The targeted synthesis of new metal-organic frameworks (MOFs) is difficult, and their discovery depends on the knowledge and creativity of the chemist. High-throughput methods allow complex synthetic parameter fields to be explored quickly and efficiently, accelerating the process of finding crystalline compounds and identifying synthetic and structural trends.
High-throughput (HT) methods are an important tool for the fast and efficient screening of synthesis parameters and the discovery of new materials. This manuscript describes the synthesis of metal-organic frameworks (MOFs) from solution using an HT reactor system, resulting in the discovery of various phosphonate-based MOFs of the composition [Al2H12-x(PMP)3]Clx∙6H2O (H4PMP = N,N '-piperazine bis(methylenephosphonic acid)) for x = 4, 6, denoted as Al-CAU-60-xHCl, containing trivalent aluminum ions. This was accomplished under solvothermal reaction conditions by systematically screening the impact of the molar ratio of the linker to the metal and the pH of the reaction mixture on the product formation. The protocol for the HT investigation includes six steps: a) synthesis planning (DOE = design of experiment) within the HT methodology, b) dosing and working with in-house developed HT reactors, c) solvothermal synthesis, d) synthesis workup using in-house developed filtration blocks, e) characterization by HT powder X-ray diffraction, and f) evaluation of the data. The HT methodology was first used to study the influence of acidity on the product formation, leading to the discovery of Al-CAU-60∙xHCl (x = 4 or 6).
Metal-organic frameworks (MOFs) are porous, crystalline compounds whose structures consist of metal-containing nodes, like metal ions or metal-oxygen clusters, which are connected by organic molecules (linkers)1. By varying the metal-containing nodes as well as the linker, a variety of compounds can be obtained that exhibit a wide range of properties and therefore have potential applications in different fields1.
The stability of a material is important for its application1,2,3. Therefore, MOFs containing tri- or tetravalent metal ions, such as Al3+, Cr3+, Ti4+, or Zr4+, with carboxylate2 or phosphonate4 linker molecules have been the focus of many investigations5,6,7. In addition to the direct synthesis of stable MOFs, the enhancement of stability through post-synthetic modifications as well as the formation of composites is a field of interest2. Phosphonate-based MOFs have been less often reported compared to carboxylate-based MOFs8. One reason is the higher coordination flexibility of the CPO32- group compared to the -CO2– group, which often leads to the formation of dense structures and greater structural diversity8,9,10,11. In addition, phosphonic acids often must be synthesized, as they are rarely available on the market. While some metal phosphonates exhibit exceptional chemical stability10, systematic access to isoreticular metal phosphonate MOFs, which allows the tuning of properties, is still a topic of high relevance12,13. Different strategies for the synthesis of porous metal phosphonates have been investigated, such as incorporating defects into otherwise dense layers, for example, by partially replacing phosphonate with phosphate ligands4,14. However, as defective structures are poorly reproducible, and the pores are not uniform, other strategies have been developed. In recent years, the use of sterically demanding or orthogonalized phosphonic acids as linker molecules have emerged as a suitable strategy for the preparation of porous metal phosphonates4,8,10,11,13,15,16,17,18. However, a universal synthesis route for porous metal phosphonates has not yet been discovered. As a result, the synthesis of metal phosphonates is often a process of trial and error, requiring the investigation of many synthesis parameters.
The parameter space of a reaction system includes chemical and process parameters and can be vast19. It consists of parameters such as the type of starting material (metal salt), molar ratios of starting materials, additives for pH adjustment, modulators, type of solvent, solvent mixtures, volumes, reaction temperatures, times, etc.19,20. A moderate number of parameter variations can easily result in several hundred individual reactions, making a carefully considered synthesis plan and well-chosen parameter space necessary. For example, a simple study using six molar ratios of the linker to the metal (e.g., M:L = 1:1, 1:2, … to 1:6) and four different concentrations of an additive and keeping the other parameter constant, leads already to 6 x 4 = 24 experiments. Using four concentrations, five solvents, and three reaction temperatures would necessitate carrying out the 24 experiments 60 times, resulting in 1,440 individual reactions.
High-throughput (HT) methods are based on the concepts of miniaturization, parallelization, and automation, to varying degrees depending on the scientific question being addressed19,20. As such, they can be used to accelerate the investigation of multi-parameter systems and are an ideal tool for the discovery of new compounds, as well as synthesis optimization19,20. HT methods have been used successfully in different fields, ranging from drug discovery to materials science20. They have also been used for the investigation of porous materials such as zeolites and MOFs in solvothermal reactions, as recently summarized20. A typical HT workflow for solvothermal synthesis consists of six steps (Figure 1)19,20,21: a) selection of the parameter space of interest (i.e., the design of experiment [DOE]), which can be done manually or by using software; b) dosing of the reagents into the vessels; c) solvothermal synthesis; d) isolation and workup; e) characterization, which is typically done with powder X-ray diffraction (PXRD); and f) data evaluation, which is followed by step one again.
