Continuous Wave Single Crystal Electron ... - ACS Publications

Article pubs.acs.org/JPCC

Continuous-Wave Single-Crystal Electron Paramagnetic Resonance of Adsorption of Gases to Cupric Ions in the Zn(II)-Doped Porous Coordination Polymer Cu2.965Zn0.035(btc)2 Stefan Friedlan̈ der,† Petko St. Petkov,‡ Felix Bolling,† Anastasia Kultaeva,† Winfried Böhlmann,§ Oleg Ovchar,∥ Anatolii G. Belous,∥ Thomas Heine,‡ and Andreas Pöppl*,† †

Abteilung für Magnetische Resonanz Komplexer Quantenfestkörper, Fakultät für Physik und Geowissenschaften, Universität Leipzig, Linnéstrasse 5, 04103 Leipzig, Germany ‡ Wilhelm-Ostwald-Institut für Physikalische und Theoretische Chemie, Fakultät für Chemie und Mineralogie, Universität Leipzig, Linnéstrasse 2, 04103 Leipzig, Germany § Abteilung für Supraleitung und Magnetismus, Fakultät für Physik und Geowissenschaften, Universität Leipzig, Linnéstrasse 5, 04103 Leipzig, Germany ∥ Institute of General and Inorganic Chemistry, National Academy of Sciences of Ukraine, Academian Palladin Avenue 32-34, 03680 Kyiv, Ukraine S Supporting Information *

ABSTRACT: Continuous-wave X-band electron paramagnetic resonance with dielectric resonators has successfully been applied to small single crystals of the metal−organic framework HKUST-1 and Cu2.965Zn0.035(btc)2 to investigate the structure of paddle-wheel building blocks with pure Cu/Cu and mixed Cu/Zn pairs. The local paramagnetic Cu2+ ion probes were used to identify the magnetic g and A tensor orientations with respect to the crystal axes. We were able to monitor changes in these tensor orientations by EPR at gas adsorption on MOFs for the first time. We explored the spectral simulations of the spin Hamilton parameters of the single crystals and found results similar to those in previous studies of powder samples, but moreover, the tensor orientations are influenced upon gas adsorption, which is represented by a distinct line broadening effect in the angular resolved single-crystal EPR spectra. The as-synthesized, dehydrated, carbon dioxide-adsorbed, carbon monoxide-adsorbed, methanol-adsorbed, and reactivated states have been analyzed to reveal the magnetic tensor orientations, and the direct coordination of the adsorbed gas to the Cu2+ ions along with consistent, corresponding DFT calculations allows us to predict an improved model for the mixed paddle-wheel structure upon the adsorption of gases to a paddle-wheel based on perturbations of the g and A principal axis orientations. Additionally, we analyzed a reversibly occurring background signal observable not only in Cu2.965Zn0.035(btc)2 but also in pure Cu3(btc)2 at very low temperatures.

■

INTRODUCTION

(DFT) calculations allow us to monitor structural changes upon gas adsorption, as the large variety of studies on MOFs shows.7−11 A well-studied MOF is Cu3(btc)2 (btc = benzene-1,3,5tricarboxylate), also known as HKUST-1.12,13 It is based on paddle-wheel (PW) secondary building units with two Cu2+ ions and forms a regular three-dimensional network with defined channels and cages. The MOF was among the first commercially available representatives of these hybrid materials and its high porosity found in numerous studies set benchmarks for adsorption capabilities of MOFs and the effectiveness of

Over the last decades metal−organic framework (MOF) compounds have attracted considerable attraction as a discrete class of porous solids.1 MOFs are formed of metal centers and organic linkers and are dedicated for gas separation and storage2 due to their characteristic crystalline three-dimensional open framework with large pore diameters and high specific micropore volumes. The materials also reveal interesting properties for catalytic application.3,4 In general, a high structural diversity with numerous material properties can be formed.5,6 Due to their high metal ion content MOFs can be investigated by magnetic spectroscopy techniques, among which electron paramagnetic resonance (EPR) helps to elucidate the nature of paramagnetic probes including their interaction with diamagnetic and paramagnetic adsorbates in MOFs and together with theoretical density-functional theory © 2016 American Chemical Society

Received: September 19, 2016 Revised: November 2, 2016 Published: November 7, 2016 27399

DOI: 10.1021/acs.jpcc.6b09456 J. Phys. Chem. C 2016, 120, 27399−27411

Article

The Journal of Physical Chemistry C separation of CO2 and CH4.14 Besides MOF-5,15 HKUST-1 is also one of the best studied MOF materials. The structure of HKUST-1 belongs to space group Fm3m12 and is shown in Figure 1. The three-dimensional network

about the structure and dynamics of the paramagnetic adsorption complexes including among others distributions of complex geometries and motional effects of the adsorbed molecules which are hardly accessible from powder experiments alone. Despite this information deficit for the elucidation of complex structures in general, adsorption complexes formed with paramagnetic ions in porous materials such as MOFs but also zeolites, AIPOs, and SAPOs have been investigated widely, yet in powders only.24−26 For single crystals, EPR has so far often not been feasible, because the low-density materials typically form small crystallites which themselves feature relatively few atoms. Therefore, single-crystal EPR had been restricted to a few studies of framework substitutions in highfield EPR experiments.27 Here, we present for the first time single-crystal EPR studies at X-band frequencies of paramagnetic adsorption complexes in a porous, microcrystalline material allowing for the determination of the orientation of the metal ion g and (HFS) (A) tensor axes frames. This has been possible with the advances in EPR sensitivity also for lower X-band frequencies.28,29 Now, studies on MOF crystals with dimensions of only some hundred micrometer edge length have become more feasible.30 DRs with low concentrations of paramagnetic impurities and high dielectric permittivities provide an easy-to-use setup for conventional Xband EPR spectrometers, where the DR is placed in a typical EPR sample tube together with the studied sample to perform experiments also at nonambient conditions.31 The DRs provide sufficient sensitivity for investigations of MOF single crystals in commercial X-band cw EPR spectrometers and are an easy and convenient way to improve the signal-to-noise ratio (SNR) without limitations for typical gas adsorption experiments. Here, we describe the application of DRs for single-crystal MOF cw EPR studies and present the advantages for nonambient, low-temperature measurements together with low-pressure activation or gas adsorption experiments on the same sample. Our experiments allow together with previous EPR studies on the HKUST-1 powder18 and DFT calculations the determination of both the principal values and the orientation of the principal axes of the g-tensor and HFS tensor A of the mononuclear Cu2+ ions with adsorbed gases. This enabled us to explore the structure of the formed paramagnetic adsorption complexes in a superior way compared to previous reports and propose an improved model for the structure of this framework, utilizing also the findings from earlier DFT studies conducted with the corresponding powder materials.20 Although in this earlier study the HFS interactions to the ligands were prioritized, now the Cu2+ center in the PWshall be addressed with more attention.

Figure 1. Schematic representation of the structure of Cu3(btc)2 showing (A) the Cu2+ ion PWbuilding block and (B) the HKUST-1 network. Only the oxygen atoms binding axially to the Cu2+ are shown here.

contains antiferromagnetically coupled Cu2+ ions13 that are coordinated by carboxylate groups to a square whereas water molecules weakly bind to the residual axial binding site of the Cu2+ ions in the hydrated Cu3(btc)2 material (Figure 1A). The Cu2+ pairs are connected by the btc linker molecules to form a three-dimensional porous network (Figure 1B) with interconnected cages having two different pores with approximate diameters of 0.9 and 0.7 nm as measured between the oxygen atoms of the carboxylate groups. The axial water molecules can be easily removed from the Cu2+ ions by a moderate heat treatment in vacuum to form structurally well-defined accessible Lewis acid copper sites for catalytic applications.4 Within the wide variety of applications of MOFs it is crucial to tune their chemical and physical properties without altering the underlying structural features. It has been shown that this can be achieved either by manipulating the organic linker molecule16 or by modifying the inorganic building unit. An example of the latter is the partial substitution of the metal species to introduce local probes or special adsorption sites.17 It is possible to synthesize partially Zn2+-substituted HKUST-1 to form binuclear Cu/Zn PW building blocks.18 Via the separation of the binuclear, antiferromagnetically coupled Cu2+ ions, the excited S = 1 electron spin state and an EPR silent S = 0 ground state19 is replaced by a mononuclear Cu2+ ion with a S = 1 2 ground state, which is accessible by EPR spectroscopy also at low temperatures with high polarizability. The verification of such an isomorphous framework substitution had been shown by powder EPR spectroscopy for low Zn2+ concentrations,18 whereas DFT is capable of describing such interactions using model complexes with coordinated probe molecules.20 The Zn substitution offers the opportunity to explore Cu2+ adsorbate interactions and determine the structure of adsorption complex, and continuous-wave (cw) and pulsed EPR have been used widely to investigate such isomorphus substitutions in MOF powders recently.18,21−23 In principle, cw and pulsed EPR spectroscopy of paramagnetic centers in disordered materials allow us to measure the principal values of the magnetic interaction tensors and the relative orientation of their principal axes frames with respect to each other. Nevertheless, powder EPR experiments cannot give insights into the orientation of magnetic tensors with respect to the crystallographic axes. Such information can only be obtained from single-crystal experiments and may reveal valuable details

