Structural Transitions of the Metal-Organic Framework DUT-49(Cu) upon Physi- and Chemisorption Studied by in Situ Electron Paramagnetic Resonance Spectroscopy.

Flexible metal-organic frameworks (MOFs) exhibit a variety of phenomena attractive for basic and applied science. DUT-49(Cu) is one of the remarkable representatives of such MOFs, where phase transitions are coupled to pressure amplification and "negative gas adsorption". In this work we report important insights into structural transitions of DUT-49(Cu) upon physi- and chemisorption of gases and volatile liquids obtained by in situ electron paramagnetic resonance (EPR) spectroscopy. In this method, phase transitions are detected via the zero-field splitting in dimeric copper(II) units. First, a new approach was validated upon physisorption of n-butane. Then, using diethyl ether, we for the first time demonstrated that chemisorption can also trigger phase transition in DUT-49(Cu). On the basis of the EPR results, the transition appears completely reversible. The developed EPR-based approach can also be extended to other flexible MOFs containing paramagnetic metal paddlewheels, where high sensitivity and spectral resolution allow in situ studies of stimuli-induced structural variability.

Metal–organic frameworks (MOFs) attract tremendous attention in chemistry and materials science.[1−5] They are crystalline porous compounds built of metal ions and organic linkers, and the diversity of these building blocks results in a wide structural variety of MOFs and their properties.[6−8] MOFs are actively studied in fields of adsorption,[3,9−16] sensing,[17,18] catalysis,[19−23] etc.

Stimuli-induced structural switchability is one of the interesting properties of MOFs.[24−27] “Breathing” MOFs are able to undergo reversible contraction and reopening transitions upon adsorption/desorption of guest molecules,[28] or upon application of external stimuli such as irradiation, pressure,[6,26,29−31] etc. This phenomenon is explained by the bistability in their free energy landscape with at least two minima, which dynamically change upon guest loading.[32,33] DUT-49(Cu) stands out against other flexible MOFs, because this MOF for the first time evidenced pressure amplification upon adsorption of fluids under certain conditions.[1] The overloading of the metastable state of the DUT-49(Cu) leads to a sudden contraction of the lattice and partial desorption of guests.[34,35] The phenomenon was called “negative gas adsorption” (NGA) due to its highly unusual manifestation in the sorption isotherms.[1] Later on, it was found that this effect is highly sensitive to the crystallite size of the MOF, being thus a highly cooperative phenomenon.[2,36]

Various structural and spectroscopic methods are routinely used to study phase transitions in flexible MOFs, including XRD, NMR, scattering and optical techniques, etc.[26,37] For instance, high pressure 129Xe NMR was used to detect phase transitions in DUT-8(Ni)[38] and in DUT-49(Cu).[38,39] X-ray diffraction and scattering techniques were actively used to investigate stimuli-induced structural hysteresis in MIL-53(Al) and other flexible MOFs.[24,26]

EPR spectroscopy has not yet become a routine tool for studying MOFs; however, its applications in this field are rapidly developing.[40−42] Bare MOFs with paramagnetic metals, embedded or adsorbed radicals, were actively studied over the past decade.[10,43−45] In particular, structural transitions in flexible MOFs such as MIL-53 and some others, were addressed.[28,46−51] Several promising applications of in situ EPR to MOFs were recently reported,[52−54] including a study of Cu-doped DUT-8(Ni).[47] DUT-49(Cu), to the best of our knowledge, was never studied using EPR. Since DUT-49(Cu) includes paramagnetic copper(II) units (paddlewheels, PWs), their EPR spectroscopy can potentially serve as a source of information on occurring structural transitions and accompanying NGA/pressure amplification phenomena. In fact, the role of the metal center in the flexibility in DUT-49 has only recently been addressed[55] and was so far only investigated in situ by EXAFS spectroscopy.[1]

Therefore, in this work we applied EPR spectroscopy to investigate structural transitions in DUT-49(Cu) induced by adsorption of various guest molecules. Two principle lattice states are known for this MOF, corresponding to open and closed pores, op and cp states (Figure a).[1] First, we developed and verified the methodology for in situ measurement of op ↔ cp transitions induced by the physisorption of n-butane. Next, we applied this methodology to study the chemisorption of diethyl ether, Et2O, that, contrary to n-butane, is able to directly coordinate to the unsaturated copper(II) sites in the PWs. To the best of our knowledge, this is the first example of using such agents to induce phase transitions in DUT-49(Cu). As will be shown below, in both cases in situ EPR spectroscopy provides excellent means for monitoring structural rearrangements and studying guest–host interactions.

