A Combined Pulsed Electron Paramagnetic Resonance Spectroscopic

Mar 28, 2013 - Nickel Based Paddle-Wheel Metal–Organic Frameworks Towards Adsorption of O3 and SO2 Molecules: Quantum-Chemical Calculations. Ali Sho...
6 downloads 6 Views 2MB Size
Article pubs.acs.org/JPCC

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 Bettina Jee,† Petko St. Petkov,‡,¶ Georgi N. Vayssilov,‡ Thomas Heine,¶ Martin Hartmann,§ and Andreas Pöppl*,† †

Institut für Experimentelle Physik II, Universität Leipzig, 04103 Leipzig, Germany Faculty of Chemistry and Pharmacy, University of Sofia, 1126 Sofia, Bulgaria ¶ School of Engineering and Science, Jacobs University Bremen, 28759 Bremen, Germany § Erlangen Catalysis Research Center (ECRC), Universität Erlangen-Nürnberg, 91058 Erlangen, Germany ‡

S Supporting Information *

ABSTRACT: Cu3(btc)2 (btc = 1,3,5-benzenetricarboxylate), also called HKUST-1, is one of the well-known representatives of the metal−organic framework (MOF) compounds. It exhibits a large surface area and a high pore volume. Due to the coordinatively unsaturated metal centers as preferential adsorption sites, Cu3(btc)2 is particularly interesting for the separation of CO2 and CO in gaseous mixtures. We studied the interactions of 13C-enriched carbon dioxide (13CO2) and carbon monoxide (13CO) with the Cu2+ centers in the zincsubstituted homologue Cu2.97Zn0.03(btc)2 using continuous wave (cw) and pulsed electron paramagnetic resonance (EPR) spectroscopy (Davies or Mims electron nuclear double resonance (ENDOR) and hyperfine sublevel correlation (HYSCORE)). Upon adsorption of 13CO2 and 13CO, the coordination geometry of the Cu2+ centers changed from square planar to square pyramidal. The cupric ion g-tensor and the 63/65Cu hyperfine coupling tensor ACu show the changes in the ligand field of Cu2+. Moreover, the interaction with the 13C nuclei of the gas molecules is reflected in the isotropic coupling constant ACiso and the dipolar coupling parameter TC⊥ which are derived from the 13C hyperfine coupling tensor AC obtained by the pulsed EPR experiments. From the experimentally obtained parameters, we derived a geometrical model for the adsorption of 13CO2 and 13CO at the Cu2+ ions that is consistent with our DFT calculations. The 13 CO molecule is found to coordinate linearly at the Cu2+ center via the 13C atom and perpendicular to the CuO4 plane with a Cu−C distance of rCuC = 2.57(10) Å (DFT, 2.42 Å). The 13CO2 molecule is coordinated slightly tilted via the O atom with a Cu−C distance of rCuC = 3.34(10) Å (DFT, 3.27 Å). The Cu−O distance for adsorbed 13CO2 is not directly accessible to EPR measurements but could be estimated from geometrical considerations in the range of rCuO = 2.53−2.73 Å (DFT, 2.39 Å). The results provide detailed insight into the geometry of adsorbed CO2 and CO in porous materials and show the potential of EPR spectroscopy for analyzing adsorption complexes.


after activation of the material. It has been calculated that polar and nonpolar molecules have different interaction energies with the coordinatively unsaturated copper centers.9−11 Thus, Cu3(btc)2 is studied experimentally and theoretically with respect to CO2 storage12−14 and the separation of CO2 and CO from industrial gas mixtures.15,16 Recently, Cu3(btc)2 has been implemented in so-called mixed-matrix membranes exhibiting considerable enhancement in terms of selectivity and permeability compared with the pure polymer membrane in the separation of CO2 from N2 and CH4 in binary mixtures.17,18

Cu3(btc)2 (btc = 1,3,5-benzenetricarboxylate), also called HKUST-1, is one of the first and best investigated metal− organic frameworks (MOFs).1 This class of porous coordination polymers gained high interest in recent years2 because of their potential applications in many industrial processes.3−5 Especially the desired storage and separation of the “greenhouse gas” CO2 induced many experimental and theoretical studies focused on the exceptional adsorption properties of metal−organic frameworks.6,7 Cu3(btc)2 is structurally well characterized and shows remarkable adsorption properties due to its high surface area and the presence of coordinatively unsaturated metal centers at the Cu/Cu paddle wheel units in the framework.1,8 These potential adsorption sites are accessible for sorbate molecules © 2013 American Chemical Society

Received: January 10, 2013 Revised: March 28, 2013 Published: March 28, 2013 8231

dx.doi.org/10.1021/jp4003033 | J. Phys. Chem. C 2013, 117, 8231−8240

The Journal of Physical Chemistry C


Mims35 electron nuclear double resonance (ENDOR) and by hyperfine sublevel correlation (HYSCORE) experiments36 at 7 K. Corresponding DFT calculations were carried out on large saturated model complexes incorporating the mixed Cu/Zn paddle wheels and the linking btc ligands.

The adsorption of carbon monoxide on Cu3(btc)2 has been studied by infrared (IR) spectroscopy using the change in intensity and the shift in wavenumber of the vibrational bands upon adsorption of different amounts of CO to characterize the adsorption complexes formed.19−22 The formation of dicarbonyls at the Cu2+ ions has been suggested,20,21 and the discussion of a potential reduction of CuII to CuI by carbon monoxide is still continuing.22,23 Using IR spectroscopy, the adsorption of CO at CuII and at CuI, which was present after synthesis or after activation at high temperatures, could be distinguished.19,21,24,25 At low temperatures, CO was found to coordinate via the O atom presumably interacting with the organic part of the framework.21 In the IR spectroscopic studies of the adsorption of CO2 in Cu3(btc)2, two cupric ion adsorption sites were discussed: the Cu2+ ions of Cu/Cu paddle wheel units at the external surface and the ones located inside the pores.14,19 Neutron diffraction experiments indicated four possible arrangements for CO2 adsorbed at the Cu2+ corresponding to the 4-fold symmetry of the paddle wheel units.26 To obtain information about the coordination geometry of adsorbed CO2 and CO molecules at the copper ions, we carried out a combined experimental−theoretical study by applying electron paramagnetic resonance (EPR) spectroscopy and density-functional theory (DFT) calculations. The Cu/Cu pairs in the paddle wheel units are known to be antiferromagnetically coupled with an excited spin state S = 1. The EPR silent S = 0 ground state dominates below 70 K as shown previously by magnetic measurements27 and EPR spectroscopic investigations.28 These experimental observations are in agreement with early theoretical predictions and recent multireference calculations.29−31 According to a common procedure in EPR spectroscopy,32 we have substituted about 1% of the copper ions by zinc ions resulting in the Zn-substituted compound Cu2.97Zn0.03(btc)2 with a few statistically distributed mixed Cu/ Zn paddle wheel units in the framework as previously reported.33 At low temperatures, the Cu/Cu pairs are in their EPR-silent S = 0 state, whereas the Cu2+ ions of the mixed Cu/Zn paddle wheel units give the typical anisotropic EPR powder pattern of cupric ions with well-resolved hyperfine splitting (hfs) due to the 63/65 Cu nuclei having a nuclear spin ICu = 3/2. As we have shown previously, the Cu2+ EPR parameters are sensitive to structural changes in the coordination sphere of the copper atom in the mixed Cu/Zn paddle wheel units occurring upon adsorption and desorption of methanol.33 Using pulsed EPR experiments, we obtained information about the coordinating atoms and the geometry of the copper−adsorbate complex. Hereby, we assume that the guest molecules would coordinate similarly to the copper centers of mixed Cu/Zn paddle wheel units as to the copper centers of the parent Cu/Cu paddle wheel units. In this work, we present our continuous wave (cw) and pulsed EPR spectroscopic studies of the adsorption of 13CO2 and 13CO over activated Cu2.97Zn0.03(btc)2 (1). A 2:1 excess of gas molecules to metal ions was adsorbed, which can also be regarded as 24:12 with respect to the number of metal atoms in a unit cell. 13C-enriched carbon monoxide and carbon dioxide were used to enhance the signal intesity due to the 13C nucleus

