylates from pyridyls in the orthogonal layer in the lattice, are also constrained.Hence, the structure cannot deform to give the favored SP geometry.
The normal and reverse Berry distortion has important consequences for the electronic structure of Cu 2+ .In C 2v symmetry, the orientations of the g-matrix axes (but not their assignment to g 1 , g 2 , and g 3 , listed in numerical order) are fixed by symmetry, and the g and A matrices must be coincident.Choosing the labels in Figure 1d, the singly occupied molecular orbital (SOMO) can be described as a linear combination: [23] | SOMO⟩ where a is the 3d coefficient to the SOMO allowing for delocalization of spin density from the metal to the ligands.The coefficient c describes the symmetry-allowed mixing of d z 2 and d x 2 −y 2 that both transform as a 1 in C 2v .It should be noted that this can also be written in terms of a formal angle: [19] | SOMO⟩ = a Equations ( 1) and (2) are equivalent with  = tan −1 c .With c = 0, the 3d contribution to the SOMO is pure d z 2 that corresponds to a TBP stereochemistry.A normal Berry rotation pathway gives a positive value for c, with the SP limit corresponding to c = +1/√3, giving the 3d contribution to the SOMO as This can also be written as d z 2 −y 2 , where x is the axial direction of the SP, and a reverse Berry rotation gives a negative value for c.
Perturbation theory gives the g-values as: where ∆g i = g i − g e , and ∆E i is the energy gap from the ground state to the excited state, which is mixed in under spin-orbit coupling in that orientation. is the spin-orbit coupling constant for Cu 2+ (830 cm −1 , free ion [24] ), and b i accounts for delocalization in the excited state.Equations ( 3)- (5) show that the near-axial g-values for MFM-520(Zn 0.995 Cu 0.005 ) with g 3 >> g 1,2 are only possible with |c| ≈ 1/√3.Given the "reverse" geometry, we must therefore be near the limit of c = −1/√3.This gives the SOMO as which can also be written as d z 2 −x 2 : in this limit the other a 1 d-orbital is d y 2 .Hence, the g-values imply that we have a near pure d z 2 −x 2 ground state and that g y (Figure 1d) is the unique and largest g-value (g 3 ).Unfortunately, we are unable to confirm this assignment by single crystal EPR because of the orthogonal layer structure in the MFM-520 lattice, but such measurements have confirmed the large g-value in the equivalent orientation (orthogonal to the terpy plane) in [Cu(terpy)(NCS) 2 ] where c = −1/3. [23][Cu(terpy and 158.2(2) o for X = Br, 98.1(3) and 158.9(2)°f or X = NCS, respectively] compared to MFM-520 [99.1(3) and 152.4(2) o , respectively], and has more rhombic g-values (and a smaller g y ).Thus, both the structural and EPR data are consistent with the Cu 2+ site in MFM-520 being "further along" the reverse Berry distortion than [Cu(terpy)(NCS) 2 ].It should be noted that simulations in red (sharp peaks at ca. 15 MHz are overtones of 1 H).Bottom: calculated ENDOR frequencies in the molecular xz plane at B 0 = 1142 mT (red;  = 0 (z) to 90 o (x) for  = 0 o ) and xy plane at B 0 = 1142 mT (black;  = 0 (x) to 90 o (y) for  = 90 o ); f) Q-band (33.76 GHz) selective ENDOR experiment spectra (black).Simulation (H1: blue, H2: red, sum: magenta), including dipolar interaction and isotropic contributions to the hyperfine resulting from electron transfer from the metal into the  system of the ligands.a d z 2 −x 2 ground state is also consistent with the large 14 N hyperfine coupling constant (ca.30 MHz) to the pyridyl lying on the x-axis (Figure 1d), and is similar to that observed in square planar [Cu(pyridine) 4 ] 2+ . [25]he ∆E i terms in Equations ( 3)-( 5 ] 2− complex). [22]The broad peak at 960 nm has a FWHM of ca.3000 cm −1 , and thus it is feasible that it consists of excitations to these multiple states.The narrow, weaker peak at higher energy can then be assigned to the d yz excitation: this transition would be formally forbidden in C 2v , but allowed in C 2 , which would also contribute to its weaker intensity.Substituting these excitation energies into Equations ( 3)-( 5) with c = −1/√3, and taking a 2 = 0.8 (from analysis of the hyperfine interaction, see below) and b 2 = a 2 (i.e., assuming delocalization is similar for each d-orbital), gives g 1 = 2.06, g 2 = 2.41, and g 3 = 2.10.This is in remarkable close agreement with the experimentally observed values given the approximations in the theory.