Figure 1

(a) Structure of DUT-49(Cu) in open and closed pore states (op/cp) and structure of the copper(II) PW unit (Cu, light-blue; O, red; C, gray). (b) X-band EPR spectrum. (c) Q-band EPR spectrum of activated bare MOF (op state) at room temperature. Red lines show simulations using the ZFS values D = 318 mT, E = 0, g = [2.045 2.045 2.285], and copper(II) hyperfine interaction [3.03 3.03 9.39] mT. The asterisk (*) denotes the background signal of X-band resonator; “#” marks minor admixtures (<1%) of monomeric copper(II).

The synthesis of the ligand and DUT-49(Cu) was performed using the procedure reported previously.[2] The details of EPR and sorption isotherms measurements are given in Supporting Information. All simulations used EasySpin.[56]

In the first step, we investigated the principle capabilities of EPR spectroscopy to detect phase transitions in DUT-49(Cu). Room-temperature X/Q-band EPR spectra (Figure ) correspond to a classical copper(II) dimer with appreciable zero-field splitting (ZFS). They can be simulated using spin Hamiltonianwhere S = 1 is the spin of a dimer, B is the magnetic field, ĝ is the g-tensor, β is the Bohr magneton, D and E are parameters of the ZFS, I = 3/2 are nuclear spins or coppers, and  is the hyperfine interaction tensor. When D ≫ E, B, Â, the splitting between two major spectral features (e.g., ∼100 and ∼470 mT in Figure b) is close to D.

The observation of the characteristic dimeric EPR spectrum indicates that copper(II) PW units are well magnetically isolated from each other, so that exchange coupling between them can be neglected. This agrees well with sufficient separation of at least 9.46 Å, provided by a long organic linker in DUT-49(Cu);[57] however, in some other PW-containing MOFs this was not the case, and at room temperatures exchange-narrowed spectra were observed, limiting the explanatory power of spectral signatures.[45,58] Since this is not happening for DUT-49(Cu), we can use ZFS parameters as potential source of information on the lattice state (op vs cp states, Figure a).

In principle, ZFS parameter D in dimers is contributed by dipole–dipole interaction (Ddip) and anisotropic exchange interaction (Dex), which can be approximated as[58,59]where g are the components of g-tensor, Δg is characteristic deviation of the g values, g = (g + g + g)/3, rCu–Cu is the distance between two copper ions, and J is the isotropic exchange interaction between two copper ions known to be ∼200 cm–1 for the PWs.[60,61] Both terms Ddip and Dex might change if the geometry of the PW changes upon phase transition. On the basis of EXAFS (rCu–Cu ≈ 2.471 A) and EPR data (g = [2.045 2.045 2.285], caption to Figure ), Ddip can be calculated as ∼225 mT. Thus, Ddip is ∼2/3 of the total D = 318 mT obtained from simulation, and the contributions of Ddip and Dex are comparable. Therefore, it is rather difficult to predict a priori how structural changes during phase transitions in DUT-49(Cu) would affect the resulting D value, and the benchmarking is necessary.

n-Butane adsorbed at 298 K induces op → cp phase transition in DUT-49(Cu) accompanied by pronounced NGA in the pressure range p = 300–350 mbar (p/p0 ∼ 0.15) (Figure ).[1]

Figure 2

Physisorption isotherms of n-butane on DUT-49(Cu) at 298 K (a) and 273 K (b) (insets: the ranges of existence of op and cp phases).