EXPERIMENTAL SECTION Sample Preparation and Adsorption of 13CO2 and 13 CO. 13CO2 and 13CO were used as purchased from Sigma Aldrich both with isotopic purity of 99% 13C. Zn-substituted HKUST-1 (1) has been prepared and activated as previously reported and stored under nitrogen according to a well-known procedure.8,33 For adsorption of 13CO2 (2), 22.2 mg (3.67 × 10−5 mol) of Cu2.97Zn0.03(btc)2 was placed in a quartz EPR tube with a Teflon valve and activated 12 h in vacuum. We used an activation temperature of 393 K, at which the reduction to CuI observed after activation at 474 K21,22 does not play a significant role. After activation, 9.9 mg of 13CO2 (2.2 × 10−4 mol) was adsorbed at 77 K. For adsorption of 13CO (3), we used 6.7 mg (1.1 × 10−5 mol) of Cu2.97Zn0.03(btc)2 and 1.93 mg (6.6 × 10−5 mol) of 13CO under the same conditions. After adsorption of the respective gases, the EPR tubes were sealed immediately. EPR Spectroscopy. All X-band cw EPR, two-pulse fieldswept electron spin echo (FS-ESE), pulsed ENDOR, and HYSCORE experiments were recorded on a Bruker ELEXYS E580 spectrometer at room temperature or at 7 K. Temperaturedependent cw EPR measurements were performed from 7 to 393 K in small steps (5−20 K) with about five minutes equilibrium time for temperature stabilization using an Oxford Instruments flow cryostat. Two pulse FS ESE and HYSCORE36 spectra were measured at X-band at 6 K using nonselective microwave (mw) pulses of tπ/2 = 16 ns and tπ = 32 ns. The (170 × 170) HYSCORE spectra were recorded with a pulse delay of τ = 136 ns at 340.8 mT to enhance the modulation signals from 13C nuclei.37 Two-dimensional (2D) Fourier transformed (FT) magnitude spectra are displayed. Pulsed ENDOR experiments of sample 3 were performed using the Davies ENDOR sequence34 applying mw pulse lengths of tπ/2 = 100 ns and tπ = 200 ns with a pulse delay τ = 1000 ns between the second and third mw pulse. The Mims ENDOR sequence35 with mw pulses of tπ/2 = 100 ns and a pulse delay τ = 800 ns between the first and the second mw pulse was employed for sample 2. The lengths of the radio-frequency (rf) pulses were trf = 10 μs in Davies ENDOR experiments and trf = 20 μs in Mims ENDOR experiments. Spectral simulations are based on the spin Hamiltonian (eq 1). Cu Cu C /̂ = βeB0⃗ ·g ·Ŝ + Ŝ ·ACu ·I ̂ − βngnB0⃗ ·I ̂ + Ŝ ·AC·I ̂ C

− βngnB0⃗ ·I ̂


The first and third term refer to the Zeeman splitting in an external magnetic field B0 of the electron spin S and the nuclear spin ICu of copper atoms, respectively (βe and βn are Bohr’s magneton and nuclear magneton; gn is nuclear g-factor), the second and fourth term describe the hyperfine coupling (hfc) with the nuclear spins of the 63/65Cu isotopes (ICu = 3/2) and 13C atoms (IC = 1/2), respectively, and the last term takes the Zeeman splitting of the nuclear spin IC of 13C atoms into account.38 The principal values of the g-tensor and the hfc tensor ACu are obtained by cw EPR spectroscopic measurements and reveal information about the coordination geometry and interactions at the Cu2+ ions. Because both isotopes, 63Cu and


with I C = 1/2, assuming that the adsorption behavior of the 13Cenriched gases is the same as for carbon monoxide and carbon dioxide with 13C atoms in natural abundance. The samples of activated Cu2.97Zn0.03(btc)2 with adsorbed 13CO2 (2) and 13CO (3) were investigated by cw EPR spectroscopy in a broad temperature range from 7 to 393 K, by pulsed Davies34 and 8232

dx.doi.org/10.1021/jp4003033 | J. Phys. Chem. C 2013, 117, 8231−8240

The Journal of Physical Chemistry C


polarization functions.51 Relativistic effects have been included by the zero-order relativistic approximation (ZORA) and spin− orbit corrections.52−54 The calculations of the hfc constants have been carried out with the ADF code. For an accurate calculation of the A-tensor, one needs a very large basis set in the core region, especially for heavy nuclei. Such large basis sets (jcpl basis set from ADF basis set library) were applied for Cu2+, which are basis sets especially designed for EPR A-tensor and NMR spin−spin coupling calculations.


Cu, have the same nuclear spin and only slightly different nuclear gn factors, their hfc is of the same size and therefore not distinguished in all spectra. The principal values of the 13C ligand hfc tensor AC are accessible with pulsed EPR experiments such as HYSCORE and Davies or Mims ENDOR and give more sophisticated information about the adsorption complexes with 13 CO and 13CO2.39 Spectral Simulations. For the simulation of the cw EPR spectra, we used the EasySpin simulation package,40 which takes the first three terms of the spin Hamiltonian (eq 1) into account. The orientation-selective pulsed ENDOR spectra are well represented by calculating the spin Hamiltonian with EasySpin including the hyperfine (hf) and nuclear Zeeman interactions of the 13C nucleus, which are described in the last two terms of eq 1. The simulations of the orientation selective HYSCORE spectra were calculated in the time domain by exact diagonalization of the spin Hamiltonian with a home-written program.41 From the obtained principal values ACxx,yy and ACzz of an axially symmetrical 13 C hfc tensor AC, it is possible to determine the isotropic hfs parameter ACiso and the dipolar hfc parameter TC⊥ according to eq 2. C C C Axx , yy = A iso − T⊥

