Understanding Cu(i) local environments in MOFs via63/65Cu NMR spectroscopy

The field of metal–organic frameworks (MOFs) includes a vast number of hybrid organic and inorganic porous materials with wide-ranging applications. In particular, the Cu(i) ion exhibits rich coordination chemistry in MOFs and can exist in two-, three-, and four-coordinate environments, which gives rise to many structural motifs and potential applications. Direct characterization of the structurally and chemically important Cu(i) local environments is essential for understanding the sources of specific MOF properties. For the first time, 63/65Cu solid-state NMR has been used to investigate a variety of Cu(i) sites and local coordination geometries in Cu MOFs. This approach is a sensitive probe of the local Cu environment, particularly when combined with density functional theory calculations. A wide range of structurally-dependent 63/65Cu NMR parameters have been observed, including 65Cu quadrupolar coupling constants ranging from 18.8 to 74.8 MHz. Using the data from this and prior studies, a correlation between Cu quadrupolar coupling constants, Cu coordination number, and local Cu coordination geometry has been established. Links between DFT-calculated and experimental Cu NMR parameters are also presented. Several case studies illustrate the feasibility of 63/65Cu NMR for investigating and resolving inequivalent Cu sites, monitoring MOF phase changes, interrogating the Cu oxidation number, and characterizing the product of a MOF chemical reaction involving Cu(ii) reduction to Cu(i). A convenient avenue to acquire accurate 65Cu NMR spectra and NMR parameters from Cu(i) MOFs at a widely accessible magnetic field of 9.4 T is described, with a demonstrated practical application for tracking Cu(i) coordination evolution during MOF anion exchange. This work showcases the power of 63/65Cu solid-state NMR spectroscopy and DFT calculations for molecular-level characterization of Cu(i) centers in MOFs, along with the potential of this protocol for investigating a wide variety of MOF structural changes and processes important for practical applications. This approach has broad applications for examining Cu(i) centers in other weight-dilute systems.


Introduction
Metal-organic frameworks (MOFs) are porous crystalline materials composed of organic and inorganic components, arranged in a motif that features metal cations or metal-inorganic clusters connected by organic linkers. 1,24][5][6] The metal-centered entities are typically referred to as secondary building units (SBUs); the SBU composition and coordination can be tailored to achieve desired MOF topologies and properties. 7,8opper(I) is a versatile metal that can adopt a multitude of coordination states; Cu(I) applications range from serving as active sites in catalysts to playing an integral role in proteins and biology.From a materials perspective, Cu(I) has the ability to form a wide variety of cluster-based compounds and MOFs. 9he copper(I) halide clusters Cu x X y (X = Cl, Br, I) exhibit unique luminescent behaviors. 91][12] Cu(I) centers in MOFs can adopt three distinct local coordination geometries: two-coordinate linear, three-coordinate trigonal planar, and four-coordinate tetrahedral.][14][15][16][17][18] Structural characterization is critical to understanding the molecular-level origins of unique MOF properties.The coordination state, geometry, local environment, and position of Cu(I) sites in the SBU and MOF inuence the properties and applications of the resulting material.Cu(I) is generally regarded as a "spectroscopically silent" target that cannot be probed through traditional routes such as EPR and UV-vis spectroscopies, which makes characterization very challenging.[53][54][55][56] 63/65 Cu solid-state NMR is a promising untapped avenue for probing the local metal structure and unravelling structure-property relationships in Cu(I) MOFs.
Copper has two NMR active isotopes, 63 Cu and 65 Cu, which are both quadrupolar nuclei with a spin number (I) of 3/2.The electric quadrupolar moments (Q) of both nuclei are relatively high, where Q( 63 Cu) = −0.220and Q( 65 Cu) = −0.204barn. 57,58The natural abundance of 63 Cu is 69.2% and 65 Cu is 30.8%, 59yet 65 Cu is generally the preferred option for solid-state NMR in systems where sensitivity is not an issue due to the smaller Q and higher gyromagnetic ratio (g, where g( 65 Cu) = 7.6104 × 10 7 rad T −1 s −1 and g( 63 Cu) = 7.1088 × 10 7 rad T −1 s −1 ). 60In situations when the Cu(I) density within a material is low (e.g., catalytic applications), the signicantly more abundant 63 Cu isotope may be a more prudent choice for NMR experiments.The sizeable Q of both isotopes renders 63/65 Cu NMR spectra very broad when Cu does not reside in a local environment of high symmetry, making spectral acquisition challenging.The same anisotropic quadrupolar and chemical shi interactions that give rise to broadened and complicated 63/65 Cu NMR spectra also encode a wealth of information regarding the local Cu environment.
The 63/65 Cu NMR signals of many materials are broadened into the "ultra-wideline" frequency regime 61 and are difficult to acquire, which has limited the use of 63/65 Cu NMR for practical applications.Non-spinning (i.e., static) experiments are wellsuited for acquiring ultra-wideline 63/65 Cu NMR spectra. 35,51hallenges associated with 63/65 Cu NMR have been partially mitigated through the use of increasingly accessible high magnetic elds (i.e., >18.8 T). 19,35,53,54 The second order quadrupolar interaction (QI) that broadens central transition (+1/2 4 −1/2) 63/65 Cu NMR spectra is inversely proportional to the magnetic eld strength, which results in narrower signals at higher elds.Higher magnetic elds also enhance the population difference between the +1/2 and −1/2 spin states, increasing NMR sensitivity.While 1 H- 63 Cu RESPDOR NMR experiments have been used to examine a Cu(I) MOF, 62 there have been no reports regarding direct Cu(I) NMR of MOFs.
In this work, we report a 63/65 Cu NMR study of Cu(I) MOFs featuring copper sites in various two-, three-and fourcoordinate environments.The 63/65 Cu NMR parameters quantied from 21.1 T data reveal key information regarding local symmetry and coordination about Cu.We use data from this work and prior studies to illustrate how the Cu quadrupolar coupling constant (C Q ) values are highly dependent on the coordination number and geometric conguration of Cu(I) in MOFs, and present a general scale to guide researchers in determining the Cu(I) coordination number from C Q (Cu) values in MOFs and many other compounds.This experimental approach can be employed to monitor the structural evolution of MOFs, such as phase transitions, via effects on Cu(I) local environments.In favorable situations, the resolution of ultrawideline 63/65 Cu solid-state NMR spectra is sufficient to resolve signals from multiple Cu(I) sites. 35Practical applications of 63/65 Cu NMR are explored with experiments on a Cu(I)/Cu(II) mixed valence MOF featuring paramagnetic metal centers that lacks single crystal X-ray diffraction (XRD) data.A comprehensive examination of density functional theory (DFT) calculations and associated geometry optimizations have been performed to better understand the structural origins of experimental electric eld gradient (EFG) tensors, along with any discrepancies between calculated and experimental NMR parameters.To nish, we show that 65 Cu solid-state NMR spectra of Cu(I) MOFs can be successfully acquired at a more accessible lower magnetic eld of 9.4 T with sufficient resolution to accurately extract Cu NMR parameters.The practical applications of this concept are illustrated by using 65 Cu NMR at 9.4 T to elucidate local structural transformations associated with anion exchange in Cu MOFs.The 63/65 Cu solid-state NMR approach in this work demonstrates a promising investigative route for the characterization of Cu(I)-based MOFs and their derivative materials, whether the crystal structure is known or unknown.

