Supramolecular chemists seek to develop materials and systems made of multiple molecular building blocks where the whole is “more than the sum of its parts.” Many have exploited the reversibility of both non-covalent and dynamic covalent systems to form multicomponent structures under thermodynamic equilibrium (Whitesides et al., 1991; Rowan et al., 2002; Otto and Severin, 2007; Han et al., 2010; Okesola and Mata, 2018). Under fully reversible equilibrium conditions, the final outcome of such processes only depends on the molecular building blocks being used and the overall stability of the products: the thermodynamic product is often observed, representing the global energy minimum (Rowan et al., 2002). However, “out-of-equilibrium” structures have recently gained more attention (Ogi et al., 2014). For these systems, the reaction environment has a strong influence on the assembly pathway and the final structures obtained. Under non-equilibrium conditions, or where diffusion is slower than the rate of reaction, the impact of mixing efficiency or local concentration gradients can be substantial (Sevim et al., 2018). Controlling mixing is very difficult to achieve in a standard flask, but can be readily achieved under flow and microfluidic conditions (Nagy et al., 2012; Ward and Fan, 2015).

Microfluidic reactors–where reactions are carried out in channels <1 mm–offer unique control over the mixing of reagents in both space and time (“spatio-temporal control”) (Sevim et al., 2018). The degree of mixing can be much more finely controlled compared to batch, ranging from extremely turbulent flow and thus fast mixing to extremely slow: in the laminar flow regime, streams of fluids flow parallel to each other, creating a defined liquid-liquid interface between them (Brivio et al., 2006). Thus, specific assembly pathways can be deliberately targeted by controlling the flow regime, and obtaining non-equilibrium structures becomes easier in microfluidic environments, facilitating the targeted synthesis of materials with desired properties. Furthermore, in flow, a higher degree of control results in greater reproducibility and a consequent avoidance of off-target reactions due to poor mixing and local concentration gradients.

One recent example where this has been successfully exploited in supramolecular chemistry is the work of Numata et al. (2015b) who were able to synthesise porphyrin microfilms stabilized by an extended two-dimensional hydrogen bonding network. In contrast to batch conditions which formed an amorphous material, the use of microflow conditions enabled the formation of regular hydrogen-bonded networks leading to micron-sized multi-layered porphyrin sheets. This suggested that the microsheets could only be formed following the non-equilibrium kinetic pathway established under microflow conditions. Microfluidic conditions have also been used to access alternative self-assembled structures in a controlled manner in the field of organic conductors (Puigmartí-Luis et al., 2010), co-ordination polymers (Rubio-Martinez et al., 2016b), covalent organic frameworks (COFs) (Rodríguez-San-Miguel et al., 2016; Singh et al., 2018), and MOFs (Ameloot et al., 2011), and to control the hierarchical supramolecular assembly of perylene bisimides (Numata et al., 2015a), nanofibers (Numata and Kozawa, 2013), and amphiphiles (Numata et al., 2015c), with the latter enabling the formation of energetically unfavourable self-assembled structures under kinetic control.

Whilst control over reaction outcomes and the assembly process is a powerful tool, it is made more powerful if it can be rapidly applied to a wide range of systems. One potential drawback of continuous flow is the challenge of parallelising experiments, which currently limits throughput. HTS allows a much faster, and arguably more efficient, exploration of a wider chemical space, and can lead to the accelerated discovery of new structures. Efforts to parallelise or improve the throughput of flow experiments include microdroplet screening (Reizman et al., 2016), continuous variation of variables coupled within inline analysis (Aroh and Jensen, 2018), and multi-channel chip architectures (Headen et al., 2018), and it is likely that these approaches will become increasingly used to combine the benefits of enhanced control with rapid screening.

Synthetic screening often results in a vast number of structures with only a select few being suitable for the application of interest. Fast and effective screening of the design space thus plays an important role in identifying materials and assemblies with desirable properties in supramolecular chemistry, and this is where HTS can offer its services.