C C Azz = A iso + 2T⊥C


RESULTS AND DISCUSSION cw EPR Spectroscopy. The cw EPR spectra at room temperature of Cu2.97Zn0.03(btc)2 with adsorbed 13CO2 (2) and with adsorbed 13CO (3) exhibit the broad isotropic signal at g0 = 2.16 (ΔBpp = 67−80 mT), which is already known from the pure Cu3(btc)2 and activated Cu2.97Zn0.03(btc)2 (1) (Figure 1, Supporting Information).28,33 This signal has been previously assigned to the excited S = 1 spin state of the antiferromagnetically coupled Cu/Cu pairs in the paddle wheel units.28 The substitution of one percent Cu2+ ions by Zn2+ ions does not affect the majority of the Cu/Cu pairs and the signal of their S = 1 state is also observable for Cu2.97Zn0.03(btc)2 (1) at room temperature.33 The occurrence of this signal of the excited S = 1 state of the antiferromagnetically coupled Cu/Cu pairs in the paddle wheel units in the room temperature cw EPR spectra of samples 2 and 3 (Figure 1, Supporting Information) indicates that the framework remained intact after adsorption of 13CO2 and 13CO. Any decomposition would destroy the electronically coupled Cu/Cu paddle wheel units resulting in a dramatic decrease of the isotropic line width at room temperature as we previously have monitored by cw EPR spectroscopy after water vapor adsorption in Cu3(btc)2.55 But because of the large line width, this signal comprises no further information about the hf coupling and the anisotropy of the Cu2+ g-tensor. For the adsorption experiments, we used the Zn-substituted material. At low temperatures, the antiferromagnetically coupled Cu/Cu pairs are in the EPR silent S = 0 ground state, whereas the S = 1/2 signal of the paramagnetic mixed-metal Cu/Zn paddle wheel units remains and becomes visible displaying a typical powder spectrum of Cu2+ ions with well-resolved hf coupling with the 63Cu and 65Cu nuclei.33 At 7 K, well-resolved signals of an S = 1/2 spin system coupled to a single nuclear spin ICu = 3/2 were observed in the parent activated sample 1 and in both adsorbed samples 2 and 3 (Figure 1). The principal g-tensor parameters as derived by spectral simulations are typical for axial symmetry at the Cu2+ ion (gxx = gyy ≠ gzz) (Table 1). The simulated cw EPR spectra of samples 2 and 3 at 7 K are shown in Figures 2 and 3, Supporting Information, respectively. After adsorption of 13CO2 (sample 2), the principle value gzz increases significantly to gzz = 2.293 compared with the activated Cu2.97Zn0.03(btc)2 (1). At the same time, the principal value ACu zz −4 of the hf tensor decreases from ACu to 182 × 10−4 zz = 190 × 10 cm−1, indicating an axial distortion of the ligand field. In the spectrum of 2, the signal of sorbate-free Cu2+ ions with square planar geometry of the mixed Cu/Zn paddle wheel units is still present (Figure 2, Supporting Information). In order to influence the ratio between both signals by enhancing a potentially hindered diffusion, sample 2 was kept for 3.5 h at 323 K and subsequently another 2 h at 343 K. After each thermal treatment, the sample was measured again at 7 K, but no changes in the intensity ratios of the two signals could be observed.

(2) 13

The distance r between the electron spin and the coupling C nucleus can be estimated from the dipolar hfc parameter TC⊥ in frequency units using the point-dipole approximation (eq 3): T⊥ =

μ0 gegnβeβn 4π ℏ



with g = gzz cos θ + gxx,yy sin θ and θ being the angle between the external magnetic field B0 and the principal z-axis of the gtensor (μ0: permeability of vacuum).39 All other constants have their usual meanings. To get some initial values for ACxx,yy and ACzz to start the simulation, we derived a rough estimate for TC⊥ from the maximum shift of the cross-peak ridges ΔνSmax from the ν1 = ν2 frequency axis in the 2D HYSCORE spectra according to eq 4 with the Larmor frequency νCL of the 13C nuclei.42 2


T⊥ =

2 3


S 8Δνmax νL 2




The outer end positions of the 13C ridges provide estimates for the principal value ACzz. For further information concerning the simulation procedure, we refer to an earlier paper.41 Computational Details. The specific adsorption site of the material was modeled as CuZn(H2btc)4 cluster saturated with terminal carboxylic acid groups. The structures of the bare cluster model and the complexes with CO and CO2 adsorbates were optimized on grounds of DFT employing the Becke threeparameter hybrid method combined with a LYP correlation functional (B3LYP).43,44 The basis set associated with45 relativistic effective core potentials46 supplemented with polarization f functions47 was employed for description of the electronic structure of Cu and Zn atoms. The standard 6311G(d) basis sets were applied for the H, C, and O atoms. Geometry optimization was performed with the Gaussian09 program suite.48 All CO and CO2 complexes were modeled in doublet state. The hyperfine coupling constants (hfc) for 13C were obtained with single-point calculations using the optimized structures of the complexes at B3LYP level.49 Hfc for 13C were calculated on grounds of DFT, employing the gradient-corrected exchange−correlation functional PBE,50 a triple-ζ basis set with 8233

dx.doi.org/10.1021/jp4003033 | J. Phys. Chem. C 2013, 117, 8231−8240

The Journal of Physical Chemistry C


Upon adsorption of 13CO (sample 3), a stronger axial distortion of the ligand field at the Cu2+ center was indicated Cu by a larger gzz (gzz = 2.300) and a smaller ACu zz value (Azz = 175 × −4 −1 10 cm ). In general, a variation of the principal values of the gand ACu-tensors of the cupric ion indicates a change in the coordination symmetry.56 Especially the increase of gzz and the decrease of ACu zz correspond to a distortion of the square planar coordination geometry toward square pyramidal and further to octahedral. In such a distortion, the ligand field splitting Δ between the dxy and the dx2−y2 orbital is reduced, causing an increase of the gzz principal value.57 Such an increase of gzz is observed in combination with a decrease of ACu zz upon adsorption of 13C-enriched carbon dioxide and carbon monoxide compared with the activated material and confirms that the square planar symmetry of the Cu2+ ion in the Cu/Zn paddle wheel is disturbed by an additional axial ligand field. Consequently, 13CO2 and 13 CO are coordinating to the Cu2+ ion in axial direction of the gtensor. This implies a change in the coordination geometry at the Cu2+ center from square planar to square pyramidal, as expected upon adsorption of molecules at the coordinatively unsaturated Cu2+ sites in mixed Cu/Zn paddle wheel units and as we have found previously for the adsorption of methanol.33 Indeed, the axial distortion of the ligand field appears to be stronger for carbon monoxide than for carbon dioxide but is less strong compared with the adsorption of methanol that resulted in an −4 −1 even stronger decrease of ACu zz to 158 × 10 cm and increase of 33 gzz to 2.336 (Table 1). Therefore, we assume that the adsorption strength might also increase in the same direction 13 CO2 < 13CO < CH3OH. Besides the nicely resolved signal, a broad underlying signal was found in the spectra of both samples 2 and 3 that has been detected also in various intensities in the sorbate-free materials 1. This signal is not fully understood yet and has tentatively been assigned to agglomerated extra-framework Cu2+ species.28,33,58

Figure 1. cw EPR spectra at 7 K of (a) activated [Cu2.97Zn0.03(btc)2]n (1) and 1 after adsorption of (b) 13CO2 (2) and (c) 13CO (3). (νmw = 9.75 GHz).