MOFs with four-coordinate Cu(I) sites
Representative 63/65 Cu NMR of tetrahedral Cu(I) sites in MOFs: [CuCl(bpy)] and [CuI(bpy)].The most common bonding conguration for Cu(I) in MOFs is in a four-coordinate tetrahedral or distorted tetrahedral fashion.The Cu-X-bpy series of MOFs (X = Cl, Br, I, bpy = 4,4 0 -bipyridine) have exhibited potential applications in photocatalytic hydrogen production. 10CuCl(bpy)] is a neutral three-dimensional framework with open pores measuring ca. 2 × 4 Å.This compound crystallizes in the I4 1 /acd space group (Fig. 1(a)). 16Cu(I) resides in a slightly distorted tetrahedral environment in which adjacent Cu(I) centers are bound to two 4,4 0 -bpy ligands and bridged by two m 2 -Cl atoms (Fig. 1(a)).The 63/65 Cu static NMR spectra of [CuCl(bpy)] at 21.1 T are shown in Fig. 1(e and i).Both the 63 Cu and 65 Cu NMR spectra exhibit typical QI-dominated powder patterns that can be simulated using a single signal arising from one unique Cu site, which is consistent with the XRD structure. 16The metallic Cu(0) signal denoted by an asterisk (*) originates from the NMR probe, rather than the sample (Fig. S3 †).There is also an additional check on the C Q obtained from simulating the spectra at 21.1 T; the condition C Q ( 63 Cu)/C Q ( 65 Cu) = 1.078 must be satised, owing to the ratio between the respective nuclear Q values.Despite the dominance of the QI, there are ne features in the 63/65 Cu NMR spectra that cannot be simulated using only quadrupolar parameters, which unambiguously conrms that Cu chemical shi anisotropy (CSA) must be present.The 63/65 Cu NMR parameters (Table 1) were determined to be C Q ( 65 63,64 With this in mind, the ability of 63/65 Cu NMR to investigate phase transitions in MOFs was explored using [CuI(bpy)].
The three-dimensional porous [CuI(bpy)] MOF is transformed to two-dimensional [Cu 2 I 2 (bpy)] with heat (Fig. 1(b, c) and S4 †). 65[CuI(bpy)] crystallizes in the I4 1 /acd space group.The single unique Cu(I) center resides in a CuN 2 I 2 slightly distorted tetrahedral local environment, which involves bonding to two N atoms from separate 4,4 0 -bpy ligands along with two bridging In (e-h), the experimental ("Exp.,"blue) and simulated ("Sim.,"red) 65 Cu static NMR spectra of [CuCl(bpy)], [CuI(bpy)], [Cu 2 I 2 (bpy)] and [Cu 2 I 2 (pyz)] at 21.1 T are shown, with the corresponding 63 Cu NMR spectra and simulations in (i-l).The asterisk (*) denotes a signal from metallic copper (Cu 0 ) and the pound (#) marks a signal from probe background.The plus symbol (+) marks a resonance arising from residual CuI at 0 ppm after thermal treatment.A background 23 Na signal is also noted in (j).The definitions of *, #, and + also apply to all other figures in this work.] is a two-dimensional layered material that crystallizes in the P 1 space group with layer stacking along the crystallographic b axis./65 Cu solid-state NMR experiments were performed to investigate the local structure at Cu in both [CuI(bpy)] and the [Cu 2 I 2 (bpy)] product from thermal treatment (Fig. 1(b and c)).These MOFs give rise to well-dened 63/65 Cu NMR powder patterns, which are dominated by the QI but also inuenced by CSA, and are both indicative of one unique Cu site.The NMR spectra of [CuI(bpy)] and [Cu 2 I 2 (bpy)] are visually distinct and yield different 63/65 Cu NMR parameters (Table 1).The C Q ( 65 Cu) value of 28.In a manner similar to [CuI(bpy)], the three-dimensional [CuCl(bpy)] MOF can also undergo a transformation to the two-dimensional [Cu 2 Cl 2 (bpy)] MOF upon thermal treatment.The 63/65 Cu NMR spectra of these two MOFs (Fig. S5 † .The reticular synthesis of MOFs from specically selected metal centers and organic linkers is an active eld of research, since the pore size and eventual material properties can be controlled to a signicant degree.Adjustment of the linker length while retaining key binding functional groups has proven an effective avenue to modify the pore size and specic surface area without changing the MOF topology. 66In the previous case, 63/65   .In many MOFs there are multiple unique metal sites, which can oen be challenging to characterize and distinguish using NMR techniques.Higherresolution solid-state NMR techniques (e.g., MAS, MQMAS) are not applicable due to the large quadrupolar interactions and broad lineshapes in 63/65 Cu solid-state NMR; however, wideline NMR experiments on nuclei such as 35 Cl have demonstrated that it is possible to distinguish multiple inequivalent sites. 68,69n this section, we show that 63/65 Cu NMR can resolve separate resonances arising from several crystallographically inequivalent Cu sites in MOFs.Many MOFs containing Cu x I y clusters feature multiple unique Cu(I) sites and have luminescent properties.The luminescent [Cu 4 I 4 (DABCO) 2 ] MOF is composed of Cu 4 I 4 clusters along with 1,4-diazabicyclo[2.2.2]octane (DABCO) linkers (Fig. 2(a)). 12This material crystallizes in the P 4 /mcc space group and has three inequivalent Cu(I) sites in the Cu 4 I 4 unit, where the Cu sites are populated in the ratio Cu1 : Cu2 : Cu3 = 1 : 2 : 1.Each inequivalent Cu(I) site resides in a CuNI 3 distorted tetrahedral environment, with the three coordinated iodine atoms originating from the Cu 4 I 4 cluster and the nitrogen atom from a DABCO linker.The 63/65 Cu NMR spectra (Fig. 2(b and c)) are both >400 kHz broad at 21.1 T, with ne features that hint at overlapping Cu resonances.Simulations of experimental data Cu a Differences between the experimental and calculated values for both the CS tensor parameters and Euler angles are considerable due to the computational difficulties involved with calculating Cu CS tensor parameters.b The "Exp." label denotes experimental Cu NMR parameters obtained from best-t simulations of 65/63 Cu NMR spectra acquired at 21.1 T. c The "Calc." label denotes the NMR parameters obtained from plane-wave DFT calculations using the CASTEP soware package.A geometry optimization of all atoms in the reported crystal structure was performed before calculation of NMR parameters; see the Materials and Methods section for additional details.Please see Table S5 for additional calculations performed using dened cluster models.d The 65/63 Cu NMR spectra were simulated independently.The experimental C Q ( 63 Cu)/C Q ( 65 Cu) ratio of 1.080 was found to be very close to the accepted quadrupole moment ratio Q( 63 Cu)/Q( 65 Cu) of 1.078, 57 which gives additional condence to the simulated ts.
conrm that there are two narrower and overlapping Cu signals of higher intensity nested within a less intense, broader underlying signal; the NMR parameters obtained from simulations are summarized in Table 1.
With the NMR parameters successfully extracted, plane-wave DFT calculations were performed to assign 63/65 Cu resonances to crystallographic sites.The calculated 63/65 Cu NMR parameters (Table 1) indicate that Cu sites 1 and 2 should exhibit similar NMR parameters, including relatively smaller C Q ( 63/65 Cu) values, while site 3 should correspond to unique NMR parameters and a larger C Q ( 63/65 Cu).Accordingly, the two narrower components of the spectrum were assigned to Cu sites 1 and 2, with the broader signal corresponding to site 3. Given the similarities in C Q values between Cu sites 1 and 2, an alternate NMR parameter, such as h Q , must be used to distinguish between them.A careful examination of the le quadrupolar "horn" of the two narrower signals located between +150 and +250 kHz in the 63 Cu NMR spectra reveals signicant detail, which differentiates the Cu1 and Cu2 powder patterns based on h Q values.A comparison of local structural parameters between Cu2 and Cu1 shows that Cu2 has both the larger Cu-I bond length distribution and :N-Cu-I distribution of all Cu sites; 12 this combination reects a relatively lower axial symmetry in the Cu2 local environment and should result in a relatively higher h Q value.Cu2 is thus assigned to the signal with h Q = 0.14(4) and Cu1 is assigned to the signal with a smaller h Q of 0.09(3).This assignment is also consistent with DFT calculations (Table 1, where  72 The very slight departure from perfect C 3 rotational symmetry indicated by the h Q value of 0.02 can be traced to one of the Cu-bonded iodine atoms, which lies slightly out of a truly C 3 symmetrical ligand arrangement.The dominance of the QI on 63/65 Cu NMR spectral appearance, paired with the high signal-to-noise ratio, indicates that there is very little paramagnetic inuence on the Cu(I) NMR parameters.The lack of paramagnetic effects can be attributed to two reasons.5][76][77][78] Second, the distance between the Cu(I) and Cu(II) dimers is 7.05 Å, which is long enough that the Cu(I) spin energy levels are only perturbed to a minor degree by any paramagnetic interaction.
In addition to the direct synthesis of Cu (I/II) MOFs, a postsynthetic approach to Cu(I/II) MOFs affords alternate avenues for tuning MOF properties; however, it is difficult or impossible to obtain diffraction-caliber single crystals of product using this approach.Potential applications for 63/65 Cu solid-state NMR in the characterization of post-reduction Cu(I)-containing MOFs were explored by examining the case of Cu 2 BDC synthesis from the reduction of CuBDC.Cu(II) sites in the two-dimensional CuBDC (BDC, 1,4-benzendicarboxylic acid) MOF can be partially reduced with L-ascorbic acid (LA acid) via postsynthetic modication to introduce Cu(I) sites, forming a three-dimensional Cu 2 BDC MOF (Fig. 4(a)). 14Powder XRD (Fig. S9 † The successful reduction of Cu(II) centers to Cu(I) was conrmed by Cu 2p 3/2 X-ray photoelectron spectroscopy (XPS, Fig. S9 †) and X-band EPR (Fig. S10 †).The 63/65 Cu NMR spectra of Cu 2 BDC (Fig. 4(b and c)) features a QI-dominated NMR powder pattern that yields C Q ( 65 Cu) = 53.0(3)MHz and h Q = 0.22(3); the well-dened spectrum with relatively sharp features is also indicative of a highly ordered local structure.Although Cu(I) is fourcoordinate in this system, the C Q value is much larger than those of other four-coordinate Cu(I) centers previously discussed.This discrepancy arises from the seesaw local geometry about Cu(I) in Cu 2 BDC, which is a much more signicant deviation from tetrahedral symmetry versus previous examples of distorted tetrahedral geometry.
A CSA span value of 1800 ppm was necessary to achieve good agreement between the simulated and experimental 63/65 Cu NMR spectra of Cu 2 BDC (Fig. 4(b and c)), owing to the hyperne interaction.0][81][82][83] The corresponding hyperne interactions between Cu(II) unpaired electrons and Cu(I) nuclei leads to very large 63/ 65 Cu NMR span values.We have performed localized molecular orbital calculations on Cu 2 BDC, which revealed that the unpaired electrons of Cu(II) are indeed able to sample regions proximate to Cu(I) (Fig. S11 †); this nding, together with the unremarkable Cu chemical shi of 200 ppm, indicates that an electron delocalization effect is present rather than a spinpolarization effect. 79,82,83Cu and (c) 63 Cu static NMR spectra of Cu 2 BDC at 21.1 T are shown, along with simulations incorporating CSA effects (red trace) and neglecting CSA effects (black trace).Note the effects of CSA on the central spectral discontinuity.The signal from 23 Na background is also indicated in the 63 Cu NMR spectrum but truncated for clarity.