A typical approach to HTS involves the use of multi-well microtiter plates and is often aided by the use of automated robotic platforms. For example, Greenaway et al. (2018) illustrated the advantages of employing an automated platform for HTS alongside computational analysis in the search for new organic cages. Overall, this led to the streamlined discovery of 33 new organic cages, with two forming bridged-catenane structures upon recrystallization, a new cage topology, highlighting the advantage of HTS in accelerating serendipitous discoveries. Their automated workflow has also been employed in the discovery of other supramolecular assemblies including an unsymmetrical organic cage (Berardo et al., 2018) and socially self-sorted organic pots (Greenaway et al., 2019). More recently, Cui et al. (2019) demonstrated how automated platforms can be used for polymorph screening, resulting in different HOFs being accessed. Additionally, Lin et al. (2021) demonstrated how HTS can screen for supramolecular diversification, resulting in a range of different hierarchical self-assemblies such as thin fibrils, helical ribbons, twisted ribbons, wide and thin ribbons, and macroscopic hydrogels.

Despite the potential advantages HTS offers, it has still not been widely adopted in the field of supramolecular chemistry. This could be due to factors such as the typically high initial cost of automated platforms, or challenges in finding suitable and equally high-throughput characterisation techniques for the supramolecular assemblies being targeted. However, with the availability of lower cost and open-source automated liquid handling platforms, and with several advances having been made in the HT characterisation of supramolecular materials (Greenaway et al., 2018; White et al., 2020), we expect these techniques to become more widely adopted in the field for screening and discovery of new “hits.”

Once a “hit” has been found, the next step in the process is to optimize for parameters such as yield, selectivity, performance, throughput, or stability; this process can take as long as the discovery phase. Here, enabling technologies offer the opportunity to shorten this timeline and assist the chemist in making key decisions to ensure the best chance of finding an optimal supramolecular system.

Flow chemistry enables real-time analysis, giving mechanistic insight that can be used to inform optimization. Intermediates can be isolated from the reaction stream and separately analysed to provide a better understanding of the reaction pathway and general self-assembly processes (Sagmeister et al., 2021). For example, recently Jones et al. (2021) exploited this to optimise the synthesis of a macrocyclic molecular hinge using a semi-continuous method informed by at-line analysis. Here, the fast heat transport possible in flow enabled a rapid temperature change mid-synthesis, leading to a threefold increase in yield and simultaneous reduction in reaction times.

HTS can also be used to screen many different parameters in parallel (e.g., reagent stoichiometries, solvents, catalysts, additives, and temperatures), and has been widely used in the pharmaceutical industry to optimise yield and selectivity (Shevlin, 2017; Mennen et al., 2019). This optimisation process can also be aided by incorporating Design of Experiments (DoE) to narrow down the number of reactions required. In addition, microfluidic platforms can also be employed for HTS which, when compared to automated platforms and microplates, can significantly reduce the quantities of reagents required for both screening and optimisations (Zhou et al., 2020). Whilst there are very few examples of applying flow, microfluidics, or automated platforms for high-throughput screening in supramolecular chemistry, there is no reason why these approaches could not be adopted for the field.

It is worth noting that whilst HTS can enable a large amount of the design space to be investigated, it is not desirable to fully empirically explore all precursor combinations or reaction conditions to ensure the optimal conditions are selected. By combining HTS and flow chemistry with machine learning algorithms and Bayesian optimisations, and in situ analysis techniques, this process can be streamlined by using closed-loop autonomous screening to optimise towards a target parameter or property (Zhou et al., 2017; Bédard et al., 2018; Clayton et al., 2019; Clayton et al., 2020). A recent example that highlights this in the area of supramolecular chemistry is the work of Cronin and co-workers who used a closed-loop autonomous “chemical robot” to explore a large combinatorial space for the discovery of coordination architectures (Porwol et al., 2020). The group calculated that it would take 4 × 10⁴ experiments to fully explore the variables chosen, and therefore proposed an autonomous decision-making approach to explore this space as rapidly as possible, discovering four new coordination complexes in the process. This example highlights the importance of decision-making in screening large parameter spaces: the task of deciding “what to explore,” or “what to explore next,” normally taken by the chemist, is a critical point that determines the success or failure of the experiment as well as how fast that point is reached. Emerging research in algorithms for process space exploration, growing from the established science of DoE approaches, are yielding incredible advances in discovery speed in adjacent fields and are likely to make a similar impact in supramolecular chemistry, especially if suitable conditions for scale-up can also be identified.