Table 1. EPR Parameters of the S = 1/2 Spin System of the Cu/Zn Pairs in the Paddle Wheel Units of [Cu2.97Zn0.03(btc)2]n (1) and after Adsorption of 13CO2 (2) and 13CO (3) Compared with 1MeOH with Adsorbed Methanola





b ACu xx,yy

b ACu zz

1act 2CO2

2.046 2.049

2.279 2.293

32 26

19033 182


2.051 2.060

2.300 2.336

23 15

175 15833

−4 −4 Errors: Δgii = ±0.002; ΔACu cm−1. bACu cm−1. ii = ±0.4 × 10 ii in 10

Figure 2. Selected HYSCORE spectra at 7 K of [Cu2.97Zn0.03(btc)2]n after adsorption of 13CO2 (2): experimental (A−C) and simulated (a−c) spectra at 280.0 (A, a), 284.5 (B, b), and 344.5 mT (C, c). For the full set of HYSCORE spectra, see Figure 6, Supporting Information (νmw = 9.75 GHz). 8234

dx.doi.org/10.1021/jp4003033 | J. Phys. Chem. C 2013, 117, 8231−8240

The Journal of Physical Chemistry C


Figure 3. (left) Experimental (black) and simulated (blue) Mims ENDOR spectra at 7 K of [Cu2.97Zn0.03(btc)2]n after adsorption of 13CO2 (2) at (a) 280.0, (b) 284.5, (c) 298.0, (d) 327.5, (e) 339.2, and (f) 344.2 mT. (right) Experimental (black) and simulated (blue) Davies ENDOR spectra at 7 K after adsorption of 13CO (3) at (a) 280.0, (b) 327.5, (c) 340.0, and (d) 344.5 mT. (νmw = 9.72 GHz). The signals are centered at the 13C Larmor frequency.

with the nuclei of the ligands having a nuclear spin I ≠ 0 to elucidate the coordination geometry of adsorbed molecules at Cu2+ ions. In this work, we studied the adsorption complexes of carbon monoxide and carbon dioxide at the Cu2+ center by pulsed EPR spectroscopy using 13C (IC = 1/2) enriched gases to enhance the sensitivity for the interaction of 13CO2 and 13CO molecules with the electron spin of Cu2+ in the mixed Cu/Zn paddle wheel units. In both experiments, HYSCORE and pulsed ENDOR, the intensities of the 13C hyperfine signals observed for both samples 2 and 3 were significantly stronger than those for the activated material 1 with 13C atoms in natural abundance. Furthermore, the signals observed for samples 2 and 3 revealed significantly different principal AC-tensor values from the activated sample 1.33 The pulsed ENDOR and HYSCORE spectra have been measured at different field positions within the anisotropic CuII powder pattern for orientation selection (Figure 5, Supporting Information). Adsorption of 13CO2. For the interaction of 13CO2 with Cu2+, intense signals at the 13C Larmor frequencies νCL with wellresolved 13C cross peak ridges were observed in the 2D HYSCORE spectra. Selected spectra are shown in Figure 2, the full set of orientation selective spectra are displayed in Figure 6, Supporting Information. The maximum shift Δνsmax = 0.03 MHz from the ν1 = ν2 frequency axis found for the ridge from (2.5, 3.5) to (3.5, 2.5) MHz in the spectrum at 280.0 mT(B0∥gzz, Figure 2a) provided a rough estimate for the dipolar coupling constant = 0.48 MHz as derived from eq 4. The initial hf of TC,initial ⊥ = 1.0 MHz was taken from the coupling parameter AC,initial zz maximum frequency spread of the cross peak ridges. Employing eq 2, the initial starting parameter AC,initial xx,yy = −0.4 MHz was used for the simulation. This corresponds to a frequency spread of about 0.7 MHz of the ridge from (4.0, 3.4) to (3.4, 4.0) MHz in the spectrum at 344.5 mT(B0∥gxx,yy, Figure 2c). The spectral

The Cu2+ spin Hamiltonian parameters of both samples 2 and 3 do not change up to 80 K, which is shown in the temperaturedependent measurements (Figure 4, Supporting Information). With increasing temperature, the intensity of the Cu2+ signals of the Cu/Zn paddle wheel units with S = 1/2 decreases as expected from the 1/T-dependence according to the Curie law.57 In the temperature-dependent spectra of sample 2, the signals of the 13CO2 adsorption complex, as well as the signals of sorbatefree Cu2+ centers, are simultaneously observed. Upon increasing the temperature, we did not observe an intensity shift from the signal of Cu2+ with adsorbed 13CO2 in favor of the signal of the sorbate-free Cu2+, as would be expected upon desorption. Therefore, we can exclude that the decrease in the signal intensity in this temperature range is caused by the desorption of 13CO or 13 CO2. Nevertheless, we observe a pronounced line broadening in the spectra for both samples 2 and 3 up to 80 K that is qualitatively described by enhanced dipolar interactions between the isolated S = 1/2 CuII spins of the mixed Cu/Zn paddle wheel units and the more and more populated excited S = 1 spin state of neighbored Cu/Cu pairs. A further quantitative analysis of these EPR line broadenings is beyond the scope of this work. From the temperature-dependent EPR measurements, it is confirmed that the sorbate molecules remain coordinated to the CuII ions at least up to 80 K (Figure 4, left, Supporting Information). Because of the decrease of the signal intensity in addition to the line broadening according to the increasing dipolar interactions with the S = 1 spin states of the Cu/Cu pairs under rising temperature, it was not possible to monitor the adsorption−desorption behavior of the gaseous molecules on the Cu2+ ions at temperatures higher than 80 K by cw EPR spectroscopy. Pulsed EPR Spectroscopy. Orientation selective pulsed ENDOR and HYSCORE experiments can probe the interactions 8235

dx.doi.org/10.1021/jp4003033 | J. Phys. Chem. C 2013, 117, 8231−8240

The Journal of Physical Chemistry C


the coordination resulting in a slightly elongated C−O bond. Considering an angle ∠Cu−O−C between 110° and 123°, as it has been calculated here (119°, see below) and with other DFT methods,14,60 the Cu−O distance can be estimated to be in the range of rCuO = 2.53 Å (123°, β = 18°) to rCuO = 2.73 Å (110°, β = 20°) (Figure 4a). We would like to point out that the Cu−O

refinement based on simulation using the exact diagonalization of the spin Hamiltonian (eq 1) still had a large uncertainty because of the additional strong signal without hf splitting at the Larmor frequencies. It is not clear from the spectral simulations whether this signal arises from the same 13C nuclei or from different 13C nuclei in greater distance. The outer edges of the ridges corresponding to the principal values ACxx,yy and ACzz were sampled more distinctly by pulsed ENDOR spectroscopy because the modulation intensity in the HYSCORE experiment drops to zero near the canonical orientations of the hf coupling tensor. For sample 2, the Mims ENDOR pulse sequence was applied. The resulting spectra at the different field positions are shown in Figure 3, left. With the pulsed ENDOR measurements, it became clear that the signal derives from two different 13C nuclei. Both components could not be reproduced by only one set of 13C hfc parameters, as shown in the simulation (blue line in Figure 3, simulation parameters are given in Table 2). Consequently, the strong signal

Figure 4. Schematic representation of the coordination geometry of 13 CO2 adsorbed at Cu2+ ions in [Cu2.97Zn0.03(btc)2]n derived from pulsed ENDOR and HYSCORE experiments. Both geometries with φ = 0° (a) and φ = 5°−7° (b) are in principle feasible within the range of the error for β and rCuC.