Parallelization and miniaturization are achieved in solvothermal reactions through the use of multiclaves, often based on the well-established 96-well plate format most commonly used in biochemistry and pharmacy19,20,22,23. Various reactor designs have been reported and several groups have constructed their own reactors19,20. Reactor choice depends on the chemical system of interest, especially the reaction temperature, (autogenous) pressure, and reactor stability19,20. For example, in a systematic study of zeolitic imidazolate frameworks (ZIFs), Banerjee et al.25 used the 96-well glass plate format to perform over 9600 reactions24. For reactions under solvothermal conditions, customized polytetrafluoroethylene (PTFE) blocks, or multiclaves with 24 or 48 individual PTFE inserts, have been described among others by the Stock group19,20. They are routinely employed, for example, in the synthesis of metal carboxylates and phosphonates. As such, Reinsch et al.25 reported the advantages of the methodology in the field of porous aluminum MOFs25. The in-house made HT reactor systems (Figure 2), which allow 24 or 48 reactions to be studied simultaneously, contain PTFE inserts with a total volume of 2.655 mL and 0.404 mL, respectively (Figure 2A,B). Usually, no more than 1 mL or 0.1 mL, respectively, is used. While these reactors are used in conventional ovens, microwave-assisted heating using SiC blocks and small glass vessels has also been reported26.
The automation of studies leads to time savings and improved reproducibility, as influence of the human factor is minimized20. The degree to which automation has been used varies strongly19,20. Fully automated commercial systems, including pipetting20 or weighting capabilities20, are known. A recent example is the use of a liquid-handling robot to study ZrMOFs, reported by the group of Rosseinsky27. Automated analysis can be performed by PXRD using a diffractometer equipped with an xy stage. In another example, a plate reader was used to screen solid-state catalysts, mainly MOFs, for HT screening of nerve agent degradation28. Samples can be characterized in a single run without the need for manual sample or position changes. Automation does not eliminate human error, but it reduces the possibility of its occurring19,20.
Ideally, all steps in a HT workflow should be adapted in terms of parallelization, miniaturization, and automation to eliminate possible bottlenecks and maximize efficiency. However, if it is not possible to establish a HT workflow in its entirety, it may be helpful to adopt selected steps/tools for one's own research. The use of multiclaves for 24 reactions is particularly useful here. The technical drawings of the in-house made equipment used in this study (as well as others) are published for the first time and can be found in Supplementary File 1, Supplementary File 2, Supplementary File 3, and Supplementary File 4.
Due to the complexity of the HT method, the individual steps and the method itself are discussed in the following sections. The first part covers the critical steps for each working step of the HT workflow (Figure 1), possible modifications, and limitations of the technique, where applicable. At the end, a general discussion also including the significance of the HT method with respect to existing methods and future applications is presented.
In the first step of…
The authors have nothing to disclose.
The work was supported by the Christian-Albrechts-University, the State of Schleswig-Holstein, and the Deutsche Forschungsgemeinschaft (especially STO-643/2, STO-643/5 and STO-643/10).
Norbert Stock would like to thank the B.Sc., M.Sc., and doctoral students, as well as the cooperation partners who have carried out many interesting projects using the high-throughput methodology, in particular Prof. Bein from the Ludwig-Maximilians-Universität in Munich, who played a major role in the development of the reactors.
AlCl3·6H2O | Grüssing | N/A | 99% |
Filter block for filtration of max. 48 reaction mixtures | In-house made | N/A | Technical drawings in the supplementary files |
Hydrochloric acid | Honeywell | 258148 | Conc. 37 %, p.a. |
Multiclaves with 24 individual Teflon inserts | In-house made | N/A | Technical drawings in the supplementary files |
N,N ‘-piperazine bis(methylenephosphonic acid | Prepared by coworkers | N/A | H4PMP, Prepared by coworkers with the method reported by Villemin et al.: D. Villemin, B. Moreau, A. Elbilali, M.-A. Didi, M.’h. Kaid, P.-A. Jaffrès, Phosphorus Sulfur Silicon Relat. Elem. 2010, 185, 2511. |
Sample Plate for PXRD | In-house made | N/A | Technical drawings in the supplementary files |
Sodium hydroxide | Grüssing | N/A | 99% |
Stoe Stadi P Combi | STOE | Stadi P Combi | Cu-Kα1 radiation (λ = 1.5406 Å); transmission geometry; MYTHEN2 1K detector; opening angle 18°; curved monochromator; xy-table |
Forced convection oven | Memmert | UFP400 |