Assessing Crystallisation Kinetics of Zr Metal–Organic Frameworks through Turbidity Measurements to Inform Rapid Microwave‐Assisted Synthesis

Abstract Controlling the crystallisation of metal‐organic frameworks (MOFs), network solids of metal ions or clusters connected by organic ligands, is often hindered by the significant number of synthetic variables inherent to their synthesis. Coordination modulation, the addition of monotopic competing ligands to solvothermal syntheses, can allow tuning of physical properties (particle size, porosity, surface chemistry), enhance crystallinity, and select desired phases, by modifying the kinetics of self‐assembly, but its mechanism(s) are poorly understood. Herein, turbidity measurements were used to assess the effects of modulation on the solvothermal synthesis of the prototypical Zr terephthalate MOF UiO‐66 and the knowledge gained was applied to its rapid microwave synthesis. The studied experimental parameters—temperature, reagent concentration, reagent aging, metal precursor, water content, and modulator addition—all influence the time taken for onset of nucleation, and subsequently allow microwave synthesis of UiO‐66 in as little as one minute. The simple, low cost turbidity measurements align closely with previously reported in situ synchrotron X‐ray diffraction studies, proving their simplicity and utility for probing the nucleation of complex materials while offering significant insights to the synthetic chemist.


Introduction
Despite the growth in interest and vast number of metal-organic frameworks (MOFs)-coordination polymers wherein organic ligands connect metal ions or clusters into network solids-synthesised over recent years, there has been relatively little investigation into the crystallisation process of these materials.A saresult,u nderstanding of the reactiona nd crystallisation mechanisms remainsl imited, with MOF synthesis often proceeding more through trial and error than by design. [1,2] Understanding how these materials assemble would allow arational approacht owards MOF design and synthesis, increasing the ability to generate highlys pecialised materials of desired size, topologya nd functionality. [3][4][5] Crystallisationi sk nown to heavily dependo nv arious different reactionp arameters, such as temperature, pH and time, [6,7] whilst addition of crystallisation promoters has been shownt o increasec rystallinity and allow particle size control. [8][9][10][11] An alternative is the use of coordination modulation-thea ddition of monotopic ligands, known as modulators, into synthetic mixtures to compete with the linkersf or metal coordination, allowing fine controlo ver an umber of physicalp roperties such as size, morphology,d efectivity and porosity in MOFs linked by high valentm etals-at echnique now ubiquitous in the self-assembly of Zr MOFs. [12][13][14][15] Knowledge on precisely how,w hy and to what extent these parameters affect crystallisation is limited however. As summarised in ar ecent review by VanV leet et al.,s everali ns itu ande xs itu techniques have been used to probe the nucleationa nd growth of MOFs. [2] An in situ total X-ray scattering study carried out by Xu et al. examined the formation of the hexanuclear Zr 6 secondaryb uilding units (SBU) duringt he solvothermals ynthesis of the archetypal zirconium terephthalate MOF UiO-66 [16] (Figure 1a), [Zr 6 O 4 (OH) 4 (BDC) 6 ]( BDC = 1,4-benzenedicarboxylate), suggesting self-assembly occursb yt he formation of SBUs in am etal salt precursor solution,f ollowed by assembly of multinuclear clusters upon the addition of the organic linkers, and finally coalescence of these clusters to form an orderedc rystalline framework. [17] An in situ synchrotron energy dispersiveX -ray diffraction (EDXRD) study into the formation of the isoreticular Zr-fum analogue, [Zr 6 O 4 (OH) 4 (fumarate) 6 ], in water showed that the addition of increasing equivalents of formic acid modulator to the aqueous synthesis resultsi nadecrease in both the nucleationr ate and crystal growth