Therefore, we first performed in situ EPR experiments with n-butane as adsorptive at temperatures close to ambient (Figure and Figure S3). Figure a shows that a stepwise increase of pressure at 298 K results in a shift of the spectral feature at ∼65 mT to the lower field, because the ZFS value D increases from D = 318 mT (0 mbar) to D = 320.6 mT (400–800 mbar) (see the Supporting Information for simulations). The major spectral change occurs at p/p0 ∼ 0.1–0.15, which is in good agreement with the op → cp transition detected in sorption isotherms (Figure a). Therefore, we can safely assign the EPR spectrum with D = 320.6 mT to the cp state of DUT-49(Cu).

Figure 3

X-band CW EPR spectra of DUT-49(Cu) as a function of the applied pressure of n-butane (values on the right) at 298 K (a) and 273 K (b). Microwave frequency νmw ≈ 9.39 GHz. The sample was activated (kept for 12 h under 10–3 mbar at 423 K) prior to measurement. Color dotted lines guide the eye for spectral transformations. Blue shading corresponds to the initial (after activation) op state, red, to the cp state, and green, to the reopened op state (re-op) at high pressure of n-butane. The values of D corresponding to each state are indicated (see Supporting Information for simulations).

Figure b shows the same experiment with sorption of n-butane at 273 K. In this case the range of the p/p0 values is broader. First, we again observe the op → cp transition between p/p0 = 0.1 and 0.2, which nicely agrees with Figures and 3a. At p/p0 ∼ 0.4–0.5 we observe one more change in the EPR spectrum that corresponds to the decrease of D to 315.5 mT. This range of p/p0 agrees reasonably well with the reopening of the pores of MOF (Figure b), and thus D = 315.5 mT can be assigned to the reopened (re-op) state.

Figure S3 shows the data at intermediate temperature ∼293 K that also manifest both op → cp and cp → re-op transitions at corresponding values of p/p0. The experimental accuracy of p/p0 is 0.01–0.015 in this temperature range; therefore, the agreement of critical pressure where each transition occurs is satisfactory (Figures and 3 and Figures S3 and S4). The low-field spectral feature at ∼65 mT is most sensitive to the op → cp → re-op transitions. The change of the PW geometry upon op → cp transition is small, and Cu–Cu distance increases by ∼0.012 Å (∼0.5%, EXAFS[1]). According to eq , such change should decrease the Ddip value by ∼4 mT, whereas we observe experimentally the opposite trend. This implies that the increase of D is due to an overwhelming increase of Dex term (see the Supporting Information for details).

The signal shift is not abrupt as a function of pressure, which might originate from the distribution of the particle sizes, as was outlined previously.[2] Interestingly, according to XRD, the geometry of PWs is similar in initial op and re-op states; however, the EPR spectra (and D values) are slightly different. There is a good matching between p/p0 values where EPR spectrum changes and the known op → cp → re-op transitions; therefore, we assume that EPR spectra are extremely sensitive to a very subtle structural differences in op state that occur at different pressures of n-butane (similar behavior is also observed for other fluid, Et2O; see below).

Thus, the study of n-butane sorption shows that CW EPR spectroscopy is capable of detecting the op → cp → re-op transitions and structural perturbation in DUT-49(Cu) in situ.

Most plausibly, n-butane molecules are physisorbed in the pores of DUT-49(Cu) without being directly coordinated to the copper(II) in the PW. However, PW units of DUT-49(Cu) have open coordination sites able to coordinate guest molecules.[62] We have examined this possibility of guest chemisorption using diethyl ether (Et2O). Prior to in situ EPR studies, adsorption/desorption of Et2O vapor on DUT-49(Cu) at 298 K was accessed ex situ (Figure a).

Figure 4

(a) Sorption of Et2O on DUT-49(Cu) at 298 K. (b) X-band CW EPR spectra vs pressure (indicated) upon in situ Et2O sorption at 298 K. The experiment started with activated sample, then the pressure of Et2O was increased stepwise up to 640 mbar. νmw ≈ 9.87 GHz. Color dotted lines guide the eye for spectral transformations. Blue shading corresponds to the initial op state, red, to the cp* state, and green, to the reopened op* state at high pressure of Et2O. The D values of each state are indicated (see the Supporting Information for simulations). Inset in (a) shows the colors of the sample in the EPR tube with/without (Ads./Des.) Et2O supplied.