■

EXPERIMENTAL METHODS Single-Crystal Preparation. The Zn2+-doped material with 3.5% Zn2+, Cu2.965Zn0.035(btc)2 (1), was prepared by a slightly modified solvothermal synthesis established in literature and used earlier for both the pure and Zn-substituted HKUST1.4,18 In a typical setup, 2.4 mmol of Cu(NO3)2·3H2O (0.5394 g) and 0.36 mmol of Zn(NO3)2·6H2O (0.108 g) were dissolved in 10 mL of a 1:1 volume mixture of water and ethanol in a 15 mL pyrex vessel, closed and placed in an autoclave. The autoclave was heated by 10 K min−1 in an oven to 398 K for 12 h. The subsequent cooling was done with a rate of −5 K min−1. The turquoise-blue product was filtered and washed with 10 mL of ethanol three times, once per day. Single-crystal XRD 27400


Article

The Journal of Physical Chemistry C

was enhanced in the vicinity of the small single crystal due to the mw concentration inside the DR by a factor of 4, allowing for the overall reduced mw power suppressing signals from impurities in the cryostat.30 Gas Adsorption for EPR. After synthesis the crystal 1AS contains about 50% ethanol and has a light blue color, as seen in Figure 2A. The activated samples 1act were dried under vacuum at 393 K for 24 h to obtain dehydrated materials with removed axial ligand molecules. A structural representation of dehydrated PWs in all orientations occurring in the structure seen along a crystallographic [112] axis is given in Figure S1 in the Supporting Information. The dehydration was performed while having the crystal fixed in the EPR sample tube as readyto-measure setup including the DR. The crystals of 1act have a dark blue color; the final vacuum after evacuation was 3 × 10−2 mbar. Subsequently, the sample tubes as depicted in Figure 2B with the dehydrated material 1act were closed and cooled to room temperature without further air contact. After dehydration, all specimens of 1act showed the reproducible results presented later. In the following, gas adsorption was performed at the same crystal and in the same setup to obtain the crystals of 1CO2, 1CO, and 1MeOH. Carbon dioxide-adsorbed samples 1CO2 were prepared from dehydrated samples by adsorbing CO2 and carbon monoxide-adsorbed samples 1CO by adsorbing CO at 10 mbar at room temperature and methanol-adsorbed samples 1MeOH were prepared by adsorbing CH3OH at its vapor pressure at room temperature for 1 h, respectively. The same and several different crystals have been used each to ensure reproducibility of the results. After slowly cooling in the EPR spectrometer to 7 K, 1CO2, 1CO, and 1MeOH have been measured by EPR. A subsequent reactivation led to samples similar to 1act and could be reused again for gas adsorption. For that purpose, the evacuation valve has been maintained during the EPR measurements. 1AS was not used for detailed EPR analysis as without further characterization the exact configuration of the free Cu2+ coordination sites is unknown and nonreproducible. Fast cool-down and warm-up cycles during the EPR measurements limited the reusability of the crystals. Best results have been achieved for slow cooling rates of about 500 K h−1. Computational Details. Periodic electronic structure calculations were carried out at the Kohn−Sham densityfunctional theory (DFT) level of theory, employing the Gaussian plane-wave (GPW) formalism as implemented in the QUICKSTEP37 module within the CP2K program suite (Version 2.7).38,39 Molecularly optimized basis sets of double-ζ quality plus polarization in their shorter-range variant (DZVPMOLOPT-SR-GTH)39 were used on all atomic species (Cu, Zn, C, H, O). The interaction between the core electrons and the valence shell was described by Goedecker−Teter−Hutter (GTH) pseudopotentials.40−42 The general gradient approximation (GGA) to the exchange−correlation functional according to Perdew−Burke−Ernzerhof (PBE)43 was used in combination with Grimme’s D3-correction for dispersion interactions.44 The auxiliary plane-wave basis set was truncated at a cutoff of 400 Ry. We are using the Γ point approximation for the 630 atoms containing super cell. Initial coordinates of CuZn(btc)2 were obtained from the experimental crystal structure of Cu2.965Zn0.035(btc)2 where one out of 48 Cu2+ ions were replaced with a Zn2+ ion and the structure was relaxed in the fixed cell dimensions. In this way the Zn/Cu ratio in the sample is the lowest possible in the simulation cell (Zn/ Cu = 0.0213, whereas in the experimental sample it is in the

has been used to ensure a successful crystal growth, ICP-OES to detect the Zn concentration in the crystals. The pure Cu3(btc)2 (2) crystals have been synthesized likewise without modification of the established literature procedure. The structure contains six magnetically nonequivalent PW orientations. As synthesized, the sample 1AS of Cu2.965Zn0.035(btc)2 contains 3.5% mixed PWs with mononuclear Cu2+ ions. The obtained crystals with typical tetrahedral shapes have a side length of around 400 μm, as shown in Figure 2. Every face is a

Figure 2. (A) Single crystal of 1 stuck in grease, which was only used for this image. (B) Stacked PTFE crystal holder with crystal mounted. The DR is not slid fully down. The external magnetic field B0 is rotated perpendicular to the indicated z-axis.

(111) plane such that a simple crystal orientation can be performed by face alignment, as shown in Figure 2A. Diamond 3.2f with POV-Ray and Avogadro were used to visualize the structures.32−34 Single-Crystal cw EPR. The single crystals of 1 and 2 were placed with a (111) face on a quartz glass fiber of 0.7 mm diameter such that a [111] axis is pointing along the axis of the EPR sample tube. The crystal is fixed but not covered with twocomponent epoxy, which has no detectable EPR resonances to allow a dehydration of the crystal and gas adsorption. Figure 2B shows the crystal mounted on the fiber and the fiber itself placed in a PTFE holder to align it in the EPR sample tube. The whole setup including the EPR sample tube, vacuum valve, DR, and single crystal mounted in the latter has been rotated around one of the crystals’ [111] axes (named z in Figure 2B) with a conventional goniometer. The cw EPR measurements at a frequency of 9.41 Ghz were carried out using a BRUKER EMXmicro X-band spectrometer equipped with a BRUKER ER 4119HS cylindrical cavity resonator and an OXFORD ESR 900 flow cryostat for low-temperature measurements. Typically, five accumulations were taken per spectrum. The spectra were recorded with very low powers of 2 μW to minimize poorly resolved signals of other paramagnetic centers such as Cu2+ impurities in the cryostat or the later so-called Cu2+ species III. All spectra were recorded at 7 K sample temperature with a modulation strength of 0.5 mT and frequency of 100 kHz. Analysis of the EPR spectra has been facilitated by the EasySpin numerical simulation package for MATLAB.35 DRs of the type K80-H on the basis of barium lanthanide titanates solid solutions (BLTss) with the general formula Ba6−xLn8+2x/3Ti18O54 (Ln = Sm, Nd) have been used.30,36 The DR has a height of 2.6 mm and a diameter of 3.4 mm and contains the sample in a central hole of 1 mm diameter. The DR’s proportions are designed to fit in a standard EPR sample tube (inner diameter 4 mm) with evacuation valve attached fitting both into the flow cryostat and to the vacuum line and adsorption setup. On the basis of results obtained earlier for the DRs, we estimate a gain in the SNR per EPR spectrum by a factor of roughly 8.6 ± 2.0.30,31 Additionally, the (mw) B1 field 27401