Chemical Science
Edge Article A successful simulation of all spectral features is challenging due to the multiple Cu powder patterns, despite the single Cu(I) site present in this MOF.Several spectral simulation strategies were explored, but only one produced a satisfactory t (Fig. S12 †), which is discussed below.The extremely broad underlying powder pattern with corresponding quadrupolar horns marked "Cu1" in Fig. 5, which has the highest integrated ratio of ca.80%, originates from the Cu(bpy  35,51 This assignment is also supported by DFT calculations, which yielded C Q,calc ( 65 Cu) = 74.5 MHz and h Q,calc = 0.17 (Table 1, Fig. 5(c and d)).Another narrower resonance labelled Cu2, with an integrated area ratio of ca.18%, is assigned to a side product; the NMR parameters are reported in Table 1 for reference.The experimental PXRD pattern of Cu(bpy) 1.5 NO 3 -$1.25H 2 O in Fig. S2 † agrees well with the pattern simulated from the reported crystal structure, yet the experimental diffractogram also exhibits some additional reections at low angles that are attributed to the Cu(I) impurity.There is also a trace amount of an unidentied impurity accounting for ca.2% of total spectral intensity that is labelled with the "&" character and assigned to a Cu3 species.While both [Cu 3 (4hypymca) 3 ] and Cu(bpy) 1.5 NO 3 $1.25H 2 O feature Cu(I) bound to three nitrogen atoms, the h Q value is signicantly higher in [Cu 3 (4hypymca) 3 ] (Table 1).Cu(bpy) 1.5 NO 3 $1.25H 2 O features Cu coordinated to three N atoms of pyridine groups, but the Cu center in [Cu 3 (4hypymca) 3 ] is connected to two pyridine-based N atoms and one nitrile N atom, which corresponds to decreased axial symmetry about Cu and a higher h Q value.The average trigonal distortion (d ¼ ð P 3 i¼1 jq i À 120 jÞ=3, where q i , i = 1, 2, 3 are the three   Al marks a signal from probe background.The signal from 23 Na background is also marked in the 63

