Digital Control of Multistep Hydrothermal Synthesis by Using 3D Printed Reactionware for the Synthesis of Metal–Organic Frameworks

Abstract Hydrothermal‐synthesis‐based reactions are normally single step owing to the difficulty of manipulating reaction mixtures at high temperatures and pressures. Herein we demonstrate a simple, cheap, and modular approach to the design reactors consisting of partitioned chambers, to achieve multi‐step synthesis under hydrothermal conditions, in digitally defined reactionware produced by 3D printing. This approach increases the number of steps that can be performed sequentially and allows an increase in the options available for the control of hydrothermal reactions. The synthetic outcomes of the multi‐stage reactions can be explored by varying reaction compositions, number of reagents, reaction steps, and reaction times, and these can be tagged to the digital blueprint. To demonstrate the potential of this approach a series of polyoxometalate (POM)‐containing metal–organic frameworks (MOFs) unavailable by “one‐pot” methods were prepared as well as a set of new MOFs.

Abstract: Hydrothermal-synthesis-based reactions are normally single step owingt ot he difficulty of manipulating reaction mixtures at high temperatures and pressures.H erein we demonstrate asimple,cheap, and modular approach to the design reactors consisting of partitioned chambers,t oa chieve multi-step synthesis under hydrothermal conditions,indigitally defined reactionware produced by 3D printing.This approach increases the number of steps that can be performed sequentially and allows an increase in the options available for the control of hydrothermal reactions.T he synthetic outcomes of the multi-stage reactions can be explored by varying reaction compositions,number of reagents,reaction steps,and reaction times,a nd these can be tagged to the digital blueprint. To demonstrate the potential of this approach as eries of polyoxometalate (POM)-containing metal-organic frameworks (MOFs) unavailable by "one-pot" methods were prepared as well as aset of new MOFs.
Hydrothermal or solvothermal reactions are normally performed in ac losed system at temperatures above the boiling point of the solvent at standard pressure. [1,2] Hydrothermal conditions have been widely used to access many classes of materials including the synthesis of metal-organic frameworks (MOFs), [3] covalent organic frameworks (COFs), [4] extended metal oxides [5] and polyoxometalates (POMs). [6] Also,t oh elp speed up the discovery and optimization of new materials,h igh-throughput hydrothermal methods which allow the systematic investigation of the reaction parameters have been developed. [7,8] In addition, continuous flow hydrothermal techniques have been introduced for scalable synthesis. [9] Despite these advances,c onventional hydrothermal syntheses are still usually carried out under one-pot, single step conditions,w hich precludes the exploration of multi-step reactions or more complex reaction conditions.T his limits the ability to introduce synthetic complexity,whereby the composition of the reaction mixture can be changed under the reaction conditions without interruption. Some approaches to multi-stage hydrothermal reactions do exist, however these are limited to bespoke designed flow systems which are beyond the capabilities and means of many researchers. [10] We hypothesized that sequential hydrothermal syntheses could be achieved by creating reactors with internal geometries that allow the compartmentalization of different reaction mixtures,p reventing their mixing until defined points in the synthesis.T his approach leads to the possibility of "trapping" otherwise inaccessible reaction intermediates or unstable building blocks.W ith this in mind, we opted to build upon our recent work that used 3D printing to give architectural control of the reactor with the design and fabrication of reactionware. [11][12][13][14][15] This is because we wanted to see if the production of bespoke hydrothermal reactors,with partitioned chambers,w as possible.T raditional "one-pot" hydrothermal techniques are limited to using asingle reaction composition (c)w hich encompasses the sum of starting materials introduced at the start of the experiment, and this evolves through the course of the reaction. By introducing separate compartments to the reactor we introduce as imple way of separating the starting materials into two compositions (c 1 and c 2 )which can evolve separately,inparallel until mixed at new time, t n . As such the mixture can produce anew composition (c*), unavailable to traditional approaches,a nd these new mixtures can then continue reacting until the end of the experiment ( Figure 1A).