Article


copper isotopes 63,65Cu. The subscript i marks all found species that differ in their tensor orientations, gi denotes the axial g tensor and Ai marks an axial HFS tensor of the respective copper nuclei with the principal axis values given in Table 1. The principal axes orientations of the respective gi and Ai tensors are found to be collinear. The total spin Hamiltonian is the sum of all Hamiltonians /i . Table 1 shows the principal axes values of the axially symmetric g and A tensors for all investigated samples of Cu2.965Zn0.035(btc)2. At present, we will focus on the first line (1act) and columns 1−4 and discuss the fifth column later. To simulate the resonance field positions presented above in Figure 4, the g and A principal axes values from the respective powder samples (repeated in Figure S2 in the Supporting Information) have been used as initial values and optimized for the single crystals. Deviations from the powder values18 will be presented and discussed in the next section. It is obvious that the resonance fields shown in Figure 4 resemble six distinct angular dependencies with an equal, 6fold symmetry. All six sets of resonance field lines have been fitted with the same g and A principle axis values, but different tensor orientations. These can be found in Table 2. The first three orientations belong to a group of species called I throughout this manuscript, the last three are called group II species, respectively. The principal axes orientations of the g and A tensors are obtained by a rotation of their z-axes in the B0-plane around the [111] axis (corresponding to the angle ϕ) to fit all resonance peaks. At first, three magnetically nonequivalent Cu2+ incorporation sites can be fitted (group Iact). They reach their largest effective HFS in a parallel orientation to the external magnetic field B0 one after another separated by 60°. This effective HFS corresponds to the maximum HFS based on former results on powder samples18 and the symmetry axes, denoted by z, of the g and A tensors of these species of group Iact are found to be embedded in the (111) plane. The angular dependencies of the resonance fields of two of the three species of the group Iact intersect every 30°. The low-field HFS lines meet for one intersection at about 285 mT and for the subsequent intersection at about 310 mT. The latter intersection marks the orientation of the highest HFS of the third orientation of the species of group Iact. Second, three further species labeled as a set of species IIact are identified in the spectra mainly in the spectral region 305−340 mT. They follow the angular dependencies of species Iact but are shifted to that by 30° such that they reach their orientation with the largest effective HFS in the considered (111) plane where the three species of group Iact do not have their maximum HFS. Hereby the z axes orientations of the g and A tensors of the group IIact species are lifted by about 55 ± 1° with respect to B0 and hence the (111) plane, each. Altogether, the fitted singlecrystal angular dependencies for all six Cu2+ species of 1act displayed in Figure 4 show a very good agreement for all identified resonance field positions. Due to the cubic symmetry of HKUST-1, no further rotations than those around [111] axes have been carried out as almost all other rotational planes would provide equal or less information. Figure 5 shows four, individual single-crystal spectra and simulations of 1act. More spectra with simulations can be found in Figure S3 in the Supporting Information. The spectra in Figure 5 are taken for four distinct orientations of the magnetic field B0 within the (111) plane. The presented spectra show the typical patterns of mononuclear Cu2+ ions with S = 1 electron spin interacting 2 with nuclear magnetic moments of 63Cu and 65Cu isotopes.

order of 0.0118). Three-dimensional periodic boundary conditions (PBC) were applied in combination with fixed unit cell parameters obtained from the experimental crystal structure (α = β = γ = 90.0° and a = b = c = 26.2896 Å). Geometries with or without adsorbate (CO, CO2 and MeOH) were fully relaxed in the fixed simulation cell. The EPR calculations were performed on cluster models cut out from the periodic structure. The cluster contains the Cu/ Zn dimer together with the four btc ligands and the axial adsorbate (without ligand, CO, CO2, and MeOH, labeled DFTact, DFTCO2, DFTCO, and DFTMeOH, respectively). Calculation of the EPR properties was done with ORCA electronic structure package45 using B3LYP hybrid exchange− correlation functional due to its excellent predictions of EPR properties.46−49 The HFS tensor A was calculated only for the Cu2+ atom described with CP(PPP) basis set documented previously,50 for all the other atoms TZVP/J basis set was applied. Spin−orbit coupling through the “complete mean field” approach with efficient Coulomb terms via resolution-ofthe-identity (RI) as it is implemented in ORCA (SOCType 3; SOCFlags 1,3,3,1) was taken into account.

■

RESULTS Dehydrated Cu2.965Zn0.035(btc)2. After dehydration of 1AS, the single crystals of 1 have been measured and denoted as 1act. After this activation process, the EPR spectra show wellresolved, angular-dependent resonance field lines of several paramagnetic Cu2+ ion species, as depicted in the upper plot of Figure 3 in comparison to the same orientation of other

Figure 3. Experimental spectra of all samples 1 with B0 along the crystallographic [110] axis.

samples of 1 below. The complete experimental and simulated angular dependencies of the resonance fields of 1act for a rotation of the external magnetic field B0 about the crystallographic [111] axis of the crystal 1act are plotted in Figure 4. At first, they can be fitted using a spin Hamiltonian which includes electron Zeeman and hyperfine interactions: /i = βe B0g iS + SA iI

(1)

Here S and I are electron spin and nuclear spin operators with the electron spin S = 1/2 and nuclear spin I = 3/2 for both 27402


Article


Figure 4. Angular-dependent resonance fields of Cu2+ species found for 1act at a quarter turn around the [111] axis (black squares). Solid lines represent simulated angular dependencies of the resonance fields of 63Cu2+ using the parameters in Tables 1 and 2, whereas 65Cu2+ leads to the line splittings observed in the low-field region along directions marked with [110]. EPR signals of six magnetically inequivalent symmetry-related Cu2+ species are observed, which can be grouped into two classes labeled by species of group Iact and IIact.

Table 1. Principal Axes Values and Angular Distribution Width Δα of the g and A Tensors of the 63Cu2+ Species of Groups I and II in Various Single Crystals of Cu2.965Zn0.035(btc)2 gzz

gxx,yy

Azz/ 10−4 cm−1

Axx,yy/ 10−4 cm−1

Δα/deg

1act 1CO2

2.281(2) 2.293(2)

2.046(1) 2.049(2)

187(2) 182(2)

32(2) 26(2)

3.0(5) 5.0(5)

1CO 1MeOH 1AS

2.300(1) 2.336(2) 2.337(2)

2.051(2) 2.058(1) 2.058(1)

175(2) 158(2) 158(2)

23(2) 15(2) 15(2)

5.0(5) 6.0(5)

sample

Table 2. Orientations of the Principal Axes (Euler Angles) of the g and A Tensors of the Cu2+ Species of Groups I and II in All Single Crystals of Cu2.965Zn0.035(btc)2 with Respect to the Crystal Frame (EasySpin Convention35), Errors Δϕ = Δθ = ±1°, and Deduced Crystal Axes as Explained in the Discussion species

ϕ/deg

θ/deg

1 2 3 4 5 6

45 0 90 45 0 90

−90 −45 −45 90 45 45

group I

II

Figure 5. Selected experimental and simulated Cu2+ EPR spectra for 1act at a rotation around the crystallographic [111] axis. For details of the simulation procedure see text.

crystal axis [110] [101] [011] [11̅ 0] [1̅01] [01̅1]

and simulated powder spectrum in Figure S2 one possible reason for a line width broadening can be identified. Although it is showing well-resolved HFS signals in the gzz part, indicating only moderate strain and line broadening effects, not all of the four HFS lines in the gzz part fit equally nice in line width. In particular, resonance field lines for higher fields of B0 have slightly broader line widths as those for lower fields. Having observed this in single crystals already earlier,31 we rated it as a typical sign for correlated g and A strain distributions.51 Consequently, such a correlated Gaussian g and A strain distribution has been taken into account in the simulations of all single-crystal spectra of 1. The correlation approximation and parameters are given in section S3 and Table S1 in the Supporting Information. However, this was not enough to explain all line widths of the tested samples of 1. Additionally, a distribution of the g and A tensors’ principal axes orientations must be assumed to account for the observed line shapes of the Cu2+ EPR signals recored in the spectra where B0 is pointing along all noncanonical orientations of the g and A tensors. This was motivated by the assumption that ligands, which are known to coordinate in a noncollinear position with the C4 axis of the PW might also be able to perturb the g and A tensor orientations of PWs with adsorbents.20 In fact, we found it useful to apply a mechanism also similar to that for 1act, although with smaller effects. In the simulations, we have

The resonance fields vary from 278 to 344 mT. Figure 5 displays the maximum HFS in this plane for species of group Iact in the spectrum recorded for B 0 parallel to the crystallographic [110] axis, which corresponds to B0 aligned along the symmetry axes of the g and A tensors. In the case of the second species in group IIact the maximum HFS in the (111) plane is obtained for B0 along [112]. For the individual spectral simulations the exact line shapes have been taken into consideration. For that purpose, a specific line broadening mechanism has been developed, which is adoptable to all investigated samples. The requirements on this line broadening model were as follows: As seen in Figure 5, the spectra with B0 along the crystallographic [110] axis suffers from almost no line broadening and the HFS lines of the63Cu and 65Cu isotopes are well resolved for the low-field signals. From this spectrum, a moderate line broadening of 1.3 mT peak-to-peak line width had been deduced as a general line width for all resonance lines. Now, at only 30° of rotation further with B0 along a crystallographic [112] axis the resonance lines have become significantly broader. From the comparison of the experimental 27403


