**1. Introduction**

Metal–organic frameworks (MOFs) are a class of advanced hybrid porous crystalline solids on which metal containing units and organic linkers are held together by strong chemical bonds [1–5]. They have gained increasing interest in recent years as an alternative to traditional nanoporous materials due to their unique combination of high porosities, large surface areas, different pore sizes, and variable topologies tailored through the use of different types of linker and metal nodes [1–5]. Among a large variety of possible applications [6–14], MOFs show great promise for separating and purifying solvent mixtures, processes of key importance in many applications [15–19]. It is expected that they could open an efficient alternative to existing separation procedures, which are expensive, time consuming, and very energy demanding [20,21]. As an example, distillation alone, the workhorse of chemical process industries, is responsible for 10–15% of the world's energy consumption and contributes significantly to the greenhouse gases emissions [20–22]. Therefore, the improvement of current adsorption-based separation methods through the implementation of more efficient adsorbent materials is of growing interest and of fundamental importance to a competitive and environmental friendly technology development.

The separation and purification efficiency of MOFs towards a certain solvent mixture is significantly altered by a variety of factors. They include the nature of the linker and the type of metal which can be chosen to specifically induce a preferential interaction with a certain solvent, the flexibility of the framework, a remarkable property of the MOF materials, as well as the pore size and shape which can be tuned towards size selective sieving [15–19,23–25]. A further factor, which can affect the separation efficiency of certain

**Citation:** Wagemann, M.; Radzik, N.; Krzyzak, A.; Adams, A. Screening ˙ Metal–Organic Frameworks for Separation of Binary Solvent Mixtures by Compact NMR Relaxometry. *Molecules* **2021**, *26*, 3481. https:// doi.org/10.3390/molecules26123481

Academic Editor: Robert Brinson

Received: 7 May 2021 Accepted: 5 June 2021 Published: 8 June 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

MOFs, is the solvent-induced breathing process [26–28]. Due to this effect, the porosity of the MOF increases upon solvent adsorption compared to that of the dry MOF such that the wet MOF enables the adsorption of larger molecules than the dry MOF. Up to now, tens of thousands of different MOFs have been reported and their number is strongly increasing each year [29]. Among them, there are probably several MOFs which show acceptable separation properties for a particular mixture and they are just waiting to be identified.

The efficiency of a particular MOF for the separation of solvents from binary mixtures is in many cases predicted by computational screening, currently as one of the most important tools for the fast screening of existing and hypothetical MOFs towards a particular application [30–32]. Another approach relies on predicting the behavior of a mixture in the presence of a MOF by taking into account the behavior of the individual solvents [33,34]. The outcomes of both approaches need however to be experimentally confirmed in the view of contradictory results reported for various solvent/MOF systems [34,35].

Transient breakthrough experiments and chromatography are the most used experimental analytical tools to investigate the performance of MOFs for mixtures separation [15,16,19]. However, they need to be combined with simulations or with other analytical tools, which are sensitive towards chemical structure to enable the identification and quantification of the individual components in the eluted mixture [15,16,19]. All these procedures are time consuming and require expensive and sophisticated equipment and trained operators. However, in order to reliably screen the huge number of existing MOFs materials, it is of key importance to identify reliable high-throughput analytical methods, which can overcome the restrictions of the existing approaches.

Among existing analytical technics, nuclear magnetic resonance (NMR) has strongly advanced our understanding about the MOF materials in terms of structure, dynamics of the linker, and motion and diffusion of adsorbed molecules by taking advantage of dedicated high-field NMR devices and sophisticated experimental techniques [36–43]. Yet, useful microscopic information can also be gained with the help of less expensive compact low-field NMR sensors with open and closed magnet geometries. Such sensors are well suited to interrogate the behavior of protonated materials and have been already successfully applied for a detailed characterization of various materials [36–40]. However, their application in the context of MOFs is still relatively scarce with existing studies focusing on the relaxation and diffusion behavior of gases and solvents in MOFs [41–43].

In this context, we introduce compact proton NMR relaxometry with an open NMR sensor as a new tool to test the potential of a certain MOF towards the separation of solvents within a binary mixture. Compact NMR sensors are based on permanent magnet technologies [36,37]. For open compact NMR sensors, with the Profile NMR-MOUSE being the most used, the permanent magnets are arranged in well-defined positions with respect to each other to generate a thin flat slice at a fixed distance above the surface of the sensor. The acquired NMR signal steams from this slice and thus the method enables a truly non-destructive analysis of samples of different sizes and shapes by simply placing them nearby the sensor. The sensors work in strongly inhomogeneous magnetic fields, which hamper the conduction of spectroscopy measurements but are well suited for relaxation and diffusion measurements [36–38]. The diffusion measurements are possible because the NMR sensors with open geometries have a strong intrinsic gradient, which is in many cases around 10 to 20 T m<sup>−</sup>1. One of the most used measuring methods with compact NMR is, however, the CPMG (Carr, Purcell, Meiboom, Gill) pulse sequence [44,45] which allows acquiring the whole transverse relaxation decay curve within a single shot. The analysis of the CPMG decay curves is done with the help of exponential fit functions or the use of inverse Laplace transform (ILT) that delivers the effective spin–spin relaxation times *T*2eff.

Our method quantifies the solvent mixture composition in terms of mass proportions before and after separation by a MOF using prior established correlation curves between the transverse relaxation times *T*2eff and the known mass proportions of a particular mixture. This new approach is tested by investigating the efficiency of UiO-66(Zr) in separating various binary mixtures and by comparison with the results obtained from 1H liquid-state

NMR on exactly same samples. The proposed methodology requires minimal amount of MOF, it is simple to conduct, and once the correlation curves are established for a certain binary solvent mixture, the separation efficiency of a MOF can be determined within few minutes of measuring time. Moreover, the relaxation measurements for various solvent mixtures can be implemented for conduction in an automatic way enabling thus a high-throughput screening.