MOFs with two-coordinate Cu(I) sites
Some MOFs feature Cu(I) in two-coordinate linear arrangements./65 Cu static solid-state NMR spectra of SLUG-22 at 21.1 T (Fig. 7(b and  c)) are extremely broad, owing to the low local symmetry of the linear two-coordinate environment at Cu and the correspondingly large C Q value.Simulations (Table 1) necessitated the use of CS parameters, but indicated only a single Cu(I) site with C Q ( 65 Cu) = 63.0(1.0)MHz and h Q = 0.34(2) was present; this contrasts with the reported crystal structure 18 that indicated there are two inequivalent Cu(I) sites.A careful examination of the 63/65 Cu NMR spectra reveals slightly broader features in the experimental spectra, which could be indicative of two nearly identical overlapping Cu powder patterns that cannot be resolved.Indeed, the crystal structure shows that the Cu centers reside in very similar local environments. 18Furthermore, Fig. 7 illustrates how the frequency of the central spectral discontinuity is affected by CSA, and in particular, the exceptionally large span value of 1500 ppm.In order to investigate the origins of the abnormally large Cu CSA in SLUG-22, EPR and XPS experiments were performed (Fig. S13 †), which indicated this material was largely free of Cu(II) or other paramagnetic impurities.The lack of any plausible hyperne interactions indicates the sizeable CSA in SLUG-22 likely arises from the linear two-coordinate local geometry at Cu. Large CSA spans have been recorded from other transition metal compounds in linear congurations (e.g., linear HgX 2 ). 85e found that cluster DFT calculations using the RHF method and 6-31++G**/6-311++G** basis sets were the most reliable avenue for calculating Cu span values (vide infra, Table S5 †); these particular calculations also predicted a substantial Cu CSA span value in SLUG-22 arising from the local linear coordination geometry at Cu.
There are more complicated MOFs featuring mixed coordinate Cu(I) local environments that can be examined using 63/ 65 Cu NMR.We investigated the [Cu 6 I 6 (DABCO) 2 ] framework, which contains four distinct Cu sites and produced a complicated Cu NMR spectrum that lacked clear singularities.The results and discussion regarding these experiments can be found in the ESI.†