Article


distribution width Δα with respect to model B to achieve the line width of the spectra. In the following Results section, we will first refer to the simple approach and discuss the other two depicted models in the Discussion section later, when also Figure 6D is issued. The latter line broadening mechanism is further discussed in the Discussion section. Adsorption of Carbon Dioxide on Cu2.965Zn0.035(btc)2. The dehydrated samples 1act have been used to adsorb various gases. Among them, the single crystals 1CO2 with adsorbed CO2 shall be presented next. The spectra of 1CO2 and 1act are comparable in a certain way, which will become clear in the subsections for 1CO and 1MeOH. At first, the resonance field lines shown in Figure S5 in the Supporting Information do not show very obvious differences compared to the results in Figure 4. Fitted by the same spin Hamiltonian (1) the values presented in line 2 of Table 1 are obtained. With these values, the singlecrystal angular dependence of 1CO2 shows again a good agreement for all identified resonance fields. The individual single-crystal spectra and simulations of 1CO2 are shown in Figure S4 in the Supporting Information. The spectra presented are similarly taken between the mirror planes in a [110] axis and a crystallographic [112] axis. The presented spectra show very well-resolved single-crystal patterns of mononuclear Cu2+ ions with S = 1 electron spin interacting with the nuclear 2 magnetic moments of the 63Cu and 65Cu isotopes. The resonance fields vary from 280 to 346 mT, which is only slightly higher than those of 1act. For the individual spectral simulations the same line broadening mechanism has been employed. The correlation parameters are given in Table S1 in the Supporting Information. The distribution of the g and A tensors’ principal axis orientations is almost twice as large compared with those of 1act and we found a distribution of the g and A tensor orientations by Δα = 5° around the respective equilibrium position α0 = 0. Again, we assumed a deflection of the g and A tensor principal axis orientations out of the C4 axis of the PW such that their projections onto the C4 plane of the PWpoint along the existing Cu−OP directions in the PW. These spectral simulations agree satisfyingly to the experimental spectra. Adsorption of Carbon Monoxide on Cu2.965Zn0.035(btc)2. Here we will present the single crystals 1CO with adsorbed CO. The previous spectra and those of 1CO are slightly different, because the spectra of the latter ones show besides the well-resolved resonance field lines of Cu2+ species

distributed the g and A tensor orientations over a small set of angles α and weighted them by a Gaussian distribution with width Δα around the respective equilibrium position α0 = 0. This is given in the fifth column of Table 1. Hereby, we assumed different models for the tilting of the g and A tensor principal axis orientations out of the C4 axis of the PWsuch that in the simplest approach their projections onto the C4 plane of the PWpoint toward a PW’s oxygen along the existing Cu−OP bond directions, forming a single Gaussian distribution per Cu−OP direction for every orientation given in Table 2 and shown in Figure 6A, each. Taking into account a moderate

Figure 6. (A) Cu2+ g and A tensor principal axis orientations (black) as deduced by spectral simulations from EPR data (shades indicate Gaussian weighting) and (B) and (C) in the more complex models, and (D) static angles delivered from DFT (gray shows mirror vectors). The tilting angle distribution Δα and static tilting angle α0 are depicted in orange.

value of Δα = 3.0°, this broadening mechanism together with the correlated strain effects can simulate all single-crystal spectra in Figure 5 including the gxx,yy part of the corresponding powder pattern with satisfying agreement. One can argue also other models could be possible and assume a somewhat more complex approach: a slightly bent adsorption complex structure with an angle α0 as deviation from the linear complex and for each bent configuration a Gaussian distribution, such as for example two conformations with α0 = ±|α0| and two Gaussian distribution with a corresponding width of Δα/2 as depicted in Figure 6B. To offer even more choices for the best model, it is also possible to rotate the projection of the tilting direction in between two Cu−OP bonds, which should also be possible for symmetry reasons and which is shown in Figure 6C. For this model, we found it necessary to roughly double the Gaussian

Figure 7. Angular-dependent resonance fields of Cu2+ species found for 1CO at a quarter turn around the [111] axis (black squares). Solid lines represent simulated angular dependencies of the resonance fields using the parameters in Tables 1 and 2. Again EPR signals of six magnetically inequivalent symmetry-related Cu2+ species are observed, which can be grouped in two classes labeled by species of groups ICO and IICO. 27404


Article


that the spectral simulations for the resolved peaks of 1CO agree with the experimental spectra with almost no exception. Cu2+ Species III. Besides the well-resolved single-crystal EPR resonance lines an underlying, broad and almost angularindependent signal of a further Cu2+ species III is present in the gxx,yy spectral range of the Cu2+ ion EPR spectra of 1CO. This very behavior has also been found for the sample 1MeOH described in the next subsection, and it is important to notice that the signal disappears repeatedly for reactivated samples 1act. In our simulations, the signal could not be identified clearly by means of the given theoretical model with tested angular perturbations of up to Δα ≤ 30°. For this reason, it is not represented in the spectral simulations anywhere in this manuscript. Adsorption of Methanol on Cu2.965Zn0.035(btc)2. At last, the single crystals 1MeOH with adsorbed CH3OH are described below. They are somewhat comparable to the previously shown spectra of 1CO, being clearly dominated by two general Cu2+ species groups, each. EPR resonance field lines of 1MeOH are shown in Figure S9 in the Supporting Information and can be fitted by eq 1, finally yielding the values presented in the fourth line of Table 1 with the directions of Table 2. Those values show a large correspondence to the values of 1AS. The relative difference to 1act is most obvious for 1MeOH, as it has already been the case for the corresponding powder, which has been studied in the previous publication for adsorbed CH3OH.18 The cw EPR powder spectra are published ibid. The values for the g and A tensor principal axis values gzz and Azz of species of group IMeOH can be distinguished very well from 1act but also 1CO/1CO2. As before, they show the typical patterns of

of groups I and II a broad spectrum that we assign to a third Cu2+ species, species III. In the following we will first focus on the well-resolved resonance field lines that are treated similarly to those of 1act and 1CO2. The very broad background spectrum will be discussed below. Cu2+ Species of Group I and II. EPR spectra of 1CO show the well-resolved resonance field lines that can be assigned to the previously described 6-fold symmetric angular dependency of a repeated four-line pattern of anisotropic EPR single-crystal spectra of the two respective mononuclear 63,65Cu isotopes. They are shown in Figure 7 and can be fitted by eq 1, yielding the values presented in the third line of Table 1 and directions of Table 2. With these values, the single-crystal angular dependence of 1CO shows again a good agreement for all identified resonance fields. Again, the simulation has shown that the principal axes orientations of the gi and Ai tensors are collinear with an uncertainty of 3° or better. The individual single-crystal spectra and simulations of 1CO are shown in Figure 8. Those spectra are taken from a larger set of EPR

mononuclear Cu2+ ions with S = 1 , I = 3 with 6-fold symmetry. 2 2 The resonance fields vary from 279 mT, which corresponds to the low-field, gzz part of the corresponding powder spectrum to 343 mT, representing the gxx,yy part. With these values, the single-crystal angular dependence of 1MeOH shows again a good agreement for all identified resonance fields. The individual single-crystal spectra and simulations of 1MeOH are shown in Figure S10 in the Supporting Information. The spectral resolution of the lines is even less compared to that for 1CO due to a once more increased line width. We found that this could be simulated by increasing both Δα to 6° and the correlation parameters given in Table S1 in the Supporting Information. Using these parameters, the spectral simulations for the resolved peaks of 1MeOH agree fairly with the experimental spectra but it is obvious that quite a large part of the spectrum is dominated by the broad background signal of the other, nonidentifiable Cu2+ species III, which disappears after a reactivation to 1act; for 1MeOH, the signal of Cu2+ species III is even more pronounced than for 1CO. Despite this, the total number of noncollapsed resonance lines, their approximate width and position of the simulated spectra coincides satisfyingly for all orientations with the experimental spectra. In the Discussion section, the line broadening mechanism is discussed further. The single crystals of 1AS, which are almost identical to those of 1MeOH (spin Hamilton parameters in last line of Table 1), are shown in section S2.5 and Figure S11 in the Supporting Information. Dehydrated Cu3(btc)2 and Adsorption of Methanol on Cu3(btc)2. For the purpose of investigation of the previously described unknown Cu2+ species III found in 1CO, 1MeOH, and 1AS we found it useful to measure pure HKUST-1 (Cu3(btc)2) single crystals in the same manner as we have measured the

Figure 8. Selected experimental and simulated Cu2+ EPR spectra for 1CO at a rotation around the crystallographic [111] axis. For details of the simulation procedure see text.

spectra shown in Figure S7 in the Supporting Information. Almost all of the presented spectra show well-resolved singlecrystal anisotropic Cu2+ spectra with the resonance fields varying from approximately 279 to 346 mT, which is not so much different from 1CO2. However, the spectral resolution of the lines is less due to an increased line width, especially toward the [121̅] crystallographic axis orientation. In terms of the line broadening mechanism described above, we found that the increased line width could rather be simulated by increasing the correlation parameters given in Table S1 in the Supporting Information than the angular perturbation Δα, which is up to half of a degree comparable to that for 1CO2. For an exclusive fitting of the resolved angular peaks then, no further adjustments are necessary compared to the simulations described before. Figure S6a in the Supporting Information shows such an example simulation in particular in its composition with all six Cu2+ incorporation sites for one random orientation, which are depicted in Figure S6b. We find 27405