This architectural approach allows us to exert control over c*byvarying t n to explore the time-dependent kinetic studies on the "trapped" reaction intermediates.
In the first instance we applied this method for sequential hydrothermal synthesis,b yg eometrical control of the 3D printed reactor, to increase the synthetic parameter space and discover new materials unavailable by traditional techniques. To start with, we explored 8new composition parameters of as et of possible MOFs from an etwork of 144 (3 6 8) possible reaction combinations using a3Dprinted monolithic reactor ( Figure 2). Then as an initial proof of concept, one of the eight new composition parameters was used to investigate the development of multi-stage hydrothermal reactions with different POM clusters in a3 Dp rinted compartmentalized reactor ( Figure 1B), leading to the formation of polyoxometalate-MOFs (POMOFs) 1 to 3 ( Figure 3) which were unable to be produced by conventional "one-pot" apparatus. To expand the scope of this reaction design and synthesis method using geometrical control further, time-dependent studies of MOFs c 1 and c 2 on c*were explored and these gave rise to the discovery of the new MOF structures 9 to 11 ( Figure 4).
Prior to commencing the exploration of multi-step hydrothermal reaction in the 3D-printed reactors,the MOF components chosen for the multi-step synthesis were screened in 2 2a rray of approximately 3mL capacity reaction chambers via traditional one-pot reactions.Following our previous work, [13] the reactors were constructed in as imilar way by digital design in CADs oftware,f ollowed by fabrication using the FDM deposition of polypropylene using a3Dprinter. These new MOF candidates incorporate flexible cationic ligands (CLs) together with ditopic or tritopic ligands (L 1 -L 6 )into the frameworks ( Figure 2A). The introduction of CLs not only brings in flexibility because of their inherent diversity of conformations, but also allows the encapsulation of functional anionic clusters,such as POMs [16][17][18][19] within the frameworks of MOFs,w hich will be discussed further below,i n relation to the multi-step synthesis.
MOF structures 1 to 8 were prepared by optimizing the synthetic parameters using the 3D printed array reactors (Figure 2Band Supporting Information Figure S2). All these compounds are new and feature open frameworks that are constructed from ligands L 1 -L 6 and/or CLs with transition metal nodes,a nd the void is occupied by guest solvents and/or decomposed cations from CLs ( Figure S4-S11). Thef ormulation of MOFs 1 to 8 was determined on the basis of elemental analysis,I Rs pectroscopy,a nd thermogravimetric analysis (TGA). Thep hase purity of the bulk sample was established by comparison of its   observed and simulated powder X-ray diffraction (PXRD) patterns.T he synthetic procedure,c rystal data and crystal structures,a nd general characterization data of MOFs 1 to 8 can be found in Supporting Information including CCDC deposition numbers for the structural database.
Once ar ange of new MOFs compounds had been discovered in our 3D-printed array of single chamber reactors,p roof of concept experiments were carried out to explore multi-step hydrothermal reactions.T op erform the sequential hydrothermal synthesis,t he reactor was designed with two partially connected chambers that would allow the compartmentalization of reaction components at the first stage,a nd the mixing of the different reaction intermediates at various points during hydrothermal process (Figure 1Band Figure S1). Thef abrication of the bespoke reactor was conducted on an Airwolf HD2x platform using polypropylene (PP). Thestarting reagents and solvents were manually added during apre-programmed pause when the printing was 80 % complete.T he fabrication process was resumed and completed to give ah ermetically sealed reactor, which was subsequently removed from the platform and transferred into an oven for multi-stage hydrothermal reaction ( Figure 1C).
