*3.2. NMR Experiments*

The NMR experiments were performed at room temperature with a single-sided, portable NMR-MOUSE sensor having a static gradient field of about 20 T m−<sup>1</sup> and working at a proton resonance frequency of 18.2 MHz (Figure 6). A Bruker minispec spectrometer (Germany) was used for pulse generation and signal acquisition. Effective 1H spin–spin relaxation times *T*2eff of the pure solvents and their mixtures were determined at room temperature by employing a CPMG pulse sequence [44,45] with an echo time of 0.07 ms. The waiting time between two scans was set for all samples to 4 s in order to avoid heating effects during the measurements. In order to decrease the uncertainty, each measurement was performed three times. The error of the extracted *T*2eff values was for all samples less than 1% (see Table S2 for typical values, in the Supplementary Materials). For understanding the trends obtained for the *T*2eff values of the pure solvents, literature values, where available, are reported for the corresponding viscosities and self-diffusion coefficients [60–63].

**Figure 6.** The experimental set-up used to measure the relaxation times of the pure solvents and their mixture before and after the solvent separation by a MOF (here UiO-66(Zr)). The sample to be investigated is simply placed on the top of the profile NMR-MOUSE.

The analysis of all CPMG decays could be best done with the help of a single exponential function for the pure solvents and the mixtures (Figure S1). The distribution of the effective transverse relaxation times was obtained by performing an inverse Laplace transform (ILT) of CPMG relaxation decays.

Proton liquid-state high-field NMR measurements were performed at room temperature using a Bruker Ultrashield magnet operating at a proton frequency of 400 MHz and controlled by an AVANCE 3 console. The pure solvents and the various solvent mixtures were transferred to a 5 mm NMR tube hosting deuterated chloroform. The 1H spectra were acquired after a single 90◦ radio-frequency pulse with a recycle delay of ten seconds. Evaluation of the mixture ratios was done via integration of component specific peaks. The 1H spectra of all mixtures before and after the separation are depicted in Figures S6–S10 (in the Supplementary Materials) along with the signal assignment and the values of the integral of interest.