DFT calculations of EFG tensors
To obtain further insight into experimental NMR results and the local Cu(I) environments, plane-wave DFT calculations using GIPAW methods were performed.7][88][89][90][91][92] In this section, we examine Cu(I) structural insights obtained from plane-wave DFT calculations of 63/65 Cu EFG tensors in MOFs.It should be noted that the XRD crystal structures were obtained at low temperatures while our Cu NMR experiments were performed at room temperature, which could lead to discrepancies between calculated and experimental Cu NMR parameters due to issues such as temperaturedependent unit cell dimensions and dynamics.To verify that temperature changes did not introduce signicant changes to Cu NMR parameters, low-temperature NMR experiments were performed on selected MOFs, which yielded Cu NMR spectra nearly identical to room temperature spectra (Fig. S15 †).This nding signied that, in this case, the experimental temperature did not play a signicant role in the accuracy of calculated NMR parameters.This result is not particularly surprising, as all the MOFs investigated in this work have rather rigid frameworks.
GIPAW DFT calculations on the Cu MOF systems were evaluated by rst examining the correlations between the calculated and experimental EFG tensor parameters in discrete Cu(I) coordination complexes with relevant Cu(I) local environments.The experimental Cu NMR data was taken from previous reports by Tang 35 and Yu, 56 using crystal structures obtained from the Cambridge Crystallographic Data Centre (CCDC).As shown in Fig. S16 and Appendix A of the ESI, † the calculated principal components of the Cu EFG tensor in this dataset exhibit a decreased G RMSE (the EFG distance metric expressing the deviation between experimental and computed EFG parameters; 93 see ESI †) aer geometry optimization, from 0.071 a.u.before to 0.053 a.u.aer.Aer geometry optimization, the slope of the respective plot is closer to 1, the y-intercept is reduced,  65 Cu and (c) 63 Cu static NMR spectra at 21.1 T are shown in blue, along with the red simulation which includes CSA effects, and the black trace which does not include CSA effects.Note the difference in the central "divot" spectral feature between the red (CSA) and black (no CSA) simulations; CSA is necessary to properly fit this feature.The asterisks (*) denote the signal from metallic copper (Cu 0 ) and the pound (#) marks the signal from probe background, which are truncated for clarity.The signal from 23 Na background is also marked in the 63 Cu NMR spectrum.and R 2 is higher, which all indicate a better agreement with experimental values.
The correlation between DFT-calculated and experimental Cu EFG tensor components jV kk j (k = 1, 2, 3) based on our current MOF dataset are shown in Fig. 8(a and b).Aer geometry optimization, the G RMSE of the MOF Cu dataset was calculated to be 0.119 a.u., which is signicantly larger than the 0.053 a.u. of the dataset constructed from prior reports.Bar plots of calculated versus experimental C Q and h Q values are shown in Fig. S25; † C Q depends on a single EFG tensor component (V 33 ) and is generally calculated quite accurately, while h Q calculations are dependent on all three EFG tensor components and are therefore less accurate.The SLUG-22 and Cu 3 (4hypymca) 3 MOFs are responsible for the increased G value in the MOF dataset (Fig. 8(c)).When excluding SLUG-22 and Cu 3 (4hypymca) 3 , the G RMSE value is a much more reasonable 0.047 a.u.These results indicate that further investigation of the local structure and geometry optimization in SLUG-22 and Cu 3 (4hypymca) 3 is warranted.
Discussion regarding the local structure of SLUG-22.There are three potential explanations for the signicant differences observed between the experimental and calculated Cu EFG parameters of SLUG-22.The rst possibility involves the presence of paramagnetic Cu(II) that could impact spectral appearance and inuence the accuracy of extracted Cu EFG parameters; however, XPS and EPR experiments (Fig. S13 †) indicated no detectable amounts of Cu(II) species.The second source is a potential temperature dependence of the SLUG-22 phase or unit cell dimensions, since the original XRD structure was obtained at low temperature and our NMR experiments were performed at room temperature.Hardware limitations prohibited low-temperature 63/65 Cu NMR experiments at 21.1 T, and 63/65 Cu WURST-CPMG NMR experiments at 9.4 T on these systems were not successful due to low T 2 values.As a surrogate, the 1 H- 13 C CP/MAS NMR spectra of SLUG-22 at 208 and 298 K were acquired and found to be quite similar (Fig. S18 †), suggesting that no signicant structural deviations or phase changes occur at lower temperatures.The third source is fundamental structural issues arising from the DFT geometry optimizations performed prior to EFG calculations, which was found to merit further investigation.
Four DFT optimization schemes of the SLUG-22 crystal structure were explored (Fig. 8(d)).All the geometry-optimized structures yielded lower SCF energies versus the XRD structure, yet the agreement between calculated and experimental Cu EFG tensor parameters using any of the calculation strategies did not show signicant improvement, which is puzzling.A more detailed examination of the experimental PXRD patterns in the original work describing SLUG-22 18 (Fig. S19(a) †) revealed that several intense reections expected at low angles from the reported single crystal structure are not apparent in the original experimental data, particularly the prominent reection at ca. 9°, while additional unexpected reections are present.Our geometry-optimized structures generated using a myriad of DFT-based approaches failed to improve the agreement between experimental and calculated XRD patterns (Fig. S19(b) †).Based on the considerable deviation between the experimental and calculated PXRD patterns, along with the inaccuracy of calculated EFG tensor parameters, it appears that the reported single crystal structure of SLUG-22 is incorrect in some manner, which illustrates another practical application of Cu NMR.
Discussion regarding the local structure of Cu 3 (4hypymca) 3 .As with SLUG-22, we observed no improvement in the agreement between calculated and experimental EFG tensor (d) The EFG distance in the optimized SLUG-22 structure is shown, as obtained after plane-wave DFT calculations using different geometry optimization approaches.The x-axis labels in (d) are as follows; XRD structure: no optimization; DFT: geometry optimization without optimization of unit cell dimensions; DFT-shape: optimization with unit cell using a fixed-shape constraint; DFT-volume: optimization with unit cell dimensions using a fixed-volume constraint; DFT-D2 and DFT-D3: optimization using the D2 and D3 dispersion corrections with an optimized damping parameter (Fig. S17 †). 86,93Note that there are two inequivalent but similar Cu sites in the reported XRD structure of SLUG-22.components of the Cu 3 (4hypymca) 3 MOF despite employing a variety of different geometry optimization strategies prior to NMR calculations (Fig. 9(a)).The most striking change in the local optimized geometry about Cu is a reduction of > 0.1 Å in the Cu-N bond length to the linker cyanide group.Variations in bond lengths and bond angles have a known inuence on EFG parameters. 94,95To investigate further, we systematically altered the Cu-N bond length and then calculated the EFG tensor parameters and relative energy of the system, as shown in Fig. 9(b).While the minimum energy is associated with a Cu-N bond distance of 1.95 Å, this distance leads to a considerable gap between calculated and experimental Cu EFG tensor parameters.In comparison, a Cu-N bond length of 2.20 Å maximizes the accuracy of the calculated EFG tensor parameters, but results in an unacceptably high system energy.In this case, the most likely situation is a Cu-N bond length that results in a mutual minimization of system energy and G, where the trends intersect in Fig. 9(b).This data suggests that the Cu-N bond length in the Cu 3 (4hypymca) 3 MOF crystal structure is slightly longer than the reported value of 2.00 Å.In addition, it is possible that plane-wave DFT calculations do not properly account for intermolecular interactions between the 2D sheets in this MOF, or there could be "slipping" of relative positions between the 2D MOF sheets that cannot be clearly identied via XRD studies.
Calculations of Cu EFG tensor orientations and CS tensor parameters.DFT calculations are known to reliably yield the orientation of the Cu EFG tensor, and the EFG tensor orientations for these Cu MOF systems along with a brief discussion are provided in Appendix B of ESI.† The calculated CS parameters are listed in Table 1 for reference, but we highlight that agreement between calculated and experimental CS values is rather poor (Fig. S26 in Appendix C †), and plane-wave DFT calculations evidently are not a reliable predictor of Cu CS parameters in MOFs at this time.While calculating EFG parameters only involves the electronic ground state of a system, CS calculations involve both the ground and excited states, which increases the computational complexity.For instance, Tang et al. performed CS calculations using different DFT basis sets and methods on discrete cluster models of Cu(I) compounds, 35 but only obtained partial agreement with experimental results depending on the particular approach.Calculating CS parameters for heavier atoms such as Cu is also challenging due to factors such as spin-orbit effects, 96 relativistic considerations, 97,98 and the many possible hybrid functionals that can be applicable. 99The relatively high uncertainty of experimental CS values (Table S6 †) as a result of multivariable tting may also inuence the agreement with calculated CS values; larger uncertainties in the experimental CS parameters will generally result in poorer agreement with accurately calculated values.
We also performed calculations on geometry-optimized cluster models.The results using several different methods and basis sets are listed in Table S5.† The CS span values calculated using RHF/6-31++G** and RHF/6-311++G** (Fig. S21 †) demonstrated better agreement with experimental span values when compared to plane-wave DFT calculations.In contrast, the calculated EFG tensors with all cluster models (Fig. S20 †) yielded poorer agreement with experimental values when compared to plane-wave DFT calculations (Fig. 8(b)).