Despite the activity in the field, currently, there are only a few supramolecular materials that are produced for commercial purposes. The many issues surrounding the scale-up process (e.g., large amounts of organic solvents, long reaction times, poor reproducibility) have hindered this transition, preventing structures from being viably used for the many promising applications they present (Rubio-Martinez et al., 2017).

Continuous flow reactors are one of the latest technologies being employed to produce supramolecular structures at scale. Their large surface area-to-volume ratio allows more efficient mass and heat transfer, giving safer and faster reactions. Improvements on the reproducibility and product quality can be obtained due to the higher level of control over the reaction. Flow reactors also generally require less solvent compared to traditional batch reactions, driving down the cost and waste which are all factors that need to be considered for an industrial scale process (Rubio-Martinez et al., 2017).

Recently, flow chemistry has been used to optimize the yield, selectivity, and limited scale-up, of both macrocyclic (Bogdan and James, 2010; Bogdan and James, 2011; Bedard et al., 2013; Bédard et al., 2015; Lucke et al., 2016; Morin et al., 2019; Jones et al., 2021; Seemann et al., 2021) and cage (Briggs et al., 2015; Kitchin et al., 2015) supramolecular structures. For example, Bédard et al. synthesized medicinally relevant (Marsault and Peterson, 2011) macrocyclic lipids under continuous flow, with yields of up to 97% in short reaction times (Bedard et al., 2013). They were also able to increase the reaction to a multigram scale for some macrocycles; high yields were retained without the need for re-optimisation. Similarly, Briggs et al. (2015) translated the synthesis of porous organic cages (POCs) from batch to flow, resulting in greatly reduced reaction times and reduced solvent use, offering a continuous method of scale-up. It should be acknowledged, however, that there is still significant method development required to achieve the multi-kilo or tonne scale needed for many industrial applications, and here collaboration with process chemists and engineers is essential.

Flow has also assisted in the scale-up of MOFs (Munn et al., 2015), which have promising applications in gas storage, gas separation and heterogeneous catalysis, but are typically produced by solvothermal batch synthesis that is challenging on a large scale (Czaja et al., 2009). For example, Rubio-Martinez et al. (2016a) were able to scale-up the synthesis of the aluminium fumate (Al-Fum) MOF from a laboratory to a pilot-plant system by more than 2 orders of magnitude giving an increased space-time-yield (STY) of 97,159 kg m⁻³ day⁻¹. Since then, significant progress has been made towards the commercial synthesis of MOFs; for example, a continuous hydrothermal synthesis is now being used by Promethean Particles, who commercially produce nine different MOFs and porous structures (Munn et al., 2015).

It is also worth noting that scale-up of supramolecular materials has also been demonstrated under mechanochemical continuous flow, namely using twin-screw extrusion (TSE), which can reduce the solvent requirement for synthesis even further. For example, both MOFs and POCs have been formed on scale with very little to no solvent using TSE (Crawford et al., 2015; Egleston et al., 2020; Casaban et al., 2021). Finally, there have also been promising developments in combining flow reactors and microwave heating for the process intensification of MOF formation (Laybourn et al., 2019). It is clear that the materials chemist has a number of tools at their disposal to ensure compounds can be made at-scale, and we anticipate this will become more widespread in supramolecular chemistry.