Article

The Journal of Physical Chemistry C single crystals of 1 with adsorption of CH3OH, as for 1MeOH the observed effects on all species in groups I−III have been most pronounced. Because one does not expect to find the antiferromagnetically coupled S = 1 state of the Cu/Cu PWs at low temperatures, it is not obvious at first sight why this is done. Nevertheless, EPR resonance field lines of 2act and 2MeOH are shown in comparison with those for 1CO and 1MeOH in Figure 9. Though the spectra of 2act can be explained with Cu2+

calculations cannot be resolved. DFTCO2 has the larges rhombic deviation with Ayy − Axx = 6 × 10−4 cm−1 with the experimental finding of Axx,yy = 26(2) × 10−4 cm−1 right in between. Overall, with an identical trend of increasing g factors with respect to an increasing magnitude of the binding energies of the adsorbed ligands and decreasing HFS, DFT does provide a reasonable correspondence to the experimental values. The binding energies of the axial ligand (if applicable) to the Cu2+ center of the PWs determined by DFT are also summarized in Table 3 and in Table S2 in the Supporting Information for comparison. In Table S2 in the Supporting Information we additionally introduced a hypothetical, isoelectronic but defective Cu/Cu PWcontaining one Cu+, which can be a possible modification in Cu2.965Zn0.035(btc)2 with its major PW configuration of Cu/Cu metal centers. Further motivation for this is given in section S4 in the Supporting Information, but for the moment we argue that the found binding energies of ligand−Cu2+(Cu/Zn PW) and ligand−Cu+ (Cu/Cu PW) are very similar and therefore the g and A tensors should not be very different. Consequently, it suggests that in cw EPR Cu/Zn and Cu/Cu PWs might be difficult to distinguish. Thus, for this manuscript, we will stick with the ligand-Cu2+coordination in Cu/Zn PWs. The optimized structure shows a distinct distortion of the isolated mixed Cu/Zn PWs. Also, principal axes directions of the g and A tensors can be analyzed from the DFT data. A schematic representation of the situation is given in Figure 10 as excerpt

Figure 9. EPR spectra recorded for a larger magnetic field range of samples 2act and 2MeOH (green) in comparison to the spectra of 1CO and 1MeOH (gray and black, respectively) as well as of the reactivated sample re-2act (bottom line).

impurities in the cryostat interfering often at low signal intensities,31 surprisingly the CH3OH gas adsorption afterward leads to a broad Cu2+ signal as with 1CO or 1MeOH, which are repeated as example here. We attribute this broad signal to a species similar to those denoted by Cu2+ species III in the previous sections and perceive neither apparent HFS signals nor clear evidence for an angular dependency. It is remarkable to note that after a reactivation the signal disappears as described earlier. Simulations with our proposed simple model from above using even distribution widths up to Δα ≈ 30° do not reproduce the signal shape. DFT Calculations. The computed values for the g tensor and HFS tensor A are listed in Table 3 along with the DFT predicted values for the tilting angle of the optimized PWstructures’ g and A tensors. Compared to the experimental findings of purely axially symmetric tensors given in Table 1, the values suggest a slight rhombic deformation, but the deviations are mostly within the uncertainties assumed in the last given digit. In fact, the experimental uncertainties for gxx,yy and Axx,yy motivate that the differences found by DFT

Figure 10. Schematic representation of DFTMeOH with indicated Cu2+ g tensor z principal axis orientation (cyan) viewed along the optimized structures’ Cu/Zn axis in the PW.

of one PWfrom the periodic, unit cell cluster of Cu2.965Zn0.035(btc)2 with about the same Zn2+ concentration as in 1. We found a tilting angle of 1.0° for DFTact, 0.4° for DFTCO2, 0.2° for DFTCO, and 2.0° for DFTMeOH of the z principal axes directions of the g tensors of Cu2+ in the static, optimized structures with respect to the Cu/Zn axis of the PW. Note that this axis does not coincide with the direction of the ligand-Cu2+bond. The z principal axis directions of the A tensors coincide reasonably well with those of the g tensors (deviations of 0.1°, 0.2°, 0.0°, and 0.3°, respectively) and can therefore be seen as collinear in accordance to the experimental results presented above. The projections of the g tensor z principal axis directions onto the plane spanned by the Cu−OP

Table 3. Principal Axes Values and Angular Tilting α0 of the Principal Axes Directions of g and A Tensors and Binding Energies of the Ligand to the Cu2+ in the Computed PWs of Cu2.965Zn0.035(btc)2 gzz

gyy

gxx

Azz/10−4 cm−1

Ayy/10−4 cm−1

Axx/10−4 cm−1

α0/deg

BE/eV

DFTact DFTCO2

2.179 2.182

2.045 2.047

2.044 2.045

−274 −273

−32 −29

−30 −23

1.0 0.4

−0.24

DFTCO DFTMeOH

2.195 2.201

2.050 2.055

2.050 2.054

−260 −270

−14 −15

−14 −15

0.3 2.0

−0.60 −0.69

computed structure

27406


Article


average out over a rotated azimuthal angle.20 The experimental finding for 1CO2 of Axx,yy = 26(2) × 10−4 cm−1, which matches surprisingly well the mean values of DFTCO2, Axx = 29(2) × 10−4 cm−1 and Ayy = 23(2) × 10−4 cm−1, is the distribution of this averaged value measured by EPR and it confirms therefore again this assumption. Geometrically, three species (group I) were found to be incorporated with the z axes of their g and A tensors parallel to the crystallographic (111) plane and each aligned along the crystallographic [11̅0], [011̅], or [101̅] axes. Additionally, three further species (group II) are identified with z axes orientations along the crystallographic [110], [011], and [101] axes of the cubic crystal lattice. HKUST-1 with its Fm3m space group12 has 36 Cu2+ pairs in the lattice cell that are forming six magnetically inequivalent sites in accordance with the observed six magnetically inequivalent Cu2+ species in the studied Cu2.965Zn0.035(btc)2 crystals of 1act, 1CO2, 1CO, 1MeOH, and 1AS. In a cubic lattice the [111] axis corresponds to a 3-fold screw axis. Indeed for the rotation of the magnetic field about the [111] axis we observed two groups of angular dependencies (I and II) and each consists of resonance lines belonging three species and showing a 3-fold symmetry. The paddle-wheel units have a local C4 rotational axis about the vector joining the two Cu2+ ions in the pair. The C4 axes for the six magnetically inequivalent sites of the PWs point along the six face diagonals of the cubic lattice cell, the [11̅0], [011̅], [101̅], [110], [011], and [101] crystallographic axes. It is natural to assume that the symmetry axis (z axis) of the cupric ion g and A tensors in the mixed PWs in which one Cu2+ ion has been substituted by Zn2+ will also point along these local C4 axes in perfect agreement with the six experimentally found z axes orientations of these tensors (Table 2) and the DFT results. Figure 11 illustrates a

bonds in the PW are rotated by 37°, 40°, 24°, and 38°, respectively, from an Cu−OP bond and therefore directed mostly in between two of the PWs’ four oxygens OP. The optimized structure itself resembles those angles by the finding that for DFTact, DFTCO, and DFTMeOH two adjacent Cu−OP bonds are about 1% longer than the two other Cu−OP bonds, forming themselves a tilted PW, whereas in the frozen structure of DFTCO2 only a single bond in shortened by that order of magnitude, leading to the rhombic g tensor.

■

DISCUSSION The rotations of the single crystals of 1 with respect to B0 in the (111) plane presented above revealed the presence of six magnetically inequivalent Cu2+ incorporation sites in the single crystal of Cu2.965Zn0.035(btc)2 with the same principal axes values of the their g and A tensors. Therefore, the six distinct Cu2+ species are chemically equivalent but incorporated at different crystallographic positions. Their spin Hamiltonian parameters differ for each of the samples 1act, 1CO2, 1CO, 1MeOH, and 1AS (Table 1) but coincide reasonably well within the given mutual uncertainties with the findings for the powder samples and indicate the influence of the adsorption process on the cupric ion coordination. In particular, we we have only found slight disagreements for Δgzz = −0.002 and ΔAzz = +3 × 10−4 cm−1 for 1act and Δgxx,yy = +0.002 for 1MeOH compared to previously published results.18,20 A variation of the principal values of the g and A tensors of the cupric ion indicates a change in the coordination geometry.52 For example, in accordance to crystal field theory a stronger axial ligand field at the Cu2+ ion manifests in larger gzz values and smaller Azz values. The experimentally obtained increase of gzz and the decrease of Azz correspond to a distortion of the square planar coordination geometry toward a square pyramidal coordination. The ligand field splitting Δ between the dxy and the dx2−y2 orbitals is reduced, causing an increase of the gzz principal axis value.53 Such an increase of gzz is observed in combination with a decrease of the HFS upon adsorption of all tested and calculated adsorbent compared to the case for activated material as found for the powder materials before.18,20 Also in the single crystals, the axial ligand field is found to be strongest for CH3OH and decreases via CO and CO2 to the activated sample 1act. The DFT calculations supported the observations by predicting a decreasing trend in the computed gzz factors in an almost symmetric g tensor for the various PWconfigurations tested, DFTMeOH, DFTCO, DFTCO2, and DFTact, and increasing Azz principal axis values of the HFS tensor as shown in Table 3. This is correlated to the equally decreasing binding energies listed for the same computed models, with the exception of DFTMeOH, which shows slightly higher principal axis values for A than DFTCO. One assumption is that the coordination of CH3OH involves several orientations with similar binding energies that could lower especially the Azz value with respect to Azz of DFTCO. Therefore, we attribute this deviation to a computational insufficiency. The rhombic deviation in gxx and gyy of DFTCO2 can be explained by the expected averaging from the CO2 ligand molecule: although the DFT calculation presents a static, frozen picture, it was shown earlier that the low energy required to rotate the CO2 molecule at its bond to Cu2+ of 2 kJ mol−1 can lead to several frozen-molecule positions with equivalent energy but almost unchanged gzz values, whereas the gxx and gyy