Table 2. Experimentally Obtained and Calculated 13C Hyperfine Coupling Parameters for Cu2.97Zn0.03(btc)2 after Adsorption of 13CO2 (2) and 13CO (3)a sample




β (deg)



rCuC [Å]

2CO2, expt 2CO2, calcd 3CO, expt 3CO, calcd















−2.00 −1.71

−2.00 −1.71

1.50 1.30

0 0

−0.833 −0.707

1.167 1.003

2.57 2.42

distance vector to 13CO2 is only exactly parallel to the C4 axis of the paddle wheel unit when the error range was fully taken into account, but it is well possible that the Cu−O distance vector is slightly tilted by an angle φ of about 5°−7° from the C4 axis (Figure 4b). A tilted coordination of the 13CO2 molecule based on EPR data agrees well with the geometries observed by neutron powder diffraction26 or by the applied computational approach here as shown below (φ = 5°), as well as by other DFT calculations.14,60 However, the distances obtained in this study are slightly larger than other calculated Cu−O distances of rCuO = 2.39 Å here and elsewhere14 or rCuO = 2.29 or 2.49 Å.60 In the EPR experiments, the four different orientations of the carbon dioxide molecule that were found in neutron powder diffraction studies could not be distinguished. The negative value of the isotropic 13C hf coupling parameter ACiso indicates that the hf interaction between the unpaired electron of the cupric ion and the 13C nucleus of 13CO2 is mediated by spin polarization as expected for a ligand coordinating axially to the cupric ion. From the width of the second 13C signal at the Larmor frequency, an upper limit of the dipolar hf coupling constant TC⊥ of the observed distant 13C nuclei was roughly estimated to be 0.123 MHz. This translates into a minimum distance of these 13C atoms from the Cu2+ ion of about 5.5 MHz. The signal is due to 13 CO2 molecules, which are not directly coordinated to the Cu2+ centers and might indicate some 13CO2···13CO2 interactions as suggested also by neutron powder data.26 Since the loading in the experiments described here was in the high coverage region with more than one molecule per metal center (corresponding to more than 12 molecules per 12 metal centers in the unit cell, 12:12), we have to take into account that secondary adsorption sites close to the organic parts of the framework are occupied. Moreover, considerable contributions to the adsorption enthalpy have been calculated for CO2···CO2 interactions.14 The second adsorption sites might be the reason that the diffusion of 13CO2 is hindered and a complete occupation of all Cu2+ sites could not be achieved. We did not find signals for more than one type of Cu2+ coordination site, for example, on the external and internal surfaces as discussed previously based on IR studies,19−21 but we have to point out that we probed only 1% of the CuII adsorption

Errors: ΔACii = ±0.07 MHz, Δβ = ±10°, ΔACiso = ±0.070 MHz, ΔTC⊥ = ±0.070 MHz, ΔrCuC = ±0.10 Å. bACii , ACiso, and TC⊥ in MHz.


at the 13C Larmor frequencies was assigned to a second signal due to distant 13C nuclei. The other signal revealing a hf coupling of about 0.6 MHz (at 344.2 mT, Figure 3f) corresponds to the shoulders of the signals observed in the HYSCORE spectra, Figure 2. From spectral simulations of both orientation selective ENDOR and HYSCORE spectra, we could finally determine the principal values of the 13C hfs tensor AC of this more strongly coupled ligand nuclei (Table 2). Using eq 2, we obtained the isotropic and dipolar hf coupling constants ACiso = −0.060 MHz and TC⊥ = 0.533 MHz, respectively. The angle β between the principal axes of the g-tensor and 13C hf tensor AC was β = 25(10)° a significant deviation from zero. Using the point-dipole approximation in eq 3, we estimated the distance r between the Cu atom and the coupling 13C atom of the adsorbed 13CO2 as rCuC = 3.34(10) Å. The pulsed EPR spectroscopic measurements confirm directly the adsorption of 13CO2 molecules at the Cu2+ open axial coordination sites by the observation of 13C hf interactions and further reveal details about the geometry of the formed adsorption complex. In the case of 13CO2, we could derive the Cu−O distances from geometrical considerations using the obtained Cu−13C distance of 3.34(10) Å and the angle β = 25(10)° between the principal axes of the g- and AC-tensors. Restricting ourselves to chemically reasonable arrangements in which the angle at the oxygen atom ∠Cu−O−C is between 90° and 180, the Cu−O distance based on the EPR results was calculated as rCuO = 2.13 Å for ∠Cu−O−C = 180° and rCuO = 3.12 Å for ∠Cu−O−C = 90° assuming a CO bond length of 1.22 Å.59 Here, we neglected that the bond strength is reduced by 8236

dx.doi.org/10.1021/jp4003033 | J. Phys. Chem. C 2013, 117, 8231−8240

The Journal of Physical Chemistry C


Figure 5. Selected HYSCORE spectra at 7 K of [Cu2.97Zn0.03(btc)2]n after adsorption of 13CO (3): experimental (A−C) and simulated (a−c) spectra at 280.0 (A, a), 284.5 (B, b), and 344.3 mT (C, c). For the full set of HYSCORE spectra, see Figure 7, Supporting Information; simulation parameters are listed in Table 2 (νmw = 9.75 HHz).

frequency from distant 13C nuclei neither in HYSCORE nor in pulsed ENDOR spectra. In the EPR experiments, we observed the coordination of carbon monoxide to a Cu2+ ion. In zeolites, CuII often is reduced to CuI in the presence of CO, and only few studies describe a CuII···CO interaction.61 Any kind of CuI···CO interaction as reported in IR studies19−23 for the adsorption of CO in Cu3(btc)2 is not accessible to EPR experiments. The EPR results clearly indicate that one 13CO molecule coordinates linearly (β = 0°) to the Cu2+ ion via its 13C atom as shown schematically in Figure 6. The linear coordination of 13 CO to Cu2+ ions experimentally observed in this study is in good agreement with the results of the DFT investigations here (see below) and elsewhere.60 The obtained Cu−13C distance of 2.57 Å fits very well with the calculated Cu−C distance of 2.42 Å.

sites that are involved in mixed Cu/Zn paddle wheel units and that the fraction of those that are located at the external surface might be too small to sample for EPR. Adsorption of 13CO. The coordination geometry of carbon monoxide in Cu2.97Zn0.03(btc)2 was also investigated by pulsed EPR experiments. The HYSCORE spectra at 280.0 mT (B0∥gzz), 284.5 mT, and 344.3 mT (B0∥gxx,yy) are shown in Figure 5, the full set of orientation selective HYSCORE spectra are displayed in Figure 7, Supporting Information. The HYSCORE spectrum at 280.0 (Figure 5a) exhibits two strong cross peaks at (3.7, 2.3) and (2.3, 3.7) MHz with a frequency spread corresponding to a hf coupling in z-direction of ACzz = 1.5 MHz. The maximum frequency spread of about 2 MHz of the strong cross peaks at (4.6, 3.1) and (3.1, 4.6) MHz observed in the spectrum at 344.5 mT (Figure 5c) corresponds to a hfc of ACxx,yy = −2.0 MHz. The principal values ACxx,yy and ACzz of the AC-tensor determined from spectral simulations of the whole set of orientation selective HYSCORE spectra, and the angle β between the principal axes of the g-tensor and AC hfc tensor are listed in Table 2 in comparison to the principal values of tensor AC and the angle β after the adsorption of 13CO2. We further applied Davies ENDOR spectroscopy at the different field positions (Figure 3, right) to verify the principal values and also the angle β. The Davies ENDOR spectra could be represented well with the tensor parameters determined by HYSCORE spectroscopy and confirmed in particular the angle β = 0°. The isotropic and dipolar hf coupling constants were calculated to be ACiso = −0.833 MHz and TC⊥ = 1.167 MHz, respectively, employing the relation in eq 2. The hf coupling tensor AC of 13CO and the CuII g-tensor were found to be coaxial (β = 0°). The distance rCuC between the Cu2+ center and the 13C atom was derived from the point-dipole approximation in eq 3 as rCuC = 2.57(10) Å. We like to emphasize that in the case of 13CO adsorption, we did not observe signals at the 13C Larmor