In the initial trial experiments,t he ingredients of MOF 1 (Co/CL 2 /L 2 )were selected as compositional control parameters in c 1 to explore multi-step reactions with the Keggintype POM {SiW 12 }asthe reagent in c 2 .The solutions of c 1 and c 2 remained in their respective chambers under hydrothermal conditions for 12 hours (t n ), and then was mixed after turning the reactor upside down to produce an ew composition c*, which was left undisturbed to continue reacting for another 36 hours (t m )u nder the same conditions.A fter cooling to room temperature,t he reactor was opened using ad rill and the reaction mixture was carefully extracted using ap ipette. Single-crystals of POMOF 1 were then obtained and analysed by X-ray diffraction, infrared spectroscopy,t hermogravimetric analysis and elemental analysis (see Supplementary  Information). In the crystal structure of POMOF 1,t wo cationic ligands CL 2 coordinate to two Co ions to generate ac ationic metallocycle ( Figure 3A). One {SiW 12 }c luster is sandwiched by two metallocycles via electrostatic interactions and multiple hydrogen bonds formed between both the terminal and bridging Oatoms on {SiW 12 }and Hatoms on the benzene ring of CL 2 .I na ddition, strong anion-p interactions are also seen between the terminal O atoms of {SiW 12 }and aromatic ring of the linker with the anion-to-ring centroid contact of 2.943 and 3.006 ,r espectively ( Figure S12). Va rying the composition parameter c 2 from Keggin anion to Anderson-type {CrMo 6 }g ave rise to POMOF 2,f eaturing a2 Dg rid-like framework built from the connection of central Co 2+ ions with terminal Oa toms of {CrMo 6 }a nd CL 2 ligands ( Figure 3B). In contrast, owing to the strong electrostatic interactions between cationic ligands and POMs,c ontrol experiments showed that amorphous precipitates formed immediately once all the starting materials of c 1 and c 2 were added into at raditional "one-pot" reactor.S uch precipitates remained unchanged even after hydrothermal reactions ( Figure S3). Therefore,t hese results unambiguously demonstrate the applicability of the two-chamber rector for multi-stage hydrothermal reaction and the advantage of this design towards preparation of new MOF structures that are inaccessible by conventional "onepot" approach.
Instead of using preformed POM clusters as acomposition parameter c 2 ,t he compartmentalized reactor also allows the variation of c 2 by in situ synthesis of POM clusters during the multi-stage hydrothermal process.T his,i np rinciple,w ill provide more potential inputs that could further extend the synthetic parameter space.T od ot his,N a 2 MoO 4 and Cr-(NO 3 ) 3 were placed in one chamber to prepare {CrMo 6 }insitu before the turning point t n ,a nd the resulting sequential synthesis led to anew compound POMOF 3;t his compound has similar 2D grid framework to POMOF 2,a nd it is constructed by connection of Na + ions with {CrMo 6 }and CL 2 ligands,w hile free Cr 3+ ions are located within the 2D open channels and exhibited octahedral geometries fulfilled by six Oa toms from two DMF,t wo water, and two methanol molecules ( Figure 3C).
To further demonstrate the potential of the geometrical approach for discovery of new materials via the introduction of more synthetic parameters,time-dependent and kinetically controlled hydrothermal reactions were performed by mixing the two compositions (c 1 and c 2 )atdifferent time points t n to produce as eries of new compositions (c*), which can potentially result in different products and thus new materials. To this end, the reagent for MOFs 5 (Co/CL 1 /L 2 )a nd 6 (Cd/ CL 1 /L 6 )w ere added into the two compartmentalized chambers to conduct multi-stage hydrothermal reactions.Asshown in Figure 4A,various crystal combinations could be obtained by varying the mixing time points t n . When the two chambers were mixed at t n = 0, only one type of crystals (MOF 9) formed. Single-crystal X-ray analysis revealed that 9 is composed of a3 Dn etwork constructed by connection of Cd and Co centres with L 2 ligands ( Figure 4B). Bridged by L 2 , at rinuclear Cd1-Co3-Cd1 secondary building unit (SBU)i s generated and further linked with adjacent trinuclear SBU and Cd2 nodes to afford ah ighly-connected 3D framework. Thec ationic ligands CL 1 are thermally unstable and decompose into methyl viologen (MV) fragments, [20] which serve as guest molecules to fill the cavities of MOF 9.B yc arefully tuning the mixing point t n from 0t o3hours, an ew type of compound with hexagonal crystals began to form in addition to MOF 9 and the yield gradually increased when t n was varied from 3t o6hours ( Figure 4A). Such crystals,d enoted as 10,were also found to possess a3Dnetwork constructed by connection of Cd and Co centres with L 2 ligands ( Figure 4C).