rate, supportingt he concept of linker/modulator competition decelerating the reaction rate (typical modulators are shown in Figure 1b). [18] More recently,i ns itu EDXRD has been used to examinet he impact of the organic linker and modulators on the crystallisation kinetics of UiO-66(Zr/Hf)-type MOFs, with results indicating the crystallisation process to be highly complex, as linker length and solubility,a nd modulator pK a ,a ll influence crystallisation. [19] Water has been shown to speed up the nucleation of UiO-66, presumablyb yp roviding as ource of the bridging O 2À and OH À ligands found in the Zr 6 SBU, while concentrated HCl (37 % w/w aqueous) slows nucleation compared to pure water, presumably by disfavouring linker deprotonation, but nucleation is still faster than unmodulated syntheses due to the incipient water. [20] Addition of aqueous HCl to solvothermal syntheses has also been found to improve crystallinityo ft he resulting Zr MOFs. [21] The majority of the current studies into crystallisation processes of MOFs are carriedo ut under solvothermal conditions. There are av ariety of synthesis methods for MOFs, however, with microwavea ssisted heating becoming ac ommonly used alternative given its ability to vastly reduce synthesis times (and consequently lower energy consumption), allow for control of particle properties, such as size andm orphology, and produce otherwise unobtainable materials. [22][23][24][25] The use of microwaveh eating does, however,i ntroduce an ew area of uncertainty in the crystallisation of MOFs. As rapid microwave heatingo ffers ad ifferent route for energy to be introduced into the system-the interaction of electromagnetic waves with polar solventm olecules, metal salts, and organic linkers in solution resulting in the production of heat, which can then also be transferred conductively-the kinetics of nucleation and crystallisation are also likely to vary. [26,27] In many instances, the use of microwaveheating leads to the reduction of particle size, resulting in nanoparticles with an arrow size distribution, [28][29][30] suggestedt ob eaconsequence of more homogeneous heating of the reaction solution leading to faster nucleation kinetics, [31,32] and effectively inducing LaMer burst nucleation. [33,34] As rapid nucleation of crystalline MOF particles occurs, reactant is quickly removedf rom the synthesis, limiting the reactanta vailable for the growth of particles.H owever, there are severale xamples of microwave heatingr esulting in particleso ft he same size to those produced solvothermally. [22] For example, when UiO-67,t he isoreticular analogue of UiO-66 with biphenyl-4,4'-dicarboxylate linkers, is produced through an l-proline modulated synthesis, the particle size shows no change withv arying heatsource, suggestingm odulators retain ac onsiderable effect on the nucleation and crystal growth process when undergoing microwave heating. [12,35] To determine crystallisation behaviour under solvothermal conditions, turbidity measurements offer as imple alternative to in situ diffraction techniques, [36][37][38] which often requireb espoke equipment and synchrotron access. The homogeneous startings olution allows the user to pinpoint crystallisation as light transmissivity decreases upon the crystallisation of solid material. This methodology has been effectively used to monitor the nucleation of covalent organic framework (COF) synthesis, [39] and very recently,a longside spectroscopict echniques, the mechanism of formation of MIL-53(Al). [40] Herein, we probe the crystallisation process of UiO-66 by turbidity measurements in ap arallel crystallisation system. Multiple reaction parameters are probed, from temperature and concentration to the addition of modulators, andt heir effects on the kinetics of crystallisation collated. Following these results, the insight gained from turbidity measurements is used to analyse the rapid microwaves ynthesis of UiO-66, once more varying reactionp arameters to better understand the crystallisation process.