Quadrupolar coupling constant and the Cu(I) coordination number
The observed quadrupolar coupling constant (C Q ) largely depends on the coordination number of Cu(I).A summary of the C Q ( 65 Cu) values in MOFs is illustrated in Fig. 10, along with relevant values in small metal-organic coordination compounds from previous reports. 35,51,56The C Q ( 65 There were no prior 63/65 Cu solid-state NMR reports of Cu(I) centers in a four-coordinate seesaw local geometry before this work; this environment appears to produce C Q values comparable to three-and two-coordinate Cu(I) arrangements.The compiled empirical results from this and prior studies in Fig. 10 provides a convenient and general NMR-based tool to estimate the coordinate state of Cu(I) in unknown environments across a variety of materials and compounds. 65Cu solid-state NMR at 9.4 T The Cu MOFs in this study were examined by 63/65 Cu NMR using a high magnetic eld of 21.1 T. Unfortunately, high elds are not always readily accessible, but there are several previous studies regarding 65 Cu ultra-wideline NMR at 9.4 T. 35,51 In MOFs, the Cu concentration is diluted, which poses an additional obstacle.
We set out to nd if 65 Cu solid-state NMR of Cu(I) MOFs was feasible at a more accessible eld of 9.4 T (i.e., n 0 ( 1 H) = 400 MHz), which would open up this technique to researchers across a broad swath of institutions.In addition, performing the Cu NMR experiments at different magnetic elds allows one to extract unambiguous CS and QI parameters along with the second-order quadrupolar isotropic shi.A major challenge at 9.4 T is spectral width; broadening from the second-order quadrupolar interaction is inversely proportional to B 0 , thus 63/65 Cu NMR spectra are spread across a signicantly larger frequency range at lower magnetic elds.To increase the signalto-noise ratio and reduce experimental times, the WURST-CPMG pulse sequence can be employed. 100The WURST-CPMG sequence yields NMR spectra composed of a series of spikelets that trace out the overall spectral manifold, rather than the smooth continuous lineshape obtained from solid echo experiments.A spikelet spectrum can be acquired signicantly faster than a solid echo spectrum.Using seven Cu(I) MOFs from this study as examples, we obtained 65 Cu NMR spectra at 9.4 T ranging from ca. 500 to 3000 kHz in breadth (Fig. 11).The 9.4 T data was then simulated independently in order to assess the reliability of these results against those obtained at 21.1 T. The 65 Cu NMR parameters obtained at 9.4 T (Table S6 †) were consistent with those obtained from 63/65 Cu experiments at 21.1 T, validating the accuracy of the extracted NMR parameters in Table 1.These ndings prove that 63/65 Cu NMR of Cu(I) MOFs and other Cu-dilute systems is experimentally viable at 9.4 T. The experimental times are listed in Table S4.† Applications of 63/65 Cu solid-state NMR for anion exchange reactions Anions are present in many MOFs to maintain charge balance with the cationic framework, which opens the door for versatile The (Cu 2 (SO 4 )(pyz) 2 (H 2 O) 2 ) MOF, termed 1, was synthesized using CuSO 4 $5H 2 O and pyrazine in hydrothermal conditions. 103,104The Cu(I) center in 1 is in a distorted tetrahedral CuN 2 O 2 local environment.Cu is connected to two pyrazine linkers, which form zig-zag one-dimensional chains that are bridged by sulfate ions, and Cu is also coordinated to water molecules that are oriented perpendicular to the 1D chains.The 65 Cu solid-state NMR spectrum at 9.4 T (Fig. 12) features a well-dened powder pattern exhibiting a C Q ( 65 Cu) of 25.2(2) MHz and a h Q of 0.54 (2), with the C Q value lying in the established range of four-coordinate Cu (Fig. 10).
While 1 is stable in air and water, this material quickly turns from dark red to orange when exposed to aqueous NaNO 3 solution, yielding 1@NO 3 − .A white insoluble precipitate is evident when BaCl 2 /HCl is added, indicating that a migration of SO 4 2− from 1 into solution has occurred.As a so acid, the Cu(I) ion prefers to coordinate with the so base of nitrogen rather than the hard base of oxygen, which explains the formation of a precipitate.The 65 Cu NMR spectrum of 1@NO 3 − is a much broader ca. 3 MHz, which corresponds to a C Q ( 65 Cu) of 46.5 (8) MHz and a h Q of 0.28(5) (Fig. 12).Using Fig. 10 as a guide, the signicant increase in C Q ( 65 Cu) from 1 to 1@NO 3 − indicates that the local coordination at Cu has changed from four-coordinate to two-or three-coordinate.When 1 was immersed in a NaClO 4 solution, the powder changed to an orange-yellow color, and the resulting product was termed 1@ClO 4

−
. The 65 Cu NMR spectrum of 1@ClO 4 − at 9.4 T has an impressive breadth of ca.7 MHz, with a C Q ( 65 Cu) of 67.0(6) MHz and h Q of 0.23 (7); the increase in C Q again indicates that Cu now resides in a two-or three-coordinate local environment.When 1 is exposed to aqueous NaCl, the original dark red color is retained, however, the 65 Cu NMR spectral breadth is considerably narrowed and the lineshape is altered in a distinct fashion.The 1@Cl − compound corresponds to a decreased C Q ( 65 Cu) of 15.5(4) MHz, which indicates that the four-coordinate tetrahedral geometry is preserved at Cu, along with a signicantly increased h Q of 0.98 (2).The well-dened 65 Cu NMR powder patterns of 1 aer exposure to NO 3 − , ClO 4 − and Cl − indicates that the local structure about Cu is relatively ordered in all instances.Using the estimated Cu coordination states obtained from the various 65 Cu NMR spectra of 1 and its derivatives in mind, a search of the CCDC database was performed for any structures potentially matching or similar to the anion-exchanged products obtained in this study.Three compounds were identied: {Cu(pyz)(NO 3 )} n which contains a two-coordinate CuN 2 moiety, 105 {Cu(pyz) 1.5 (ClO 4 )} n with a threecoordinate CuN 3 local structure, 105 and {CuCl(pyz)} n with a fourcoordinate CuCl 2 N 2 environment. 106The reported structures had been synthesized independently through solvothermal routes, and not via the anion-exchange approach we employed.The experimental PXRD patterns of 1@NO 3 − , 1@ClO 4 − , 1@Cl − , along with the calculated PXRD patterns of {Cu(pyz)(NO 3 )} n , {Cu(pyz) 1.5 (ClO 4 )} n , and {CuCl(pyz)} n from the reported solvothermal approaches are shown in Fig. S22.† The experimental PXRD pattern of 1@Cl − matches perfectly with the calculated pattern of {CuCl(pyz)} n , indicating the anionexchanged product 1@Cl − is identical to solvothermally synthesized {CuCl(pyz)} n.This also conrms that a fourcoordinate Cu(I) tetrahedral environment exists in 1@Cl − , as predicted from C Q ( 65 Cu) NMR values.In contrast, the PXRD patterns of anion-exchanged 1@NO 3 − and 1@ClO 4 − look similar to those of solvothermally synthesized {Cu(pyz)(NO 3 )} n and {Cu(pyz) 1.5 (ClO 4 )} n , but are not identical.It appears that the pairs of 1@NO 3 − and {Cu(pyz)(NO 3 )} n MOFs, and the 1@ClO 4 − , and {Cu(pyz) 1.5 (ClO 4 )} n MOFs, are of similar connectivities but reside in different crystal structures (i.e., space groups).The Cu(I) coordination numbers are two and three in {Cu(pyz)(NO 3 )} n and {Cu(pyz) 1.5 (ClO 4 )} n , respectively, which is consistent with expectations based on the experimental 1@NO 3 − and 1@ClO 4 − C Q ( 65 Cu) values.The Cu(I) center is generally considered to be a so acid, and preferentially binds with so base ligands such as N donors and halogen ions, rather than with hard bases such as O donors.In good agreement, we observed that the formation of 1@Cl − from 200 mg of 1 in a saturated aqueous solution concluded within ca. 30 min.
In the context of hard and so acids and bases, the cleavage of Cu(I)-O bonds to H 2 O and SO 4 2− and the formation of Cu(I)-Cl bonds to yield a CuCl 2 N 2 tetrahedral coordination environment in 1@Cl − is favorable and should proceed quickly.In a similar nding, the reaction of 1 with NO 3 − was also noted to conclude within 30 min; the zig-zag one-dimensional chains are sufficiently stable enough to exist without sulfate ions or coordinated water molecules, which then yields a two-coordinated Cu(I)N 2 conguration with NO 3 − solely as a charge balancing anion.In stark contrast, the formation of 1@ClO 4 − requires ca., the formation of Cu-N bonds to pyrazine linkers in a new trigonal geometry requires a much longer duration because perchlorate is a very weakly coordinating anion and is not directly bound to Cu(I).