The adsorption behavior in this case is similar to the adsorption of n-butane at 273 K; only the op → cp phase transition occurs at p = 54 mbar without any signs of NGA. The structure reopens at p = 270–450 mbar and reaches the plateau, in which the existence of op phase is anticipated. Desorption of the Et2O from the framework is characterized by the op → cp transition at p = 100–150 mbar.

EPR signal change was stronger vs pressure upon adsorption of Et2O compared to n-butane (Figure b). Note that the position of the low-field spectral feature that we monitor can differ in different series of experiments, due to the difference in microwave frequency (νmw) used. However, each simulation yields D value as an independent spectroscopic parameter correlated with the geometry of the PW unit (see the Supporting Information), which is independent of microwave frequency; e.g., D = 318.0 mT is found for activated op state in any experiment.

Starting with the op state (Figure b), the first transition to a new state is obtained at pressure of 50 mbar (p/p0 = 0.08) and manifests itself in the shift of the ZFS component from ∼100 to ∼50 mT (corresponds to the closed pores and is marked as cp*, Figure b) and an increase of D from 318 mT to 341.5 mT. We use here a designation cp* to distinguish this state from the cp state upon n-butane sorption, as the two states have strongly different D values; below we justify that the states marked with (*) refer to the MOF with Et2O coordinated to the PW. As the pressure of Et2O further increases up to 400 mbar, the MOF undergoes reopening of the pores and arrives at the op* state with D = 337.5 mT. The D values for both cp* and op* states are drastically different from those obtained upon n-butane sorption. Moreover, the g component of the g-tensor exhibits a noticeable increase upon adsorption of Et2O (g = 2.295 in op state and 2.340 in cp*/op* states; see Supporting Information), whereas in case of n-butane the corresponding changes of the g-tensor are negligible. The four-coordinated copper(II) ion in PW has a distorted planar geometry, for which g values are typically smaller than those for five- and six-coordinated copper(II) ions.[63,64] Therefore, the increase of the g component upon adsorption of Et2O supports the assumption that Et2O molecule is axially coordinated to the PW. Moreover, according to eq , such an increase of g should lead to the increase of both terms Ddip and Dex. This is in perfect agreement with much more pronounced increase of D upon chemisorption of Et2O vs physisorption of n-butane (see the Supporting Information for more details).

Finally, the adsorption of Et2O in DUT-49(Cu) was accompanied by a much stronger color change (from bright violet to light blue; see Figure a, inset) compared to sorption of n-butane. Thus, taking into account all above arguments, we assume that Et2O molecules occupy open copper(II) coordination sites in the PWs and induce changes in geometry and electronic structure of the PWs.

As follows from Figure , the MOF with coordinated Et2O undergoes phase transition at lower relative gas pressure p/p0 (p < 50 mbar, p/p0 < 0.08) than the same transition in the case of noncoordinating solvent (p ∼ 300 mbar, p/p0 ∼ 0.12–0.15).[1] It seems that the chemisorption might lead to a more efficient pore filling,[34] thus initiating the op → cp* phase transition at lower p/p0.

Remarkably, chemisorption of Et2O was found to be reversible based on EPR results (Figures and 5). Et2O could be successfully desorbed upon activation at high vacuum and room temperature (1 × 10–3 mbar, 298 K) without MOF being destroyed (spectrum reverted, color reverted). The complete desorption of Et2O from DUT-49(Cu) required evacuation at 1 × 10–3 mbar for 3 days (see below, Figure ), while with n-butane it occurred just immediately (Figure ). This also confirms that we observe chemisorption with Et2O, while for n-butane there is no coordination to open metal sites and pure physisorption takes place. Note that, prior to our experiments in the present work, there were no examples of reversible chemisorption in DUT-49(Cu). However, this assignment is based on EPR spectra and does not necessarily reflect changes in the long-range order of the crystal structure. In fact, the hysteresis in the isotherm at low pressure indicates significant differences along the adsorption and desorption trajectory. Further in situ XRD studies may be necessary to confirm reopening of DUT-49 via desorption.