On the basis of a near pure d z 2 −x 2 ground state, consistent with the g-values, and approximating to axial symmetry [with A ⊥ = (A x + A z )/2 and g ⊥ = (g x + g z )/2 ; A ∥ = A y and g ∥ = g y ], the following expressions can be derived for the Cu hyperfine interactions: where P d is the electron-nuclear dipolar coupling parameter for 63 Cu 3d orbitals (+1197 MHz [26] ).There are three contributions to the hyperfine constants in Equations ( 6) and ( 7): i) an isotropic Fermi contact term (proportional to ); ii) a spin-dipolar contribution (proportional to a 2 ) relating to ground state d-orbital electron spin density (see Equation ( 1)); and iii) orbital dipolar contributions via spin-orbit coupling (proportional to ∆g).The orbital contribution is opposite in sign to the spin dipolar part and hence a large g-shift, reflecting a low-lying excited state (Equations ( 3)-( 5)), will lead to a reduction in the observed hyperfine coupling.This latter effect, rather than extensive delocalization (small a 2 ) or significant d z 2 mixing, is the cause of the unusually small hyperfine couplings observed in some other systems, e.g., in D 2d distorted [CuCl 4 ] 2− and in ″type 0″ copper proteins. [27,28]quations ( 6) and ( 7) can be combined to give: which removes uncertainty associated with the magnitude of .Equation ( 8) can be solved for a 2 by substituting in the experimental g and A parameters.Assuming that the leading component of the Cu hyperfine coupling (A ‖ ) is negative as predicted by theory for Cu, two solutions are possible.Taking A ┴ to be posi-tive or negative gives a 2 = 0.77 or 0.52, respectively (i.e., ≈80% or 50% of the spin density is on the metal).The former is in keeping with the many simple Cu 2+ complexes with O/N-donor sets that have been analyzed by EPR spectroscopy, and thus this solution seems more likely.This would imply that the isotropic part of the hyperfine [A iso = (2A ⊥ + A ∥ ) /3] is rather small: this can be due to the large spin-orbit contribution to A iso [of the order P d (2Δg ⊥ + Δg ∥ )/3] that is opposite in sign to the Fermi contact term, [27] although valence shell spin polarisation effects can also be significant. [29]n summary, the unusual form of the CW EPR spectrum of MFM-520(Zn 0.995 Cu 0.005 ) can be rationalized by an unusual reverse Berry distorted coordination geometry, and we can assign the observed g-values with respect to the structure.The large g and small values for |A| are consistent with the weak ligand fields dominated by the four carboxylate donors, combined with the unusual geometry.Moreover, the data are consistent with the Cu 2+ dopant adopting the parent structure of the [ZnNO 4 ] moiety within the lattice, even though this would not be its preferred geometry.This is supported by the fact that we have not been able thus far to synthesize the neat Cu-based analog MFM-520(Cu).It should be noted that this study was limited to Cu doping level of 5% or lower.Higher levels of Cu 2+ doping will distort the structure of the parent Zn 2+ material and promote the formation of a 2D material through the direct combination of the H 4 L with Cu 2+ with a more usual and expected distorted 6-coordination at Cu 2+ .The observed reverse Berry distorted coordination geometry at Cu 2+ , which deviates from the preferred and usual geometry at Cu 2+ , can only