Conclusions
A series of Cu(I)-containing MOFs featuring Cu sites in different coordination environments have been examined using 63/65 Cu ultra-wideline NMR and DFT calculations.The diversity of local environments of Cu(I) centers in MOFs leads to C Q ( 65 Cu) values ranging from 18.8 to 74.8 MHz, which are diagnostic of the local Cu coordination environment and geometry./65 Cu NMR spectroscopy provides direct evidence regarding the evolution of local Cu environments during MOF structural transformation processes.The sensitivity of this technique can also be exploited to monitor and characterize MOF phase transitions.Even in the challenging case of Cu(I/II) mixed valence MOFs, 63/65 Cu NMR spectra can be obtained, and are inuenced by paramagnetic interactions when Cu(I) is especially proximate to Cu(II).We have proven 63/65 Cu NMR can be performed within reasonable experimental times at a lower magnetic eld of 9.4 T despite the weight dilution of Cu(I) centers in MOFs.DFT-calculated 63/65 Cu EFG tensor parameters have been presented and rigorously compared with experimental values; calculations using geometry-optimized structures generally lead to better agreement with experimental results except for two instances, and the origins of these disagreements were explored.We have established a list of C Q (Cu) values from this and previous studies that permits estimation of local Cu coordination using only the C Q value, which is broadly applicable to many other Cu systems.This study highlights the versatility of 63/65 Cu solid-state NMR, extending its relevance beyond MOFs and towards any chemical systems containing either abundant or dilute Cu(I) centers, with applications in elds such as catalysis, surface chemistry, solar cells, and biochemistry.

Sample preparation
[16]18,65,107 All details regarding synthesis and non-NMR characterization of the Cu compounds can be found in the ESI.†

Solid-state NMR experiments
In general, 63/65 Cu NMR spectra were acquired under static conditions using the solid echo or WURST-CPMG (Wideband Uniform Rate Smooth Truncation-Carr Purcell Meiboom Gill) 100,108 pulse sequences.Solid echo experiments give rise to a smooth continuous lineshape, while WURST-CPMG experiments concentrate the signal into discrete spikelets that trace out the overall manifold of the powder pattern.Most of the ultra-wideline 63/65 Cu NMR spectra in this work were too broad to be acquired in a single experiment, which necessitated the use of the VOCS (variable-offset cumulative spectra) method. 109he VOCS approach involves acquiring several sub-spectra at evenly spaced transmitter offsets using otherwise identical experimental parameters, and then co-adding the subspectra together to obtain the total 63/65 Cu NMR spectrum.All 63 Cu and 65 Cu NMR spectra were referenced to solid CuCl at 0 ppm.
Solid-state NMR experiments at 21.1 T. Experiments at 21.1 T were conducted at the National Ultrahigh-eld NMR Facility for Solids in Ottawa, Canada using a Bruker Avance II spectrometer. 63Cu and 65 Cu NMR spectra were acquired using a home-built solenoid single-channel probe with a silver NMR coil (n 0 ( 63 Cu) = 238.73MHz, n 0 ( 65 Cu) = 255.74MHz).All 63/65 Cu NMR spectra were acquired using a solid-echo pulse sequence (90°-90°).The interpulse delay was set to 30 ms.Additional experimental parameters are listed in Tables S2 and S3.† Background signals of the probe and various sample containers are discussed in footnote b of Table S3 and shown in Fig. S3.† Solid-state NMR experiments at 9.4 T. All experimental parameters can be found in ESI and Table S4.†

7 ( 3 )
MHz in [CuI(bpy)] is reduced to 24.0(4) MHz in [Cu 2 I 2 (bpy)], with the increased symmetry at Cu attributed to the change from a CuN 2 I 2 to a CuNI 3 local environment.The h Q parameter is also sensitive to the phase change, falling from 0.50(4) in [CuI(bpy)] to 0.18(3) in [Cu 2 I 2 (bpy)], which is indicative of increased axial symmetry about the Cu center in [Cu 2 I 2 (bpy)].
(pyz)] (Fig. 1), and C Q ( 65 Cu) falls from 24.0(4) MHz in [Cu 2 I 2 (bpy)] to 18.8(4) MHz in [Cu 2 I 2 (pyz)].One reason for the decrease in C Q is the smaller bond angle and bond length distributions involving Cu, while another possibility lies in long-range inuences on the EFG that originate beyond the rst coordination sphere of Cu (i.e., the effect of different N-bound linker groups).A more detailed discussion can be found in the ESI.† 63/65 Cu solid-state NMR for resolving inequivalent Cu(I) sites in MOFs: [Cu 4 I 4 (DABCO) 2 ] For a more detailed discussion, please see the ESI.† 63/65 Cu NMR of mixed valence Cu(I/II) MOFs: {[Cu(I)] [Cu(II)(pdc)(H 2 O)]$1.5MeCN$H 2 O} n and Cu 2 BDC.There is a distinct family of mixed-valence MOFs that incorporate both Cu(I) and Cu(II) metal centers, which have a variety of diverse structures, unique electronic properties, and catalytic applications. 70,71The mixed valence {[Cu(I)][Cu(II)(pdc)(H 2 O)]$ 1.5MeCN$H 2 O} n (where pdc = pyridine-3,5-dicarboxylic acid) MOF was selected as a test compound to investigate if 63/65 Cu NMR could be used to probe materials containing both Cu(I) and Cu(II). 72The pdc linker contains both nitrogen and carboxylate groups, which can form MOFs with two separate types of metal nodes upon reaction with CuI.The {[Cu(I)][Cu(II)(pdc)(H 2 O)]$ 1.5MeCN$H 2 O} n MOF features a paddlewheel-type local structure incorporating Cu(II) centers, along with a Cu 4 I 4 cluster containing Cu(I) (Fig. 3(a)).Cu 2p 3/2 XPS spectra (Fig. S7(a) †) conrmed that both Cu(I) and Cu(II) centers were present in the sample.The single crystal XRD structure of {[Cu(I)][Cu(II)(pdc)(H 2 O)]$ 1.5MeCN$H 2 O} n 72 features Cu(I) sites in CuNI 3 distorted tetrahedral environments.The 63/65 Cu NMR spectra (Fig. 3(b and c)) were simulated using one Cu signal with a small amount of CSA (Fig. S8 †).Nevertheless, the prominent CuI signal in the middle of the spectrum introduces uncertainty in CSA quantication and may obscure additional Cu signals, although the likelihood of their presence is low.A C Q ( 65 Cu) value of 22.0(3) MHz and h Q of 0.02(2) were obtained from {[Cu(I)][Cu(II)(pdc)(H 2 O)]$1.5MeCN$H 2 O} n ; the near-zero h Q value is in good agreement with the high local rotational symmetry at Cu(I) indicated from the single crystal XRD structure.