Figure 5

X-band CW EPR spectra vs time (indicated) upon in situ Et2O sorption at 298 K. The experiment was started with activated sample, and then the pressure of Et2O was applied: (a) 25 mbar (p/p0 = 0.03) and (b) 200 mbar (p/p0 = 0.3). After 80 min, evacuation of Et2O (10−3 mbar) was applied. νmw ≈ 9.87 GHz. The intensity of the yellow shading illustrates the amount of adsorbed Et2O, which is at a maximum just before evacuation begins.

As a further support of above conclusions, EPR-detected time-resolved sorption/desorption experiments at two pressures of Et2O were performed: at ∼20–25 and 200 mbar, corresponding to p/p0 ≈ 0.03 and 0.3, respectively. At p/p0 ≈ 0.03 (Figure a) only the adsorption-induced changes op → op* without phase transition should occur (compare with Figure ). Indeed, D changes from 318 to 331.5 mT, which is noticeably smaller compared to op → cp* transition above (318 → 341.5 mT). Thus, the adsorption of Et2O at p/p0 ≈ 0.03 occurs without triggering phase transition. After 80 min of adsorption, evacuation was started, leading to a gradual reverse transformation of the spectrum (Figure a). This behavior was assigned to desorption occurring on a time scale of hours to days.

In the case of higher Et2O pressure of 200 mbar (Figure b), upon adsorption we observe gradual conversion of op state into superposition of two other states. One of them has the same D = 331.5 mT as that in Figure a, assigned above to op* state. Another one has D = 341.5 mT assigned above to the cp* state (Figure b).

Remarkably, shortly upon the beginning of evacuation (in 5 min) the cp* state converts to op* state, proving that this is a phase transition (Figure b). Then the conversion from the op* to op state occurs on a much longer time scale, proving this process to be a desorption of diethyl ether from the PWs. Thus, we clearly see different kinetic behaviors for the phase transition (occurs promptly upon evacuation) and for desorption (occurs on a time scale of days).

Similar to the n-butane sorption experiments, we observe with Et2O that the spectrum (D value) of op* state is slightly different at low pressure (p = 20 mbar; p/p0 = 0.03; Figure a, D = 331.5 mT) and at high pressure (p = 300–640 mbar; p/p0 = 0.63–1, Figure , D = 337.5 mT). The relative magnitude of this difference is not high, ΔD/D < 2%, and is probably due to a minor perturbation of structure and electron density distribution by gas molecules at high pressure. In the future, the pressure dependence of the PW structure can be investigated in more detail experimentally and theoretically.

In summary, we demonstrated in situ EPR as a potent tool to investigate phase transitions and physi-/chemisorption in a highly porous MOF, DUT-49(Cu). The developed approach relies on the changes in zero-field splitting of intrinsic PW units upon structural transitions and/or adsorption/desorption; therefore, neither dopants nor additional EPR-reporter molecules are required. Further advantages of this approach are the high sensitivity and the ability to perform measurements in situ from low (typically, above 100 K) to room temperature. The latter is often challenging but at the same time very relevant for sorption of many molecules of interest into MOFs. In addition, kinetics of slow processes of adsorption/desorption could be easily monitored, which may or may not be coupled to phase transitions. This approach can be extended to study structural transitions in other flexible MOFs containing copper(II) PWs separated by sufficiently long linkers, as well as the chemisorption in both flexible and rigid MOFs of this type. Finally, X-band spectrometers are nowadays available in compact desktop versions, and further miniaturization of EPR toward microresonators and lab-on-a-chip devices is underway. Remarkably, the position of the X-band low-field ZFS line in DUT-49(Cu), which is a marker of structural transition, varies within ∼50–100 mT, which is a very low magnetic field easily accessible with small permanent magnets. Therefore, considering possible sensing applications of DUT-49(Cu) and other paramagnetic PW-based MOFs, EPR might be a method of choice in terms of simplicity of practical implementations. Overall, the present study opens new perspectives for applications of in situ EPR in basic and applied research of structurally flexible MOFs.