Crystallisation kinetics
Experimental runs proceeded through the reported UiO-66s olvothermal synthesis, [16] consistingo fa1:1r atio of ZrCl 4 and BDC in N,N-dimethylformamide (DMF, unpurified ACS grade unless otherwise stated) at ar ange of concentrations, divided Figure 1. a) Schematic of variables to consideri nt he self-assemblyofU iO-66: (i) the thermal decomposition of DMF to release base and formic acid, a potentialcoordination modulator, with associated proton balance (temperature); (ii) the deprotonation of the 1,4-benzenedicarboxylate (BDC)linker (pH); (iii) the formation of the hexanuclear[Zr 6 O 4 (OH) 4 (RCO 2 ) 12 ]s econdary building unit (Zr source, water content); (iv) the level of incorporationo ft he BDC linker in these SBUs and competition from additional ligands (modulators); (v) final substitution of transientl igands/modulators to allow coalescence into the UiO-66s tructure, redrawn from CCDC deposition RUBTAK, [16] and its subsequent nucleation (reagentc oncentration). b) Typicalc oordination modulators used in the self-assembly of Zr MOFs. into sixteens ealed 1mLv ials. To mimic typical laboratorys yntheses of UiO-66, standard laboratory grade reagents and solvents were used withoutp urification.T hese vials are then simultaneously heated within the Crystal16 parallel crystalliser and held at the desired reactiont emperature, with turbidity measurements monitoringt he crystallisation process. Each run yields 16 independent measurements of the induction time, defined as the time until detection of crystals under constant process conditions like temperature, from which the induction time probability distribution can be determined, producing a plot of induction time probability versus time. The generated solids were collected, combined, and examined by powder-Xray diffraction to confirmt he formation of UiO-66 (see Supporting Information, Section S2). In addition, previous studies investigating synthesis and modulation of UiO-66h ave shown no transients pecies crystallising prior to UiO-66. [7,19,20] It should be noted that some run-to-run variability in absolute induction time was observed;c areful attempts were made to limit experimentalv ariability within as ingle run as much as possible (inter alia)-thed ata represented in each individual figure were collected at the same time using the same reagentsand the trends observed are consistent, allowing valid comparisons to be made. We are therefore confident that the stochastic natureo fn ucleation [37] is the primary reason for variations in specific induction time distributions.
The first parameter studied was that of reaction temperature. MOF syntheses in formamide solvents are believed to be driven by the thermal decomposition of the reactions olvent. As the solvent( in this case DMF) breaks down, [41] ab ase (in this case dimethylamine) is released, which is of sufficient basicity to deprotonate the organic linker,a llowing for assembly with the inorganic clusteru ltimately forming the framework. As ar esult, increasing the reactiont emperature should lead to af aster nucleation rate, as has been observed for DMF-based syntheses of Zr-fum by EDXRD. [20] Solutions of varying concentration were thus run at 373 K, 383 Ka nd 393 Kt op robe the effect of temperature and to validate the turbidity measurement protocol, and the samples are named Zr-BDC (concentration/temperature) to denote the conditions probed. The measurements produce induction time distributions exemplified by those in Figure 2, where it is possible to assess the onset time-the appearance of the first data point-and the variation in the overall distribution of the points. We hypothesise that al arger time of onset is caused by as lower rate of reaction and al arger variation is duet oas lower nucleation rate.
As expected, the onset of the induction time distribution occurs more rapidly as the reactiont emperature is increased, with materials synthesised at 373 Ks howing an average induction time of about 25 000 sc ompared to approximately 2300 s for the 393 Ks yntheses, as ac onsequenceo ft he more rapid decomposition of DMF at higher temperatures. Reduction in temperature also appearst oi ntroduce greater variance in inductiont imes, with materials ynthesised at 373K crystallising over al arger time range, suggestive of slower nucleation, compared to at 393 K.
Whilst maintaining a1 :1 ratio of Zr 4 + to ligand,s yntheses with concentrationsr anging from 11.25 mm to 90 mm were carriedo ut across the temperature range. Ag eneral trend was observed, whereby higher concentration solutions typically displayedl ongerinduction times. Although this result appearsinitially counterintuitive, this could be due to the higher concentrationo fZ rCl 4 in the synthesis leadingt oagreaterr elease of HCl, potentially hindering the deprotonation of the carboxylate linker and leading to slower assembly.I nterestingly,t he increase in concentration did not reduce significantly the variance in the induction time distribution, indicating that nucleation rates were similar.