Figure 6. Schematic representation of the coordination geometry of 13 CO adsorbed at Cu2+ ions in [Cu2.97Zn0.03(btc)2]n derived from pulsed ENDOR and HYSCORE experiments. The z-axes of the g- and ACtensor are coaxial (β = 0°). The Cu−13C distance is rCuC = 2.57 Å. 8237

dx.doi.org/10.1021/jp4003033 | J. Phys. Chem. C 2013, 117, 8231−8240

The Journal of Physical Chemistry C


Figure 7. Optimized structures of the model complexes with adsorbed CO and CO2. Color coding: gray, C; red, O; ochre, Cu2+; blue, Zn2+. C The isotropic 13C hf coupling parameter Aiso is negative, indicating also spin polarization for the hf interaction in the 13 CO adsorption complex as found for the 13CO2 complex, but to a larger extent in the case of 13CO due to the linear and direct coordination of the 13C atom to the cupric ion. A direct spin density transfer to the carbon s-orbital leading to a positive ACiso parameter was not observed since the geometrical orientation of the involved dx2−y2 orbital of the Cu2+ ion and the σ*-orbital of carbon monoxide do not coincide for positive and binding orbital C overlap. Altogether, Aiso < 0 indicates likewise an axial 2+ coordination at the Cu ions with a dx2−y2 ground state. A tilted coordination with an angle ∠Cu−C−O < 180° but with the Cu−13C distance vector still aligned along the C4 axis of the paddle wheel unit, is not excluded from EPR data, but it is not common for nonbridging carbonyl ligands and also not emphasized by DFT calculations here and elsewhere.60 However, a side-on coordination of 13CO to the Cu2+ center is obviously not supported by EPR data because in this case the angle β would have a significant deviation from zero. For the formation of dicarbonyls as suggested in IR studies,19 two different sets of 13C hf coupling parameters would be expected to be found in the orientation selective experiments but were not observed under the conditions applied in this study. We also can exclude the possibility of a coordination via the O atom23 to the Cu2+ ions because the short Cu−13C distance obtained from EPR data does not allow the insertion of an O atom. We did not find any indication for different Cu2+···CO adsorption sites as already mentioned for the adsorption of 13CO2, taking again into account that we probe only 1% of the coordinatively unsaturated metal centers that are involved in mixed Cu/Zn paddle wheel units. Thus, we might not be able to distinguish internal from external Cu/Zn paddle wheel units. A possible coordination via the O atom to the organic parts of the framework is not accessible by EPR spectroscopy. However, in the case of 13CO adsorption, signals of a second 13C species at larger distance from the Cu2+ centers were not observed as for 13CO2. Thus, the metal ions seem to be the only adsorption sites in Cu2.97Zn0.03(btc)2. DFT Calculations. The optimized structures (Figure 7) of the complexes with CO and CO2 coordinated to Cu2+ site are very similar to those proposed from the experimental EPR measurement. Selected bond distances and bond angles for the optimized model complexes of bare CuZn(H2btc)4 and of CuZn(H2btc)4CO and CuZn(H2btc)4CO2 are compared in Table 3 in the Supporting Information. The carbon monoxide is coordinated linearly to the metal ion (Cu2+) with angle β = 0°, as it is found in the ENDOR experiment. The carbon dioxide molecule is tilted with respect to the Cu−Zn axis with angle β = 22°, in complete agreement with the experimentally determined angle of 25(10)°, and the angle φ = 5° is also close to the

experimental value of 5°−7°. The CO2 molecule keeps its linearity and is tilted above one of the Cu2+−O bond with a btc ligand molecule. The optimized carbon−copper distances for CO2 and CO adsorbates are shorter by 0.07 and 0.15 Å, respectively, than the corresponding distances estimated from the pulsed EPR data (Table 2). The adsorption of CO does not affect considerably the structure of the paddle wheel unit as the Cu−O distances to the ligand are elongated by 0.01−0.03 Å. This is related to the symmetric coordination of CO along the C4 symmetry axis of the complex and the weak binding of the adsorbates to the Cu2+ ion (26 kJ mol−1 for CO). The Cu−O distances in the paddle wheel structure are affected more strongly in the case of CO2 adsorption. Two of the Cu−O bonds are elongated by 0.1 Å and the corresponding O centers are shifted upward, while the other two Cu−O bonds are shortened by 0.04 Å and the O centers are shifted downward. Due to the CO2 adsorbate, the paddle wheel reduces its symmetry from C4h to C2v. The calculated hf coupling constants for 13C in the adsorbed CO agree well with the experimentally measured values. The 13C hfc tensor was found to be axially symmetric with −1.71 MHz for the ACxx,yy components, while the ACzz component was calculated to be 1.3 MHz. The calculated ACxx,yy and ACzz components are lower by 0.29 and 0.19 MHz, respectively, than the experimentally measured values but represent the major characteristics of the tensor, axial symmetry, and dipolar hyperfine coupling of about 1 MHz, and negative isotropic hyperfine coupling (Table 2). The adsorption energy of CO2 coordinated to the Cu2+ ion at a mixed Cu/Zn paddle wheel unit is slightly stronger (by 3 kJ mol−1) than that of CO. However, the nearly free rotation of CO2, as discussed below, imposes a higher zero-point energy resulting in slightly weaker binding. Similar interaction energies of 29 kJ mol−1 have been found experimentally for both CO and CO2 coordinated to the Cu2+ ions of pure Cu/Cu paddle wheel units.14,25,62 Otherwise, the calculated adsorption energies for Cu/Cu paddle wheel units show a larger variation from 25.8 to 35.0 kJ mol−1 for CO210,14,15,26,63 and from 16.5 to 28 kJ mol−1 for CO.10,60 This indicates that the larger distortion of the ligand field by 13CO as observed by EPR can not straightforwardly be correlated with the adsorption energy of the coordinated CO and CO2 molecules. Further studies are necessary to understand in more detail the interrelationship between ligand fields and adsorption energies at Cu2+. Due to the tilted coordination of CO2 and lower symmetry of the complex, additional sets of calculations were performed. The maximum energy variation for rotation of CO2 around Zn−Cu axis was estimated to be very low, about 2 kJ mol−1 for partially optimized structures and 6 kJ mol−1 for the structures without relaxed paddle wheel constituents. This finding suggests that 8238

dx.doi.org/10.1021/jp4003033 | J. Phys. Chem. C 2013, 117, 8231−8240

The Journal of Physical Chemistry C


investigations. We obtained subtle geometrical details about the adsorption complexes from orientation selective HYSCORE and pulsed ENDOR spectra. A thorough analysis of the EPR data shows that 13CO forms a linear adsorption complex with coordination via the carbon atom of the molecule conserving the C4 symmetry of the paddle wheel units. In the case of 13CO2, a tilted adsorption complex with coordination via the oxygen atom is found. Furthermore, the computed 13C hf coupling tensors agree well with the experimentally obtained tensors. Also, the structural data obtained from EPR experiments are in excellent agreement with DFT predictions. Both 13CO and 13CO2 adsorption complexes are stable up to 80 K. Pulsed EPR data confirm the cw EPR results in which an axial coordination of the adsorbate molecule is observed according to the change in the gtensor parameters. The g-tensor parameters indicate that the distortion of the axial ligand field of Cu2+ ions in the mixed Cu/ Zn paddle wheel units is stronger upon adsorption of 13CO than upon adsorption of 13CO2.