Fig. 2 A
Fig. 2 A schematic illustration of the long-range structure of [Cu 4 -I 4 (DABCO) 2 ] along with the local structure about Cu is shown in (a).The experimental (b)63 Cu and (c)65 Cu static NMR spectra (blue), cumulative simulations (red), and individual Cu site simulations (black, purple, green) of [Cu 4 I 4 (DABCO) 2 ] at 21.1 T are also included.

Fig. 3
Fig. 3 (a) A schematic illustration of the long-range and local structure in the {[Cu(I)][Cu(II)(pdc)(H 2 O)]$1.5MeCN$H 2 O} n MOF, including the Cu(I) and Cu(II) clusters.The blue experimental and red simulated (b)65 Cu and (c)63 Cu static NMR spectra at 21.1 T are also shown.
) clearly indicates that Cu 2 BDC resides in a different phase than the parent CuBDC MOF, but further analysis of the Cu local environment is hampered by the difficulties in obtaining Cu 2 BDC single crystals.The parent CuBDC MOF contains a paddlewheel local structure about Cu, where each Cu(II) center is linked to four carboxylic groups from BDC linkers along with one water molecule, forming a stacked layered structure held together through intermolecular interactions.Aer LA-acid post-synthetic modication to produce the Cu 2 BDC MOF, half of the Cu(II) sites in the MOF were reduced to Cu(I) (Fig. 4(a)).The four-coordinate Cu(I) center in Cu 2 BDC resides in a local CuO 4 environment of seesaw geometry, with Cu(I) connected to two carboxylic oxygen atoms (O1, O2) from two BDC ligands, one oxygen atom (O3) of a water molecule, and one oxygen atom (O4) of a bridging OH group.
Fig. 4 (a) The reduction of CuBDC to Cu 2 BDC and the local environment of Cu(I) in CuBDC and Cu 2 BDC is pictured.(b) Blue experimental (b)65 Cu and (c)63 Cu static NMR spectra of Cu 2 BDC at 21.1 T are shown, along with simulations incorporating CSA effects (red trace) and neglecting CSA effects (black trace).Note the effects of CSA on the central spectral discontinuity.The signal from23 Na background is also indicated in the63 Cu NMR spectrum but truncated for clarity.

Fig. 5
Fig. 5 (a) The 3D framework structure and (b) local structure of the Cu(bpy) 1.5 NO 3 $1.25H 2 O MOF.The charge-balancing NO 3− anion and guest water molecules are omitted for clarity.(c)65 Cu and (d)63 Cu static NMR spectra of Cu(bpy) 1.5 NO 3 $1.25H 2 O at 21.1 T; the blue traces are experimental, black are simulated, and red are simulated using NMR parameters from DFT calculations.

Fig. 6
Fig. 6 (a) The long-range and local structure of Cu 3 (4hypymca) 3 .The experimental (blue) and simulated (red) 65 Cu static NMR spectra and 63 Cu static NMR spectra at 21.1 T are shown in (b and c), respectively.

Fig. 7
Fig. 7 (a) The 3D framework structure and local structure of the SLUG-22 MOF.The (b)65 Cu and (c)63 Cu static NMR spectra at 21.1 T are shown in blue, along with the red simulation which includes CSA effects, and the black trace which does not include CSA effects.Note the difference in the central "divot" spectral feature between the red (CSA) and black (no CSA) simulations; CSA is necessary to properly fit this feature.The asterisks (*) denote the signal from metallic copper (Cu 0 ) and the pound (#) marks the signal from probe background, which are truncated for clarity.The signal from23 Na background is also marked in the63 Cu NMR spectrum.

Fig. 8
Fig. 8 Relationships between the principal components (jV kk j, k = 1,2,3) of calculated and experimental 63/65 Cu EFG tensors in MOFs (a) before and (b) after geometry optimization.(c) The G values for all MOFs after geometry optimization.The G RMSE including the results from all MOFs is shown as a green dashed line, while the G RMSE excluding SLUG-22 and Cu 3 (4hypymca) 3 is shown as a red dashed line.(d)The EFG distance in the optimized SLUG-22 structure is shown, as obtained after plane-wave DFT calculations using different geometry optimization approaches.The x-axis labels in (d) are as follows; XRD structure: no optimization; DFT: geometry optimization without optimization of unit cell dimensions; DFT-shape: optimization with unit cell using a fixed-shape constraint; DFT-volume: optimization with unit cell dimensions using a fixed-volume constraint; DFT-D2 and DFT-D3: optimization using the D2 and D3 dispersion corrections with an optimized damping parameter (Fig.S17 †).86,93Note that there are two inequivalent but similar Cu sites in the reported XRD structure of SLUG-22.

Fig. 9
Fig. 9 (a) The EFG distance of optimized Cu 3 (4hypymca) 3 structure with different approaches, along with (b) the dependence of G on Cu-N bond length.

Fig. 10
Fig. 10 Copper C Q ( 65 Cu) values of Cu(I) MOFs, along with those previously reported for other Cu(I) compounds.
Cu) values of four-coordinate tetrahedral Cu(I) centers are generally <40 MHz.The C Q ( 65 Cu) values of three-coordinate Cu(I) range from 40 MHz to 80 MHz.Four-coordinate Cu(I) in a pseudo-three coordinate environment is correlated to C Q ( 65 Cu) values between 40 and 50 MHz, 51 which lies just between the bulk of four-and three-coordinate Cu environments. 63/65 Cu NMR reports on twocoordinate Cu(I) ions are not common; both the SLUG-22 MOF in this work and the previously reported small molecule ClCuP(2,4,6) 3 35 yielded C Q ( 65 Cu) values between 60 and 65 MHz.

Fig. 12
Fig. 12 The 65 Cu solid-state NMR spectra of 1 (Cu 2 (SO 4 )(pyz) 2 (H 2 O) 2 ) and associated products after exposure to different aqueous solutions, as measured at 9.4 T. The asterisk (*) denotes a signal from metallic copper (Cu 0 ) and the pound (#) indicates a resonance from a by-product containing Cu 2 O.
12 hours of reaction time.Aer the cleavage of Cu-O bonds to H 2 O and SO 4 2−

Table 1
Experimental and calculated a 63/65 Cu NMR parameters ) are distinct and diagnostic of the phase change.While the C Q (Cu) values are very similar in both forms, h Q changes from 0.45(3) in [CuCl(bpy)] to 0.25(2)in [Cu 2 Cl 2 (bpy)] (Table1), producing a clear spectral difference indicative of a signicant increase in local axial symmetry.The results indicate that63/65Cu NMR is a viable spectroscopic route for tracking phase changes in Cu MOFs.