As synthesis temperature was shown to greatly affect the rate of crystal nucleation,f urtherr eactionp arameters were then examined at ac onstant temperature of 393 K. For subsequent experiments, the higherc oncentration syntheses (90 mm)w ere replaced with an intermediate concentration (33.75 mm)d ue to solubility issues when certain additives were used, as homogeneous starting solutionsare essential foraccurate nucleation determination. All experimentsw ere carried out at 393 K, and so temperature is not denoted in sample names forthwith. The addition of water is routinelyu sed in the synthesis of zirconium based MOFs, with the nucleationo ft etragonal ZrO 2 nanoparticles implied as ap recursor to forming the hexanuclear secondary buildingu nit. [42] Al ack of water can also lead to the formation of an alternative Zr terephthalate MOF,M IL-140A, [43] at highert emperatures (MIL-140A is said to be the thermodynamic product, while UiO-66i st he kinetic one, in this system). Certainly,asource of oxygen for the bridging oxo and hydroxo units in the clusters of both Zr terephthalates is required, and often not controlled for: sourceso fa dventitious water may include partially hydrolysed or hydrated metal salts, wet solvents, or absorption of atmospheric moisture.
The use of dry DMF (water content < 0.005 %, Zr-BDC-dry, blue symbols in Figure 3) led to overall longer induction times compared with reagent grade DMF (waterc ontent < 0.2 %, Zr-BDC,r ed symbolsi nF igure 3) over the reaction concentration range studied, and higherc oncentration solutions again generally nucleatedm ore slowly.W hilst it is difficult to ensure completely dry conditions, these results show ar educed presence of water hinderst he nucleation and formation of UiO-66. Syn- theses using dry DMF deliberately spiked with 5equiv of water (Zr-BDC-5H 2 O,b lack symbolsi nF igure 3) show av ast decrease in the induction time, which occurs after only 1300s for Zr-BDC-5H 2 O( 45 mM),a nd with all 16 vials routinelyn ucleating within av ery short distribution time frame compared with the Zr-BDC-dry experiments. In contrastt oo ur previouso bservations, the induction onset time now shows ag eneral increase with decreasing reagent concentration, complying with conventional nucleation theories that predict highern ucleation rates at increased reagent concentrations. This couldi ndicate that the rate of formation of the metal clusters, aided by the addition of water,o verrides the effect of the increased release of HCl by the starting materials (and indeedc oncomitant consumption of water)a th igherc oncentrations, which limits deprotonation of the linker.
The effect of HCl on crystallisation was examined (Figure 3e) via the addition of 1equiv of 37 %H Cl to the reaction solutions using the same dry DMFs olvent (Zr-BDC-1HCl,g reen symbolsi nF igure 3e). Addition of HCl results in quicker onset of nucleationt han the control, Zr-BDC,w ith an induction time of approximately 3600 sf or Zr-BDC-1HCl (11.25 mM) and higher concentrations tending towards quicker crystallisation. As previously outlined, the addition of an externalH Cl source is commonly thought to speed up the crystallisation process, [21] which these results appear to confirm, as the addition of 37 % HCl resultsi nconcomitant addition of about3 .3 moles of water per mole of HCl. However, nucleation onset occurs later than for the Zr-BDC-5H 2 O samples, suggesting that HCl can also slow nucleation by modifying reaction pH and hindering linker deprotonation. These results are af urther indication that the rate of crystallisation may depend more heavily on the rate of formation of the metal clusters rather than the diprotonation of the linker.