be maintained at lower Cu 2+ doping concentrations, where all the Cu 2+ ions are confined within the parent framework structure defined by MFM-520(Zn).We also note that the variable temperature linewidths and g-values in the CW EPR spectra at higher temperature are indicative of dynamic effects involving the Cu 2+ coordination sphere, which are then frozen out below ca. 100 K. Given the steric restrictions imposed by the lattice, it is highly unlikely that we are witnessing a dynamic average over two SP structures with O1A and O1B along the axes.However, distortions that open the O2A•••Cu•••O2B and/or O1A•••Cu•••O1B angles would be on the pathway toward normal Berry rotation and would result in a decrease in g y and an increase in g x .Such a dynamic process at higher temperatures would be consistent with the observed broadened linewidths and the upfield shift of the lowest-field features in the EPR spectra.At higher loadings of Cu 2+ (x = 0.05), there is a very broad signal of low intensity, which is more apparent in Q-band spectra.This accounts for a small amount of the total signal intensity (<10%) (Figure S13, Supporting Information) and may be due to Cu 2+ not incorporated into the lattice but in extra-framework surface or pore sites and with higher mobility.Such species are often observed in EPR studies of Cu-doped zeolites. [30]

ENDOR Spectra
In the CW X-band EPR spectrum, there is partial resolution of 14 N hyperfine (nuclear spin I = 1) on the g ∥ feature that arises from the bound pyridyl (Figure 1e).To define this better, we have performed pulsed Davies ENDOR spectroscopy measurements.In ENDOR, neglecting quadrupole effects, peaks are observed at  S27, Supporting Information), but are well-resolved at Q-band ( H and  N = 51 and 3.7 MHz, respectively, at B 0 = 1200 mT).With the inclusion of 14 N nuclear quadrupole coupling, each peak is further split by |3P| where P is the effective quadrupole coupling for a given orientation with respect to B 0 .Davies ENDOR spectra were measured on MFM-520(Zn 0.995 Cu 0.005 ) at the B 0 positions indicated in Figure S18 (Supporting Information), corresponding to orientations along the molecular y axis (along "g ∥ "), and in the molecular xz ("g ⊥ ") plane with the axes as defined in Figure 1d.Along g ∥ the simplest spectra are observed, with four well-defined peaks centered at ca. 15 MHz (Figure 1e).In the g ┴ plane, more complex spectra are observed as all orientations in the xz plane are being selected.Four of these almost coincide with those measured at g, implying a near axial set of 14 N hyperfine and quadrupole parameters where the unique axis is within the xz plane.Because the 14 N…Cu vector lies on the C 2 axis with C 2v site symmetry, the 14 N hyperfine and quadrupole matrices (A N and P, respectively) must have their principal axes coincident with each other and those of g and A Cu .The data can be simulated with |A x N | = 38.0,|A y N | = 30.0and |A z N | = 31.5MHz, with a nuclear quadrupole coupling constant e 2 Qq/h = −4.4MHz and asymmetry parameter  = 0 corresponding to the quadrupole matrix principal values P x = −2.2,P x = 1.1 and P z = 1.1 MHz (Figure 1e).The large component of the hyperfine is coincident with the large component of the quadrupole interaction, along g x , i.e., along the direction of the Cu•••N bond (Figure 1e, bottom).