Figure 11. Schematic representation of the structure of all PW orientations occurring in the hydrated MOF Cu3(btc)2 viewed along the crystallographic [111] axis (indicated by the black dot in the middle). All PWs show g and A tensors pointing from Zn2+ to Cu2+ along the PWs’ C4 axes. Species of group I (red) are incorporated in the (111) plane, species of II (cyan) are raised from it.

projection of the orientations of the six magnetically inequivalent PWs units onto the (111) plane together with experimentally determined z axis orientations of the coaxial Cu2+ g and A tensors of the mixed Cu/Zn pairs illustrated before in Figure S6b for the simulation of the individual tensor orientation of all six incorporation sites. Consequently, we may conclude that our single-crystal measurements further confirm the powder measurements18 by finding the Cu2+ species in Cu2.965Zn0.035(btc)2 at all crystallograhically possible, former dinuclear Cu/Cu sites of Cu3(btc)2, representing another 27407


Article


fashion with an average position α0 = 0. The conducted simulation can now be used to find a Gaussian distribution width that the spectral simulations would favor. We have compared all spectra of 1 taken for the orientations of the crystals and B0 between the two mirror planes along the crystallographic [110] and [112] axes with that model and found that the tilt of the z axes of the g and A tensors toward one of the Cu−OP bonds provides a reasonably accurately broadened line shapes of the Cu2+ EPR resonance lines with Δα = 3.0(5)−6.0(5)° in comparison to further models tested: another model involves a bent adsorption structure with two equilibrium positions of the z principal axis orientations with an intermediate splitting of 2α0, their projection at first along the Cu−OP direction and a distribution of width Δα/2 (Figure 6B) and further with the projections exactly between two of the Cu−OP bonds (45° from one Cu−OP bond in the plane perpendicular to the C4 axis) and a distribution of width Δα again (Figure 6C, Δα according to Table 1). We have not found any evidence that either of these models improves the achieved line broadening of the spectral simulations significantly. For the simulations with the model of Figure 6B the computational efforts doubled in favor of a decreasing distribution width. For the model described by Figure 6C the distribution width was found to be approximately Δα again with still the 2-fold number of spectra, which have to be computed for the fit. Naturally, even further mixtures of these models and directions might occur. Nevertheless, the experimentally recorded EPR resonance lines did not allow for a more certain choice between these models, as they all provided reasonably well broadened spectral simulations. Therefore, all spectral simulations shown here and in the Supporting Information involve only the simplest model. Taking into account the unresolved background signal of Cu2+ species III, even the simple model provides sufficiently accurately broadened resonance field lines. On the contrary, the more complex models tested for the spectral simulations (Figure 6B,C) also lead to similarly broadened EPR resonance lines, as Figure S8 in the Supporting Information shows for 1MeOH as an example. Here, we chose the equilibrium positions α0 = ± 2° ≠ 0 and decreased Δα to only half the value of the original spectral simulation of 1MeOH for the two Gaussian curves, each. As one can see, such a small distribution of the tilting angle already leads to almost sufficiently broadened EPR lines. Therefore, from the spectral simulations alone it is difficult to decide for an appropriate model, if both the perturbation direction and the distribution width are so some extends arbitrary. On the basis of the overall accordingly reproduced z principal axis values of the g and A tensors for DFTMeOH, DFTCO, DFTCO2, and DFTact we evaluated the optimized structures and the direction cosines of the g and A tensors with respect to that structure. They were given in Table 3 and have shown likewise a small tilting of the principal axes directions compared to the Cu/Zn direction in the respective PWs, as shown in Figure 10. The overall findings of the DFT computations confirm the natural assumption that a ligand oxygen OL or carbon CL atom of an adsorbed molecule binding to the cupric ion favors positions far most off the PW’s own oxygens’ OP repelling potentials, which is also resembled by the g and A z principal axis projection directions and it was surprising to find distortions of the local C4 symmetry of the PW units even for the dehydrated structure DFTact. However,

method to verify a successful substitution of Cu2+ by Zn2+ ions in the PWs of this MOF. In addition, these experiments show the effectiveness of our cw EPR setup using DRs for studies of small single-crystal samples with low spin concentrations at nonambient conditions such as low temperatures and evacuated samples. Though a former study addressed the HFS interaction to various ligands,54 in this work the influence of the adsorption of various gases on the Cu/Zn PWs in 1 has been explored by DR aided single-crystal cw EPR spectroscopy. These measurements demonstrate likewise the versatility of the DRs for EPR measurements at nonambient conditions, in this case again at low temperatures and now in the presence of various gas atmospheres. The adsorption of CO2, CO, and CH3OH induces changes in the principal values of the g and A tensors of the Cu2+ ions as well as in the observed EPR line patterns in both the powder18 and single-crystal spectra of 1. Otherwise, our single-crystal EPR studies of 1 upon adsorption of various gases indicate that the overall orientations of the cupric ion g and A tensors of the mixed Cu/Zn PWs is not significantly affected by the gas adsorption process. Powder spectra of zinc-doped Cu3(btc)2, both of activated materials and upon adsorption of CO2, CO, and CH3OH, can be satisfactorily simulated using correlated g and A strain effects (Figure S12 in the Supporting Information). However, to explain the line shapes of the Cu2+ EPR signals in the recorded single-crystal spectra of 1, we found that it is not enough to apply just such correlated strain effects, and that also unresolved 1H hyperfine couplings can be excluded on the basis of an earlier ENDOR study of the material.18 Instead, we tested additionally a simple model based on on a tilt of the z symmetry axes of the Cu2+ ions’ g and A tensors due to the adsorption process. In a first approach of such a model we assumed perturbations of the local C4 symmetry of the PW unit. The z axes of both tensors are oriented such that their projections onto the plane perpendicular to the C4 axis and spanned by the four carboxylate oxygen atoms OP of the PW unit binding to the Cu2+ ion are along one of the four Cu−OP bonds. Our expectation was that not only adsorbate molecules with a low symmetry such as CH3OH but also various linear molecules such as CO2 will form a bent adsorption complex with the Cu2+ ions, as has been verified experimentally recently by pulsed EPR spectroscopy20 and will destroy the local C4 symmetry of the PW unit. Consequently, the z axes orientation of the g and A tensors might deflect from the former C4 symmetry axes and can be described by this model. The idea is represented in Figure 6 (left-hand side). Otherwise, a small linear symmetric molecule such as CO or a highly symmetric one such as H2O adsorbed at the host PW’s Cu2+ binding site is expected to point straight along the C4 axis, as depicted in Figure 11 for water (represented by the ligand’s central OL atom) for the hydrated structure of Cu3(btc)2.20 Therefore, the z axes of the g and A tensors should also point along the local C4 symmetry axis in such cases. Then, any distortion from this z axis alignment would indicate deviations of the real structure from the ideal single-crystal structure, e.g., due to structural defects, and should be significantly less than in the above case of bent adsorption complexes formed by molecules with lower symmetry. In our simulations implementing that first, simple model from Figure 6A we assume that all four possible orientations of the tensors z axes would occur with equal probability and that they can be distributed around the C4 axis by Δα in a Gaussian 27408


Article


time. We identified nicely resolved, distinct, angular-dependent EPR spectra of six mononuclear Cu2+ ion incorporation sites according to the crystal structure in the single crystals of the framework Cu2.965Zn0.035(btc)2 and confirmed their origin from mixed Cu/Zn PWs. Further, we monitored a line broadening effect which can reasonably well be explained by a tilting of the g and A principle axis orientations of the host Cu2+ ions in the PWs by a few degrees. We propose different models for this perturbation and suggest, inspired by DFT calculations, a probable tilting direction between two Cu−O bonds of the PWand equilibrium tilting angle of a few degrees. Moreover, we propose a distribution in the real structure and find differences for the monitored adsorption of CO, CO2, and CH3OH with respect to each of these parameters. It it important to mention that those minor changes in the principle axis orientations could never be observed by powder EPR before in MOFs.