essentially a barrier-free rotation of CO2 around the Cu−Zn axis occurs spontaneously at elevated temperatures. However, the pulsed EPR experiments have been performed at low temperatures where a potential CO2 rotation freezes out. Therefore, the position of the CO2 with respect to the azimuthal angle about the Cu−Zn axis is not well localized, and a random orientation of the adsorbed molecules about the Cu−Zn axis must be expected leading to a distribution of 13C hyperfine coupling values. Taking into account such a disorder in the arrangement of the adsorbed CO2 molecules, the components of 13C hfc tensor were estimated using different orientations of CO2 around Cu−Zn axis, and the ACxx,yy and ACzz components were calculated for five different orientations of CO2 around Zn−Cu axis at 20°, 40°, 60°, 80°, and 90° deviation from the local minimum, where the CO2 molecule is tilted toward Cu−O(btc) bond. The computed 13C hyperfine coupling tensors are slightly orthorhombic with |ACxx − ACxx| < 0.13 MHz (Figure 8). As it can be seen


S Supporting Information *

Additional cw EPR spectra of 1, 2, and 3 at room temperature, simulations of the two Cu2+ species observed in the cw EPR spectrum of 2 at 7 K, simulation of cw EPR spectrum of 3 at 7 K, temperature-dependent cw EPR spectra of 2 and 3, 2p FS ESE spectra of 2 and 3, full sets of orientation-selective HYSCORE spectra of 2 and 3, and calculated interatomic distances for the Cu/Zn paddle wheel units. This material is available free of charge via the Internet at http://pubs.acs.org.


Corresponding Author

*E-mail: [email protected].

Figure 8. Variation of ACxx, ACyy, and ACzz components of the AC-tensor for adsorbed CO2, calculated for different orientations of the CO2 molecule around the Cu−Zn axis. The in-plane components ACxx and ACyy are averaged and presented in Table 2.


The authors declare no competing financial interest.

ACKNOWLEDGMENTS We thank the Deutsch Forschungsgemeinschaft (DFG) for financial support within the priority program 1362 and the joint Bulgarian−German exchange program of DAAD and the Bulgarian Science Fund (Project PROBEMOF). P.St.P. and G.N.V. are grateful to FP7 project BeyondEverest and Bulgarian Science Fund (DCVP 02/1).

in Figure 8, the ACxx and ACyy components

are more sensitive to the position of CO2 around Cu−Zn axis while ACzz component of the 13 C from CO2 remains almost unchanged during the rotation of CO2. For comparison with the experimental data, the calculated 13 C ACxx, ACyy, and ACzz components were estimated as an average on the computed five different orientations and presented in Table 2. However, because we have to assume a distribution of 13C hf couplings for the adsorbed 13CO, such a small orthorhombic distortion is beyond the resolution in the orientation selective HYSCORE and pulsed ENDOR experiments and presumably within the experimental error. The average value of ACii for 13C from CO2 shows that the ACxx,yy components are lower by 0.5 MHz in absolute value than ACzz, which follows the corresponding experimentally observed trend in ACii values of 13C, where the ACxx,yy components are lower by absolute value of 0.4 MHz than ACzz (Table 2).


(1) Chui, S. S.-Y.; Lo, S. M.-F.; Charmant, J. P. H.; Orpen, A. G.; Williams, I. D. Science 1999, 283, 1148−1150. (2) Kitagawa, S.; Matsuda, R. Coord. Chem. Rev. 2007, 251, 2490− 2509. (3) Müller, U.; Schubert, M.; Teich, F.; Pütter, H.; Schierle-Arndt, K.; Pastré, J. J. Mater. Chem. 2006, 16, 626−636. (4) Kuppler, R. J.; Timmons, D. J.; Fang, Q.-R.; Li, J.-R.; Makal, T. A.; Young, M. D.; Yuan, D.; Zhao, W.; Zhou, H.-C. Coord. Chem. Rev. 2009, 253, 3042−3066. (5) Ma, S.; Zhou, H.-C. Chem. Commun. 2010, 46, 44−53. (6) Liu, J.; Thallapally, P. K.; McGrail, B. P.; Brown, D. R.; Liu, J. Chem. Soc. Rev. 2012, 41, 2308−2322. (7) Li, J.-R.; Ma, Y.; McCarthy, M. C.; Sculley, J.; Yu, L.; Jeong, H.-K.; Balbuena, P. B.; Zhou, H.-C. Coord. Chem. Rev. 2011, 255, 1791−1823. (8) Schlichte, K.; Kratzke, T.; Kaskel, S. Microporous Mesoporous Mater. 2004, 73, 81−88. (9) Yang, Q.; Zhong, C. J. Phys. Chem. B 2006, 110, 17776−17783. (10) Karra, J. R.; Walton, K. S. J. Phys. Chem. C 2010, 114, 15735− 15740.

CONCLUSIONS It was possible to derive structural information from the parameters of the AC-tensor obtained by pulsed EPR experiments combined with DFT calculations, and therefore we were able to deduce a model for the coordination geometry of carbon monoxide and carbon dioxide at the Cu2+ centers, which also should hold for the pure Cu compound. The adsorption of 13CO and 13CO2 at axial copper sites is proven by EPR spectroscopic 8239