Monocarboxylic acid modulators, such as benzoica nd acetic acid, have become routine components in the syntheses of zir-coniumM OFs,h elping to achieve highly crystalline material while tailoring properties such as particle size and defectivity. [12,13] The addition of modulators of this kind is theorised to slow crystallisation by coordinative competition and reversibility,l eading to the formation of largers ized particles, althoughi tm ay also induceZ r 6 cluster formation and could influence reaction pH. The effect of benzoica cid concentration on the inductiont ime was determined;t he addition of 5equiv of benzoic acid (Figure4)l ed to an unexpected faster onset of nucleation,w ith Zr-BDC-5BA( 22.5 mM) crystallising approximately 1800 sb efore Zr-BDC (22.5 mM),s uggesting ac omplex interplayo fd ifferent processes. At this lower concentration, the modulator may be aiding in pre-organising the Zr 6 SBUs in the pre-crystallisation solutions, but not be at ah ighe nough concentration to inhibit the onset of nucleation.A st he benzoic acid concentration is increased to 10 and 20 equiv,t he inductiont ime can be seen to increase, with the onset being latest for Zr-BDC-20BA (22.5 mM),suggesting at higher modulator concentrations, the dominant effect is coordinative competition of the modulator for the Zr sites. It became apparent throughout the course of the study that repeating experiments on different days with different solutions often resultedi nd iscrepanciesb etween absolute induction times, although the reportedt rends are highly reproducible. It has previously been observed that aging stock solutions of precursors affects physicalp roperties such as particle size and porosity of the resultant UiO-66 material formed. [44] The time taken between preparings olutions and running turbidity measurements at increased temperaturec ould be akey experimental variable, but aging of the reagents before solution preparation was also thought to be ap otential notable variable. For example, the possibility of ZrCl 4 hydrolysing in humid air,r eacting with one mole of H 2 Ot of orm ZrOCl 2 and two moles of HCl, is ap rocess which may lead to reproducibility issues as potential molecular influences on kinetics are consumed (H 2 O) and released (HCl). [45] Additionally,Z rOCl 2 is thought to be ap recursor to the hexanuclear SBU of UiO-66; its direct use as source of Zr leads to very small nanoparticles and gels, [20,45,46] suggesting rapid nucleation.
Repeating turbidity measurements ( Figure 5) with af reshly openedb ottle of ZrCl 4 (Zr-BDC-fresh,r ed symbolsi nF igure 5) caused as ignificant differencei ni nduction times compared to syntheses using an "aged"b ottle (Zr-BDC-aged,b lack symbols in Figure 5), [47] with UiO-66n ucleatingm uch more slowly in the former case, confirming that partial hydrolysis of aged reagents can dramatically modify nucleation. Attempts to measure induction times by turbidity with ZrOCl 2 as the Zr source were hampered not only by the gelation of the solutionso n heating, but also by the production of very small nanoparticles reducingt he accuracy of the turbidity measurement. Qualitatively,i nductiont imes decreased substantially-theh ighest solution concentration crystallised before the hold temperature of 393 Kc ould even be reached-confirming the effect of changing Zr source to ZrOCl 2 .
Despite the large differencesi no verall crystallisation time between Zr-BDC-aged and Zr-BDC-fresh experiments,t he trend of the higher concentrations olutionsc rystallising slower is stilla pparent, along with lowt emperature leadingt os lower crystallisation. The differencesi na bsolute induction time across the experiments demonstrate the complexity of the crystallisation process of UiO-66, with slight differences in reagentsp otentially resulting in vast differences in nucleation. However, the induction time distributions reproduce the trendsa nd influences observed previously by in situ diffraction work, confirming that the technique is valuable for examining MOF nucleation behaviour.Asummary of the different influences on nucleation kinetics is given in Table 1.  earlier onset due to water content;less significant than water alone variability decreases as reagentc oncentrationincreases lowersp Hb ut also introducesw ater to synthesis;competinge ffects, but water additionh as greater influence modulation earlier onset at low modulator concentration, later at higher no change in variabilitym ultiple competing processes in solution;SBU pre-organisationw ould give earlier onset,b ut competition with linkers would give later onset aging of ZrCl 4 earlier onset less variability partial hydrolysisofZ rCl 4 during agingeffectively adds water to synthesis