these results must be seen as a single excerpt from the structure involving already 630 atomic parameters to compute. Therefore, it is likely that from Figure 10 one can deduce Figure 6D as for symmetry reasons the single computed orientation (black) is only one of all orientations occurring (gray) in all PWs. Then, the remaining discussion joins the DFT calculations and the spectral simulations of the single-crystal cw EPR spectra of 1, as we now can decide that the most complex model presented in Figure 6C resembles the DFT results best. It is therefore likely that we can describe the changes of the spin Hamilton g and A principal axis directions to involve at least one main tilting direction between two of the Cu−OP bonds, each leading to four possible tilting directions per PW. The equilibrium positions α0 given in Table 3 as predicted by DFT can be a reasonable assumption also for the spectral simulations and, as the DFT data is derived from a single, perfect unit cell of Cu2.965Zn0.035(btc)2, we suppose that unaccounted strain effects within the MOF framework, presumably caused by lattice defects, may lead to the deviations in the g and A tensor orientations observed by EPR and described by the Gaussian distributions in the spectral simulations with a distribution width amounted to Δα as given in Table 1. Therefore, we assess the DFT data to agree sufficiently well to the model described in Figure 6C for the experimental observations which on the basis of the SNR of individual line shapes alone otherwise would not llow us to confirm a specific model. To our knowledge, such structural changes during activation and upon adsorption on Zn2+substituted Cu2.965Zn0.035(btc)2 single crystals have never been monitored with competing techniques to EPR that could give estimates about this kind of behavior. Nevertheless, it is important to note that we have only observed this effect in mixed Cu/Zn PWs of 1 but not for the parent Cu3(btc)2 sample 2, which is not accessible by EPR in the same way. However, we did find coincidences for the measurements of 1 and 2 regarding a nonresolved Cu2+ species III with a repeatedly disappearing broad signal after reactivation observed in the spectra of 1CO and 1MeOH (and 1AS) as well as 2MeOH (and 2AS) but not 1CO2 (and (re-)1act/2act). As the Cu2+ species III occurs in both materials 1 and 2, it cannot be related to the Cu/Zn PWand seems to be a general feature of HKUST1. Therefore, we have to conclude that it is not influencing the observations we made for the species of group I and II upon adsorption, but instead is a different effect and further considerations about this are issued in section S4 in the Supporting Information.

ACKNOWLEDGMENTS The authors gratefully acknowledge financial support within the priority programs SPP 1362 (Poröse metallorganische Gerüstverbindungen) and 1601 (Increasing Sensitivity with EPR) of the Deutsche Forschungsgemeinschaft (DFG) and support by the Center for Information Services and High Performance Computing (ZIH) at Technische Universität Dresden for the provided computational resources. The authors thank the group of Stefan Kaskel, Technische Universität Dresden, for providing and characterizing the samples.

CONCLUSIONS The g and A tensors of the Cu2+ species in the compounds of 1 are very sensitive to minor structural changes in the framework induced by the different adsorption procedures. This shows the sensitivity of both tensors with respect to small changes in the coordination environment of the Cu2+ ions in the PW units and the high SNR, which is provided by the DRs needed to observe it. All single-crystal measurements have been performed in a conventional X-band EPR spectrometer at 7 K sample temperature using DR aided sensitivity-enhancing cw EPR. The successful application of this technique at nonambient temperatures demonstrates the feasibility and potential of EPR spectroscopy on conventional X-band EPR spectrometers for the study of small single crystals even of porous, low-density MOF materials under nonambient atmospheres for the first

(1) Janiak, C.; Vieth, J. K. MOFs, MILs and More: Concepts, Properties and Applications for Porous Coordination Networks (PCNs)rous Coordination Networks (PCNs). New J. Chem. 2010, 34, 2366−2388. (2) Hartmann, M.; Kunz, S.; Himsl, D.; Tangermann, O.; Ernst, S.; Wagener, A. Adsorptive Separation of Isobutene and Isobutane on Cu3 (BTC)2. Langmuir 2008, 24, 8634−8642. (3) Corma, A.; Garcia, H.; Llabrés i Xamena, F. Engineering Metal Organic Frameworks for Heterogeneous Catalysis. Chem. Rev. 2010, 110, 4606−4655. (4) Schlichte, K.; Kratzke, T.; Kaskel, S. Improved Synthesis, Thermal Stability and Catalytic Properties of the Metal-Organic Framework Compound Cu3(BTC)2. Microporous Mesoporous Mater. 2004, 73, 81−88. (5) Kitagawa, S.; Kitaura, R.; Noro, S.-i. Functional Porous Coordination Polymers. Angew. Chem., Int. Ed. 2004, 43, 2334−2375.

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.jpcc.6b09456. Structure of PW orientations, additional EPR spectra on all samples, experimental resonance fields, correlated g and A strain effects, spin Hamiltonians, and details for the DFT calculations (PDF)

■

AUTHOR INFORMATION

Corresponding Author

*A. Pöppl. E-mail: [email protected]. Phone: +49 (0)341 9732608. Fax: +49 (0)341 9732649. ORCID

Stefan Friedländer: 0000-0002-7703-719X Notes

The authors declare no competing financial interest.

■

■

■

27409

REFERENCES


Article

The Journal of Physical Chemistry C (6) Shen, L.; Yang, S.-W.; Xiang, S.; Liu, T.; Zhao, B.; Ng, M.-F.; Göettlicher, J.; Yi, J.; Li, S.; Wang, L.; et al. Origin of Long-Range Ferromagnetic Ordering in Metal-Organic Frameworks with Antiferromagnetic Dimeric-Cu(II) Building Units. J. Am. Chem. Soc. 2012, 134, 17286. (7) Kurmoo, M. Magnetic Metal-Organic Frameworks. Chem. Soc. Rev. 2009, 38, 1353−1379. (8) Jee, B.; Koch, K.; Moschkowitz, L.; Himsl, D.; Hartman, M.; Pöppl, A. Electron Spin Resonance Study of Nitroxide Radical Adsorption at Cupric Ions in the Metal-Organic Framework Compound Cu3(btc)2. J. Phys. Chem. Lett. 2011, 2, 357−361. (9) Mendt, M.; Jee, B.; Himsl, D.; Moschkowitz, L.; Ahnfeldt, T.; Stock, N.; Hartmann, M.; Pöppl, A. A Continuous-Wave Electron Paramagnetic Resonance Study of Carbon Dioxide Adsorption on the Metal-Organic Frame-Work MIL-53. Appl. Magn. Reson. 2014, 45, 269−285. (10) Mavrandonakis, A.; Vogiatzis, K. D.; Boese, A. D.; Fink, K.; Heine, T.; Klopper, W. Ab Initio Study of the Adsorption of Small Molecules on Metal-Organic Frameworks with Oxo-centered Trimetallic Building Units: The Role of the Undercoordinated Metal Ion. Inorg. Chem. 2015, 54, 8251−8263. (11) Friedländer, S.; Liu, J.; Addicoat, M.; Petkov, P.; Vankova, N.; Rüger, R.; Kuc, A.; Guo, W.; Zhou, W.; Lukose, B.; et al. Linear Chains of Magnetic Ions Stacked with Variable Distance: Ferromagnetic Ordering with a Curie Temperature above 20K. Angew. Chem., Int. Ed. 2016, 55, 12683−12687. (12) Chui, S. S.-Y.; Lo, S. M.-F.; Charmant, J. P.; Orpen, A. G.; Williams, I. D. A Chemically Functionalizable Nanoporous Material [Cu3 (TMA)2 (H2O)3]n. Science 1999, 283, 1148−1150. (13) Zhang, X. X.; Chui, S. S.-Y.; Williams, I. D. Cooperative Magnetic Behavior in the Coordination Polymers [Cu3 (TMA)2L3](L= H2O, Pyridine). J. Appl. Phys. 2000, 87, 6007−6009. (14) Chowdhury, P.; Mekala, S.; Dreisbach, F.; Gumma, S. Adsorption of CO, CO2 and CH4 on Cu-BTC and MIL-101 Metal Organic Frameworks: Effect of Open Metal Sites and Adsorbate Polarity. Microporous Mesoporous Mater. 2012, 152, 246−252. (15) Li, H.; Eddaoudi, M.; O’Keeffe, M.; Yaghi, O. M. Design and Synthesis of an Exceptionally Stable and Highly Porous Metal-Organic Framework. Nature 1999, 402, 276−279. (16) Wang, Z.; Cohen, S. M. Postsynthetic Modification of MetalOrganic Frameworks. Chem. Soc. Rev. 2009, 38, 1315−1329. (17) Zeng, M.-H.; Wang, B.; Wang, X.-Y.; Zhang, W.-X.; Chen, X.M.; Gao, S. Chiral Magnetic Metal-Organic Frameworks of Dimetal Subunits: Magnetism Tuning by Mixed-Metal Compositions of the Solid Solutions. Inorg. Chem. 2006, 45, 7069−7076. (18) Jee, B.; Eisinger, K.; Gul-E-Noor, F.; Bertmer, M.; Hartmann, M.; Himsl, D.; Pöppl, A. Continuous Wave and Pulsed Electron Spin Resonance Spectroscopy of Paramagnetic Framework Cupric Ions in the Zn(II) Doped Porous Coordination Polymer Cu3−xZnx(btc)2. J. Phys. Chem. C 2010, 114, 16630−16639. (19) Pöppl, A.; Kunz, S.; Himsl, D.; Hartmann, M. CW and Pulsed ESR Spectroscopy of Cupric Ions in the Metal-Organic Framework Compound Cu3 (BTC) 2. J. Phys. Chem. C 2008, 112, 2678−2684. (20) Jee, B.; St. Petkov, P.; Vayssilov, G. N.; Heine, T.; Hartmann, M.; Pöppl, A. A Combined Pulsed Electron Paramagnetic Resonance Spectroscopic and DFT Analysis of the 13CO2 and 13CO Adsorption on the Metal-Organic Framework Cu2.97Zn0.03(btc)2. J. Phys. Chem. C 2013, 117, 8231−8240. (21) Šimėnas, M.; Kobalz, M.; Mendt, M.; Eckold, P.; Krautscheid, H.; Banys, J.; Pöppl, A. Synthesis, Structure, and Electron Paramagnetic Resonance Study of a Mixed Valent Metal-Organic Framework Containing Cu2 Paddle-Wheel Units. J. Phys. Chem. C 2015, 119, 4898−4907. (22) Nevjestić, I.; Depauw, H.; Leus, K.; Kalendra, V.; Caretti, I.; Jeschke, G.; Van Doorslaer, S.; Callens, F.; Van Der Voort, P.; Vrielinck, H. Multi-frequency (S, X, Q and W-band) EPR and ENDOR Study of Vanadium (IV) Incorporation in the Aluminium Metal-Organic Framework MIL-53. ChemPhysChem 2015, 16, 2968− 2973.