dx.doi.org/10.1021/jp4003033 | J. Phys. Chem. C 2013, 117, 8231−8240

The Journal of Physical Chemistry C


(11) Karra, J. R.; Walton, K. S. Langmuir 2008, 24, 8620−8626. (12) Millward, A. R.; Yaghi, O. M. J. Am. Chem. Soc. 2005, 127, 17998− 17999. (13) Wilmer, C. E.; Snurr, R. Q. Chem. Eng. J. 2011, 171, 775−781. (14) Grajciar, L.; Wiersum, P.; Llewellyn, A.; Chang, J.-S.; Nachtigall, P. J. Phys. Chem. C 2011, 115, 17925−17933. (15) Cavenati, S.; Grande, C. A.; Rodrigues, A. E.; Kiener, C.; Müller, U. Ind. Eng. Chem. Res. 2008, 47, 6333−6335. (16) Liu, J.; Wang, Y.; Benin, A. I.; Jakubczak, P.; Willis, R. R.; LeVan, M. D. Langmuir 2010, 26, 14301−14307. (17) Basu, S.; Cano-Odena, A.; Vankelecom, I. F. J. J. Membr. Sci. 2010, 362, 478−487. (18) Erucar, I.; Keskin, S. Ind. Eng. Chem. Res. 2011, 50, 12606−12616. (19) Prestipino, C.; Regli, L.; Vitillo, J. G.; Bonino, F.; Damin, A.; Lamberti, C.; Zecchina, A.; Solari, P. L.; Kongshaug, K. O.; Bordiga, S. Chem. Mater. 2006, 18, 1337−1346. (20) Bordiga, S.; Regli, L.; Bonino, F.; Groppo, E.; Lamberti, C.; Xiao, B.; Wheatley, P. S.; Morris, R. E.; Zecchina, A. Phys. Chem. Chem. Phys. 2007, 9, 2676−2685. (21) Drenchev, N.; Ivanova, I.; Mihaylov, M.; Hadjiivanov, K. Phys. Chem. Chem. Phys. 2010, 12, 6423−6427. (22) Szanyi, J.; Daturi, M.; Clet, G.; Baer, D. R.; Peden, C. H. F. Phys. Chem. Chem. Phys. 2012, 14, 4383−4390. (23) Alaerts, L.; Séguin, E.; Poelmann, H.; Thibauld-Starzyk, F.; Jacobs, P.; De Vos, D. E. Chem.Eur. J. 2006, 12, 7353−7363. (24) Petkov, P. S.; Vayssilov, G. N.; Liu, J.; Shekhah, O.; Wang, Y.; Wöll, C.; Heine, T. Chem. Phys. Chem. 2012, 13, 2025−2029. (25) Rubes, M.; Grajciar, L.; Bludsky, O.; Wiersum, A. D.; Llewellyn, P. L.; Nachtigall, P. Chem. Phys. Chem. 2012, 13, 488−495. (26) Wu, H.; Simmons, J.; Srinivas, G.; Zhou, W.; Yildrim, T. J. Phys. Chem. Lett. 2010, 1, 1946−1951. (27) Zhang, X. X.; Chui, S. S.-Y.; Williams, I. D. J. Appl. Phys. 2000, 87, 6007−6009. (28) Pöppl, A.; Kunz, S.; Himsl, D.; Hartmann, M. J. Phys. Chem. C 2008, 112, 2678−2684. (29) Vogiatzis, K. D.; Klopper, W.; Mavrandonakis, A.; Fink, K. Chem. Phys. Chem. 2011, 12, 3307−3319. (30) Hay, P. J.; Thibeault, J. C.; Hoffmann, R. J. Am. Chem. Soc. 1975, 97, 4884−4899. (31) De Loth, P.; Cassoux, P.; Daudey, J. P.; Malrieu, J. P. J. Am. Chem. Soc. 1981, 103, 4007−4016. (32) Kokoszka, G. F.; Allen, H. C. J. Chem. Phys. 1965, 42, 3693−3697. (33) Jee, B.; Eisinger, K.; Gul-E-Noor, F.; Bertmer, M.; Hartmann, M.; Himsl, D.; Pöppl, A. J. Phys. Chem. C 2010, 114, 16630−16639. (34) Davies, E. R. Phys. Lett. A 1974, 47, 1−2. (35) Mims, W. B. Proc. R. Soc. London 1965, 283, 452−457. (36) Höfer, P.; Grupp, A.; Nebenführ, H.; Mehring, M. Chem. Phys. Lett. 1986, 132, 279−282. (37) Höfer, P. J. Magn. Reson., Ser. A 1994, 111, 77−86. (38) Weil, J. A.; Bolton, J. R.; Wertz, J. E. Electron Paramagnetic Resonance; Wiley: New York, 1994. (39) Schweiger, A.; Jeschke, G. Principles of Pulse Electron Paramagnetic Resonance; University Press: Oxford, U.K., 2001. (40) Stoll, S.; Schweiger, A. J. Magn. Reson. 2006, 178, 42−55. (41) Pöppl, A.; Hartmann, M.; Böhlmann, W.; Böttcher, R. J. Phys. Chem. A 1998, 102, 3599−3606. (42) Pöppl, A.; Hartmann, M.; Kevan, L. Appl. Magn. Reson. 1996, 10, 494−506. (43) Lee, C. T.; Yang, W. T.; Parr, R. G. Phys. Rev. A 1988, 37, 785− 789. (44) Becke, A. D. Phys. Rev. A 1988, 38, 3098−3100. (45) Wadt, W. R.; Hay, P. J. J. Chem. Phys. 1985, 82, 284−298. (46) Roy, L. E.; Hay, P. J.; Martin, R. L. J. Chem. Theory Comput. 2008, 4, 1029−1031. (47) Ehlers, A. W.; Böhme, M.; Dapprich, S.; Gobbi, A.; Höllwarth, A.; Jonas, V.; Köhler, K. F.; Stegmann, R.; Veldkamp, A.; Frenking, G. Chem. Phys. Lett. 1993, 208, 111−114. (48) Frisch, M. J. et al. Gaussian 09; Gaussian, Inc.: Wallingford, CT, 2009.

(49) van Lenthe, E.; van der Avoird, A.; Wormer, P. E. S. J. Chem. Phys. 1998, 108, 4783−4797. (50) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996, 77, 3865−3869. (51) van Lenthe, E.; Baerends, E. J. J. Comput. Chem. 2003, 24, 1142− 1156. (52) van Lenthe, E.; Baerends, E. J.; Snijders, J. G. J. Chem. Phys. 1993, 99, 4597−4611. (53) van Lenthe, E.; Baerends, E. J.; Snijders, J. G. J. Chem. Phys. 1996, 101, 9783−9793. (54) van Lenthe, E.; Ehlers, A. E.; Baerends, E. J. J. Chem. Phys. 1999, 110, 8943−8954. (55) Jee, B.; Himsl, D.; Icker, M.; Hartmann, M.; Pöppl, A. Chem. Ing. Tech. 2010, 82, 1025−1029. (56) Hathaway, B. J.; Billing, D. E. Coord. Chem. Rev. 1970, 5, 143−207. (57) Atherton, N. M. Principles of Electron Spin resonance; Ellis Horwood Limited: Chichester, U.K., 1993. (58) Gul-E-Noor, F.; Jee, B.; Mendt, M.; Himsl, D.; Pöppl, A.; Hartmann, M.; Haase, J.; Krautscheid, H.; Bertmer, M. J. Phys. Chem. C 2012, 116, 20866−20873. (59) Holleman, A. F.; Wiberg, E.; Wiberg, N. Lehrbuch der Anorganischen Chemie; Walter de Gruyter: Berlin, New York, 1995. (60) Lukose, B.; Supronowicz, B.; Petkov, P. S.; Frenzel, J.; Kuc, A. B.; Seifert, G.; Vassilov, G. N.; Heine, T. Phys. Status Solidi B 2012, 249, 335−342. (61) Hadjiivanov, K.; Knötzinger, H. J. Catal. 2000, 191, 480−485. (62) Moellmer, J.; Moeller, A.; Dreisbach, F.; Glaeser, R.; Staudt, R. Microporous Mesoporous Mater. 2011, 138, 140−148. (63) Wang, Q. M.; Shen, D.; Bülow, M.; Lau, M. L.; Deng, S.; Fitch, F. R.; Lemcoff, R. O.; Semanscin, J. Microporous Mesoporous Mater. 2002, 55, 217−230.


dx.doi.org/10.1021/jp4003033 | J. Phys. Chem. C 2013, 117, 8231−8240