Tw ok inds of Cd centres and one type of Co centre are found in compound 10.C d1 and Cd2 are connected by L 2 to build adinuclear SBU extending into a2Dframework along a axis,w ith the peripheral L 2 further bound by Co3 to construct a3 Do pen framework. Similar to 9,t he open channel is filled by solvent and MV molecules.F urther increasing t n to 9hours afforded rod-like crystals together with 9 and 10.X -ray analysis revealed that such rod crystals, named herein as MOF 11,are only comprise eight-coordinate Cd1 centres and L 2 .T he basic SBU is based on ad inuclear motif built from two symmetry-related Cd1 centres (Figure 4D). Each SBUi ss urrounded by six L 2 and each L 2 connects with three SBUs, thus giving rise to a3 Do pen framework. AMVmolecule is found to be located within the channels of compound 11.O nce the two compartmentalized reaction systems were mixed after 24 ho fr eaction, mixed crystals of 5 and 6 were obtained, indicating these MOF crystals have already formed after 24 hinseparate chambers, and dominated as the stable species that cannot be accessible to either ligand or metal exchange.
These experiments clearly demonstrate that by mixing c 1 and c 2 at variable points t n ,itispossible to exert control over the c*t oc reate aw hole new complex system that allows access to different reaction intermediates,l eading to exploitation of new materials more easily and efficiently.I ts hould be mentioned that MOFs 9 to 11 can be reproduced in large scale using traditional "one-pot" method once the synthetic recipes are established by 3D-printed multi-stage reactions. However,t oe xtend the time-dependent synthetic strategy, similar procedures were also adopted for the discovery of new MOFs by replacing the initial parameters c 1 and c 2 with 4 (Co/ CL 1 /L 4 )a nd 6 (Cd/CL 1 /L 6 ), respectively.T his gives different results compared to the approach using 5 and 6,i nw hich MOF 12 is obtained as the sole product irrespective of mixing points t n . MOF 12 features a2 Dl ayer structure constructed from Cd 2+ ions and L 4 ,which is pillared by L 6 to form box-like 1D channel filled by MV molecules ( Figure S13). Ac ontrol experiment was done using "one-pot" reactors to produce am ixture of MOF 12 and MOF 13.X -ray analysis showed that 13 is assembled from Cd ions and L 6 ,a nd displays 1D zigzag chain structure.I nc ontrast to 12,t he guest MV molecules are wrapped within the pitches between adjacent chains ( Figure S14). These results indicate that the timedependent strategy has the potential to be used for manipulating the production distribution, and thus enable the selective formation of targeted product from ac omplex mixture.
In summary,w eh ave demonstrated an ew conceptual approach for the hydrothermal synthesis of compounds, controlled by geometrical design, and implemented using 3D printed reactionware for multi-stage hydrothermal reactions.T his 3D method enables us to divide the reaction control into more compositional parameters (c 1 , c 2 …c n ), and also introduces anew time parameter t n for mixing c 1 and c 2 to produce c*. Thus,byincreasing the number of parameters,we have shown it is possible to open up new areas of the synthetic space,w hich should allow researchers to search for new materials,i mpossible to make using conventional methods. Finally,the digital design of the reactors,coupled with tagging the synthetic meta-data will lead to improved collaboration, and reproducibility of results.T his will be made possible via development of ad igital-chemical code for hydrothermal synthesis we will develop in the future based upon the reactionware approach. [11]