Microwave synthesis
Following the study into how synthesis parameters alter the rate at which nucleation occurs, conditions were transferred to microwave assisted rapid synthesis. Microwave heating is common in MOF synthesis, producingh ighly crystalline material in af ractiono ft he time. [22,23] Whilst this heating mode does differ to conventional heating, we aimed to implement knowledge gained from the induction time study in order to facilitate rapid synthesis. It is postulated that the difference in heating, for example more localised, direct heating within the microwavec ompared with conventionalh eating,i st he leading reason for reduced synthesis times.I ts hould be noted that the induction times account fora20 min temperature ramp period,w ith time 0b eing the point at whicht he synthesis temperature is achieved, rather than the onset of heating. Comparatively,asample in ac onventionally heated oven will have am uch longeri nitial heating period due to the increased size of both the oven and sample volume. Microwave prepared samples, however,have amuch reduced heating period,reaching the required temperature in as little as 2min. This difference in heatingr ate should be kept in mind when comparing synthesis methods. The syntheses of Zr-BDC (45 mM), Zr-BDC-5H 2 O( 45 mM) and Zr-BDC-1HCl (45mM) were repeated in the microwavea t 393 K( see Supporting Information, Section S3), increasing the scale from 1mLt o1 0mL, to give MW-Zr-BDC-Unmod, MW-Zr-BDC-5H 2 O,a nd MW-Zr-BDC-1HCl,i nt urn. Samples modulated with 100 equivalents of acetic acid (MW-Zr-BDC-100AA) were also prepared to assess the effect of modulation.I no rder to ensure ac lose comparison, reactionp arameters were kept as close as possible when changing from Crystal16 to the microwave; 45 mm concentration solutions of ZrCl 4 and BDC in a 1:1r atio, with reactiont imes ranging from 1min to 2h.T he solids were assessed by PXRD, N 2 adsorption,a nd scanning electron microscopy (SEM), and notable differences were observed in the different microwave syntheses, varying with both reactiont ime and synthetic conditions (see summary in Figure 6a nd comprehensive data in the Supporting Information).
The Zr-BDC (45 mM) conditions were successfully reproduced via microwave heating (MW-Zr-BDC-Unmod), producing crystalline material in as little as 5min, although the crystallinity of the 5min sample is lessened comparedt ot he samples of longer synthesis times.W hen synthesis time was reducedt o 1min the resultant materialw as amorphous, as shown by PXRD (Figure 6a,a nd Supporting Information Figure S4). The surfacea reas of the materials were determined by BETanalysis, with samples produced in 10 min to 2h ranging from 889 m 2 g À1 to 984 m 2 g À1 ,a bout 80 %o ft he calculated surface area of pristine UiO-66. [23] As the synthesis time is reduced to 5min, the surfacearea drops to 533 m 2 g À1 ,with af urther drop to 195 m 2 g À1 when synthesis time is reduced to 1min (Figure 6b). This large decreasei ns urface area can be attributed to the productiono fa ni ncreasing amount of non-porous amorphous materialast he synthesis time decreases.
MW-Zr-BDC-Unmod-2 hr shows an approximate particle size of 200 nm, with generally well-defined octahedral morphology, however,aconsiderable amount of the sample is intergrown, forminga gglomerates ( Figure 6c). As the synthesis time decreases, the definition of the material can also be seen to decrease,w ith MW-Zr-BDC-Unmod-10 min being considerably more intergrown, and MW-Zr-BDC-Unmod-1 min lacking the distinctive octahedral shape, instead producing agglomerates of spherical material, as expected from the amorphous PXRD pattern.