Quadrupole parameters are sensitive to the electric field gradient at the nucleus.NQR studies on pyridyl adducts with Lewis acids (LA) confirm the largest component to be along the direction of the lone pair of electrons at the 14 N atom, i.e., along the N…LA axis. [31]This has been confirmed in Cu…pyridyl complexes studied by orientation-selective ENDOR, [32][33][34][35][36] and is what we observe here for MFM-520(Zn 0.995 Cu 0.005 ).[34] For example, planar [Cu(chelidamate)(dmf)] (dmf = N,Ndimethlyformamide) is an interesting comparison to the Cu site in MFM-520(Zn 0.995 Cu 0.005 ) because chelidamate is a derivative of 2,6-dipicolinate and has a value of e 2 Qq/h = −2.8MHz. [33]In the framework of the Townes-Dailey model [31] a smaller |e 2 Qq/h| implies a greater extent of electron donation from the 14 N donor (sp 2 lone pair) orbital to the Lewis acid (LA), i.e., a stronger N…LA bond. [30]Hence, the larger |e 2 Qq/h| value here would suggest a weaker Cu…N bond.A weaker N→Cu -donation could also be described as smaller Cu "hole"→N transfer, and would be expected to give rise to a smaller 14 N hyperfine coupling.This is consistent with the 14 N hyperfine values for MFM-520(Zn 0.995 Cu 0.005 ) that are smaller than for other related complexes (Table S6, Supporting Information). [33]A crude calculation of the N 2p x spin density can be obtained from: where c 2 is the spin density, and P p is the electron-nuclear dipolar coupling parameter for N 2p orbitals (138.8MHz). [26]This gives c 2 = 4.4% for the Cu ion in MFM-520(Zn 0.995 Cu 0.005 ), lower than the 7.3% calculated from the hyperfine parameters for [Cu(chelidamte)(dmf)] which, as discussed above, is consistent with greater N2p→Cu donation in the latter.The isotropic part of the hyperfine is also smaller, but analysis of this in terms of valence s-orbital density is more uncertain due to core polarisation effects.The small 14 N hyperfine couplings and value of c 2 illustrate that the Cu─N bond in MFM-520(Zn 0.995 Cu 0.005 ) is relatively weak compared to other Cu─N coordinated molecules.Orientation selective ENDOR spectra centered at the proton Larmor frequency contain features of several doublets.Spectra collected at g x,z are dominated by frequencies of ca. 4 MHz, while spectra at g y are dominated by frequencies of ca. 2 MHz (Figure 1f).The 1 H hyperfine matrix (A) at each proton atom includes contributions from an isotropic hyperfine interaction, A Hiso , and point dipole interactions (A dip ).We calculated the point-dipole interactions (A dip ) for the nearest two sets of protons, the two protons of the pyridyl bound to the Cu (Cu•••H1, ca.4.9 Å), and those on the nearest pyridyl moieties in the orthogonal layer that bind to Cu via the carboxylates (Cu•••H2, ca.4.5 Å) (Figure S19 and Table S5, Supporting Information).Calculated spectra based on these parameters fail to reproduce the experimental spectra (Figure S20, Supporting Information).However, addition of an isotropic contribution of a H1 = 1.85 MHz and a H2 = 0.2 MHz gives good agreement (Figure 1f), consistent with a leaking of spin density into the bound pyridyl.The positive isotropic hyperfine coupling constant occurs when the metal and the ligand are bonded covalently, and this isotropic value is dominated by the unpaired electron transfer from the metal into the  system of the ligands. [32]
In situ EPR spectroscopy was used to track the changes at Cu 2+ before and after NO 2 loading.On NO 2 loading, similar variable temperature CW EPR spectra are observed (Figure 3a), with narrowing of the spectra on cooling.At low temperatures, two distinct Cu 2+ signals are observed (Figure 3b), both with similar parameters to those for the unloaded material.The g ∥ regions overlap, but with sufficient resolution at Q-band to define g ∥ and A ∥ (Table 1).These differences indicate slight changes of the Cu 2+ environment, consistent with a change in structural modulation or direct interaction between Cu 2+ and NO 2 molecules.Furthermore, an additional weak signal of monomeric NO 2 was observed at low temperature (Figure 3c), consistent with the N•••N distances from X-ray diffraction results (Figure 2b,c).The EPR parameters for the NO 2 signal are comparable with those of adsorbed NO 2 in other porous media (Table 1; Table S7, Supporting Information), [3] suggesting that the introduction of Cu 2+ centers into the framework plays an important role in tuning the dimerization of NO 2 molecules via formation of specific modulated host-guest interactions.
The introduction of NO 2 also results to changes to 1 H and 14 N hyperfine, measured at the Cu 2+ signal in the Davis ENDOR spectra (Figure 3d,e).The 14 N hyperfine is slightly increased, suggesting a slightly stronger Cu•••N bond upon loading of NO 2 ; a consistent change was also observed for the quadrupole parameter that decreases from 4.4 to 4.0 upon loading of NO 2 (Table S5, Supporting Information).Comparing the 1 H signal before and after NO 2 loading, we observe subtle changes in the shape of spectra, as well as a broadening in the g y orientation.Such effects can result from the distribution of molecular conformations due to structural heterogeneities, commonly observed in biological systems. [37,38]This is very likely to occur in NO 2 -loaded MFM-520(Zn 0.995 Cu 0.005 ) because of the structural heterogeneities from the incommensurate modulation of the structure.ENDOR spectra at the field position of the NO 2 signal are consistent with nearest interactions with framework protons (based on a point dipole O 2 N•••H interaction) of 3.2 Å, which is in excellent agreement with the N•••H distance of 3.13-3.41Å from the X-ray diffraction analysis (Figure 1f; Figure S8a, Supporting Information).