(23) Kozachuk, O.; Khaletskaya, K.; Halbherr, M.; Bétard, A.; Meilikhov, M.; Seidel, R. W.; Jee, B.; Pöppl, A.; Fischer, R. A. Microporous Mixed-Metal Layer-Pillared [Zn1−xCux(bdc) (dabco)0.5] MOFs: Preparation and Characterization. Eur. J. Inorg. Chem. 2012, 2012, 1688−1695. (24) Goldfarb, D.; Bernardo, M.; Strohmaier, K.; Vaughan, D.; Thomann, H. Characterization of Iron in Zeolites by X-Band and QBand ESR, Pulsed ESR, and UV-Visible Spectroscopies. J. Am. Chem. Soc. 1994, 116, 6344−6353. (25) Hartmann, M.; Kevan, L. Generation of Ion-Exchange Capacity by Silicon Incorporation into the Aluminophosphate VPI-5/AIPO 4− 8 Molecular Sieve System. J. Chem. Soc., Faraday Trans. 1996, 92, 3661−3667. (26) Hartmann, M.; Kevan, L. Transition-Metal Ions in Aluminophosphate and Silicoaluminophosphate Molecular Sieves: Location, Interaction with Adsorbates and Catalytic Properties. Chem. Rev. (Washington, DC, U. S.) 1999, 99, 635−664. (27) Arieli, D.; Prisner, T.; Hertel, M.; Goldfarb, D. Resolving Mn Framework Sites in Large Cage Aluminophosphate Zeotypes by High Field EPR and ENDOR Spectroscopy. Phys. Chem. Chem. Phys. 2004, 6, 172−181. (28) Narkowicz, R.; Suter, D.; Stonies, R. Planar microresonators for EPR experiments. J. Magn. Reson. 2005, 175, 275−284. (29) Sienkiewicz, A.; Vileno, B.; Garaj, S.; Jaworski, M.; Forro, L. Dielectric Resonator-Based Resonant Structure for Sensitive ESR Measurements at High-Hydrostatic Pressures. J. Magn. Reson. 2005, 177, 261−273. (30) Friedländer, S.; Ovchar, O.; Voigt, H.; Böttcher, R.; Belous, A.; Pöppl, A. Dielectric Ceramic EPR Resonators for Low Temperature Spectroscopy at X-band Frequencies. Appl. Magn. Reson. 2015, 46, 33−48. (31) Friedländer, S.; Šimėnas, M.; Kobalz, M.; Eckold, P.; Ovchar, O.; Belous, A. G.; Banys, J.; Krautscheid, H.; Pöppl, A. Single Crystal Electron Paramagnetic Resonance with Dielectric Resonators of Mononuclear Cu2+ Ions in a Metal-Organic Framework Containing Cu2 Paddle Wheel Units. J. Phys. Chem. C 2015, 119, 19171−19179. (32) Brandenburg, K. Diamond 3.0; Crystal Impact GbR: Bonn, 2004. (33) Persistence of Vision Raytracer, V. 3.7; Persistence of Vision Pty. Ltd.: Williamstown, Victoria, Australia, 2016. (34) Hanwell, M. D.; Curtis, D. E.; Lonie, D. C.; Vandermeersch, T.; Zurek, E.; Hutchison, G. R. Avogadro: An Advanced Semantic Chemical Editor, Visualization, and Analysis Platform. J. Cheminf. 2012, 4, 17. (35) Stoll, S.; Schweiger, A. EasySpin, a Comprehensive Software Package for Spectral Simulation and Analysis in EPR. J. Magn. Reson. 2006, 178, 42−55. (36) Belous, A.; Ovchar, O.; Valant, M.; Suvorov, D. Abnormal behavior of the dielectric parameters of Ba6−xLn8+2x/3Ti18O54 (Ln= LaGd) solid solutions. J. Appl. Phys. 2002, 92, 3917−3922. (37) VandeVondele, J.; Krack, M.; Mohamed, F.; Parrinello, M.; Chassaing, T.; Hutter, J. Quickstep: Fast and Accurate Density Functional Calculations Using a Mixed Gaussian and Plane Waves Approach. Comput. Phys. Commun. 2005, 167, 103−128. (38) The CP2K developers group, http://www.cp2k.org. (39) Hutter, J.; Iannuzzi, M.; Schiffmann, F.; VandeVondele, J. CP2K: Atomistic Simulations of Condensed Matter Systems. Wiley Interdiscip. Rev.: Comput. Mol. Sci. 2014, 4, 15−25. (40) Hartwigsen, C.; Gœdecker, S.; Hutter, J. Relativistic Separable Dual-Space Gaussian Pseudopotentials from H to Rn. Phys. Rev. B: Condens. Matter Mater. Phys. 1998, 58, 3641. (41) Goedecker, S.; Teter, M.; Hutter, J. Separable Dual-Space Gaussian Pseudopotentials. Phys. Rev. B: Condens. Matter Mater. Phys. 1996, 54, 1703. (42) Krack, M. Pseudopotentials for H to Kr Optimized for GradientCorrected Exchange-Correlation Functionals. Theor. Chem. Acc. 2005, 114, 145−152. (43) Perdew, J. P.; Burke, K.; Ernzerhof, M. Generalized Gradient Approximation Made Simple. Phys. Rev. Lett. 1996, 77, 3865. 27410


Article

The Journal of Physical Chemistry C (44) Grimme, S.; Antony, J.; Ehrlich, S.; Krieg, H. A Consistent and Accurate Ab Initio Parametrization of Density Functional Dispersion Correction (DFT-D) for the 94 Elements H-Pu. J. Chem. Phys. 2010, 132, 154104. (45) Neese, F. The ORCA Program System. Wiley Interdiscip. Rev.: Comput. Mol. Sci. 2012, 2, 73−78. (46) Neese, F. Theoretical Study of Ligand Superhyperfine Structure. Application to Cu (II) Complexes. J. Phys. Chem. A 2001, 105, 4290− 4299. (47) Neese, F. Prediction of Electron Paramagnetic Resonance g Values Using Coupled Perturbed Hartree-Fock and Kohn-Sham Theory. J. Chem. Phys. 2001, 115, 11080−11096. (48) Neese, F. Metal and Ligand Hyperfine Couplings in Transition Metal Complexes: the Effect of Spin-Orbit Coupling as Studied by Coupled Perturbed Kohn-Sham Theory. J. Chem. Phys. 2003, 118, 3939−3948. (49) Kaupp, M.; Reviakine, R.; Malkina, O. L.; Arbuznikov, A.; Schimmelpfennig, B.; Malkin, V. G. Calculation of Electronic gTensors for Transition Metal Complexes Using Hybrid Density Functionals and Atomic Meanfield Spin-Orbit Operators. J. Comput. Chem. 2002, 23, 794−803. (50) Neese, F. Prediction and Interpretation of the 57Fe Isomer Shift in Mössbauer Spectra by Density Functional Theory. Inorg. Chim. Acta 2002, 337, 181−192. (51) Carl, P. J.; Larsen, S. C. EPR Study of Copper-Exchanged Zeolites: Effects of Correlated g-and A-Strain, Si/Al Ratio, and Parent Zeolite. J. Phys. Chem. B 2000, 104, 6568−6575. (52) Hathaway, B.; Billing, D. The Electronic Properties and Stereochemistry of Mono-Nuclear Complexes of the Copper (II) Ion. Coord. Chem. Rev. 1970, 5, 143−207. (53) Atherton, N. Principles of Electron Paramagnetic Resonance; Ellis Horwood, Ltd.: London, 1993. (54) Gul-E-Noor, F.; Jee, B.; Mendt, M.; Himsl, D.; Pöppl, A.; Hartmann, M.; Haase, J.; Krautscheid, H.; Bertmer, M. Formation of Mixed Metal Cu3-xZnx(btc)2 Frameworks with Different Zinc Contents: Incorporation of Zn2+ into the Metal-Organic Framework Structure as Studied by Solid-State NMR. J. Phys. Chem. C 2012, 116, 20866−20873.

27411