Modulated syntheses typically resulted in more rapid microwave synthesis of crystalline UiO-66, as can be seen by plotting the BET surface areas of the samples versus synthesis time, which is consistent with the turbidityd ata (Figure 6b). Doping syntheses with 5equiv of water (MW-Zr-BDC-5H 2 O), led to the successful synthesis of materialinaslittle as 1min. Areduction in particle size to approximately 100 nm is also observed, implying fast nucleationf ollowedb yl imited growth as reactants are used up in the nucleation step. The surface areas of MW-Zr-BDC-5H 2 O samples are again slightly lower than predicted, rangingf rom 791 m 2 g À1 at 2hrs to 965 m 2 g À1 at 10 min. The thermals tability of both the unmodulated and water doped samples was examined, with all materials showingv ery high Figure 6. a) StackedP XRD patterns of selected microwavesyntheses compared with the pattern for UiO-66 predicted from CSD depositionRUBTAK. [17] b) Plot of BET surface area (N 2 ,7 7K)v ersus reaction time, showing that modulation in all cases producesc rystalline UiO-66 more rapidly at short reactiont ime. c) SEM imageso fselected samples showing the differingp article size and morphology induced by modulation. thermals tabilityu pt o7 73 K, typical of UiO materials (see Supporting Information, Figure S5). [48] Surprisingly,t he addition of 1equiv HCl in the microwave synthesis did not produce significant amountso fc rystalline materialr egardless of synthesis time (MW-Zr-BDC-1HCl),w ith PXRD patterns lacking distinct peaks and suggesting amorphous samples,a lthough smallr eflections consistent with UiO-66 may be visible, likelyc orresponding to as malla mount of nanocrystalline UiO-66a mongst primarily amorphous material (see Supporting Information, Figure S6 and exemplar in Figure 6a). The IR spectra of MW-Zr-BDC-1HCl samples also lack distinctive bands, whileS EM images show agglomerates of poorly defined particles of approximately 300 nm. N 2 adsorption analysis of MW-Zr-BDC-1HCl-1 min and MW-Zr-BDC-1HCl-2hr suggest that the samples are not be completely void of space, with surface areas of 374 m 2 g À1 and 360 m 2 g À1 ,r espectively,f urthers uggestive of am ixture of some crystalline UiO-66 and some amorphous coordination polymer material. The same conditions subjected to 24 hh eatingi nac onventional oven produces crystalline material; transferring conventional solvothermals yntheses to microwave heating is clearly not a guarantee of success.
The addition of 100 equiv acetic acid, also routinely used in Zr MOF syntheses, enabled the synthesis of highly crystalline material( MW-Zr-BDC-100AA)i na sq uickly as 1min (Figure 6a). The surface areas of MW-Zr-BDC-100AA show good agreement with the previouss amples, ranging from 946 m 2 g À1 at 1min to 1087 m 2 g À1 at 2h,w ith the isotherms suggesting considerable interparticle space (Figure 6b,a nd Supporting Information,F igure S7). SEM shows there to be ag ood particle size distribution, with particles being approx. 100 nm and becomingm ore discrete rathert han intergrown agglomerates (Figure 6c). Upon attempts to repeatt he synthesis through conventional heating, crystallisation was not achieved within the 24 hs ynthesis time, likelya st he acetic acid content is relatively high compared with other previously reported modulated syntheseso fU iO-66. [13,14,49] As such, synthesis with 10 equiv of acetic acid was attempted through both conventionala nd microwave heating; Zr-BDC-10AA under conventional heating produced UiO-66, while MW-Zr-BDC-10AA reproducibly yielded MIL-140A, [ZrO(BDC)] n ,amore condensed phase with infinite one-dimensional ZrO SBUs [43] (Figure 7). Typical syntheses for MIL-140A require much higher temperatures, reaching up to 220 8C, with suggestion that MIL-140 is the thermodynamic product and UiO-66 the kinetic. [43,50] In this case, the temperature is the lowest reported so far for isolation of MIL-140A.
Our analogousw ork with Fe MOFs has shown that modulation can induce formation of the thermodynamic over the kinetic product, [51] which may explain the low temperature synthesis of MIL-140A in this case. Microwave heatingh as also previously been proposed to selectively produce kinetic phases rather than thermodynamic phases;s ynthesis of MIL-53(Cr) and MIL-101(Cr), both formed from chromium precursors and terephthalic acid, produces mixed phases via conventional heating, whilem icrowave heating selectively produces phasepure MIL-101(Cr), the kinetically favoured product, most likely due to the faster kinetics of nucleation and crystallisation. [52] Clearly modulation complicates this process further,a sm icrowaveh eating in this case forms the thermodynamic product preferentially.T his is further illustrated by the formation of UiO-66a tl ower (MW-Zr-BDC-Unmod)a nd higher acetic acid concentrations (MW-Zr-BDC-100AA), suggesting some modulator-induced pre-clustering in the latter,a nd this is under further investigation.