Conclusion
The geometry and local environment of Cu 2+ in MFM-520(Zn 0.995 Cu 0.005 ) was probed by CW and pulsed EPR spectroscopy, and the unusual form of the CW EPR spectrum rationalized by proposing a reverse Berry distorted coordination geometry, the first such example in a periodic material. 14N and 1 H hyperfine couplings observed in the pulsed ENDOR spectra further confirm the successful doping of Cu 2+ into the Zn 2+ framework sites with even distribution.Introduction of Cu 2+ to MFM-520(Zn) gives further structural incommensurate modulation and changes the degree of dimerization of adsorbed NO 2 molecules.Enhanced adsorption of NO 2 has been observed upon doping of Cu 2+ centers in MFM-520(Zn 1-x Cu x ) materials from 4.52 to 5.02 mmol g −1 and loading of NO 2 generates two Cu 2+ signals with slightly different spin-Hamilton parameters confirming changes of electron distribution of Cu 2+ sites on loading of NO 2 .Monomeric NO 2 molecules are observed spectroscopically in NO 2 -loaded MFM-520(Cu 0.005 Zn 0.995 ) even at low temperature (T = 5 K), in contrast to the parent MFM-520(Zn) material.
[CCDC 2259399 and 2259252 contain the supplementary crystallographic data for this paper.These data can be obtained free of charge from The Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/data_request/cif .] 6 6 -type open framework with different modulation vectors [0.1240(3) (a* + b*) + 0.5c*] compared to MFM-520(Zn) (Figure S7 and Table S3, Supporting Information), derived by partial replacement of Zn 2+ by Cu 2+ and variations in the distortion of the [Zn 0.95 Cu 0.05 O 4 N] moieties.Bond distances of M─N1, M─O1, and M─O2 (M = Zn 0.95 Cu 0.05 ) lie in ranges of 1.92-2.09,1.88-2.02,and 2.11-2.30Å, respectively, accompanied by modulation of the electron density around the metal center with the formal oxidation state ranging from 1.79(3) to 2.19(1), as determined by bond valence sum (BVS) calculations (Figure S9, Supporting Information).

Figure 2 .
Figure 2. a) View of the crystal structures of NO 2 -loaded MFM-520(Zn 0.99 Cu 0.01 ).[8 × 8 × 2] lattice indicates that the unit cell of the modulated structure is eight times that of the average structure along the a and b axes and twice along the c axis.b) Expanded view of (a), with different O 2 N•••NO 2 distances highlighted; c) Maps of the bond valence sum (BVS) analyses of the N•••N distance of N 2 O 4 guest molecules as a function of the modulation vector u and t in NO 2 -loaded MFM-520(Zn 0.99 Cu 0.01 ).The geometries of the N 2 O 4 for the lowest and highest (highlighted in deep and light blue, respectively) BVS analysis maps are illustrated above the contour maps.Bond distances are in Å; d) NO 2 adsorption isotherms of desolvated MFM-520(Zn 0.95 Cu 0.05 ) and MFM-520(Zn) at 298 K.

Table 1 .
14in-Hamiltonian parameter set extracted from Q-band CW and ENDOR EPR spectra at 5 K.  n in the strong coupling regime when |A| > 2 n , where A and  n are the hyperfine coupling and the nuclear Larmor frequency, respectively.A 14 N splitting of ca. 30 MHz is observed in the CW EPR spectra suggesting that the14N peaks ( N = 1.1 MHz at B 0 = 350 mT) overlap with 1 H peaks ( H = 14.9 MHz at B 0 = 350 mT) at X-band (Figure