Following our previous use of l-proline in solvothermal reactions, as well as the microwaves ynthesis of UiO-67, we assessed the convertibilityt ot he microwave synthesis of UiO-66 (MW-Zr-BDC-5L-Pro). [12] With 5equiv of l-proline added, alongside 1equiv of HCl as used in our previous report, we were again able to produce crystalline materiali na sl ittle as 1 minute, with particle size ranging from 50 nm to 500 nm, showingl arge spherical agglomerates of particles with characteristic octahedral morphology (Figure 6c,a nd Supporting Information,F igure S8). Thermogravimetric analysiss hows a gradualm ass loss startingf rom 473 K, indicating the inclusion of proline/prolinederivatives within the framework. This is confirmedb y 1 HNMR spectroscopy of digested samples,a sp eaks for proline and formyl-proline can be seen, along with formate peaks (Supporting Information, Figure S9). Despite this, the surfacea reas of MW-Zr-BDC-5L-Pro samples do not drop com- Figure 7. a) StackedP XRD patternsf or MW-Zr-BDC-10AA compared with that predicted for MIL-140A (CSD depositionZONBAH), [43] confirmingthe formation of MIL-140A ratherthan UiO-66.b)SEM imagess howing tetragonal plate morphologyo fs electeds amples, characteristic of MIL-140A. Scale bars 1 mm. pared with previous samples, suggesting l-proline to be incorporated into the framework rather than the pores, along with potentially an increasei nd efects. Similars amples produced solvothermally have recentlyb een utilised aso rganocatalysts. [53] As expected,t he use of zirconyl chloride (MW-Zr-BDC-ZrOCl 2 )r esultsi np articularly small nanoparticleso f< 20 nm in diameter,w hichi sr eflectedi nt he broad PXRD patterns,a nd indicative of rapid nucleation ( Supporting Information, Figure S10). Scherrer analysis of the main Bragg reflection at 2q = 7.328 gives ap articles ize of 10.4 nm, correlating wellw ith the SEM images. Nanocrystal formation is consistentw ith the use of zirconyl chloride in solvothermal syntheses which also produces particularly small particles. [20,46] These conditions led to the synthesis of ag el,w hich upon washing and drying gave a solid pellet which was ground down for analysis.C haracterisation of the material produced in reduced synthesis times matches closely with that of previously reported solvothermal material, confirming the convertibilityo ft hese reaction parameters.

Conclusions
The crystallisation kinetics of UiO-66 have been investigated using turbidity measurements, ac omparatively simple and fast technique which has been validated as an alternative route to probe MOF self-assembly compared to in situ diffraction analysis. Insight was gained into how severalr eactionp arameters, such as temperature and the use of common modulators, affect the speed of crystallisation. Varying reactiont emperature had al arge effect on the rate of crystallisation, with ar eduction from 393 Kt o3 73 Kl eadingt oc onsiderably slower formation of material. Theu se of both water and HCl as modulators sped up crystallisation, further confirming the same conclusions drawn by Ragon et al. in their energy-dispersive X-ray diffraction study. [20] Carboxylate-containing modulators showed more complex behaviour,w hilst the age of reagents, notably the hydrolysis of ZrCl 4 ,a lso resulted in significant variability in nucleation time. We expect that similara ging of DMF-breakdown, water absorption, and so on-will also induce variability, and so should be controlledi nf uture studies.
The information gained from the turbidity experimentso n the effect of modulator addition on nucleation onset was used to aid the rapid synthesis of UiO-66 through microwave assisted heating. Various syntheses were examined,e ach taking reaction time down to one minute, with addition of modulators resultingi nm ore rapid synthesis of crystalline, porousU iO-66. Despite reports of microwaves ynthesis often leading to smaller particles due to rapid nucleation, UiO-66 did not appear to show ar eduction in particles ize with reduced synthesis time. However,s urface areas were slightly lower than expected, suggesting rapid synthesis may result in some poorly crystalline or amorphousb y-products. Twon otable differences were observed comparing conventionalh eatingt om icrowaves ynthesis. Firstly, HCl did not effectively modulate microwave synthesis of UiO-66 in our hands, producing mostly amorphous material. Secondly, ar apid, low temperature, microwave-induced, acetic acid modulated synthesis of the alternative phase MIL-140A was discovered, suggesting that combining modulation with microwave heatingc ould lead to enhanced efficient syntheses of new phases and control between kinetic and thermodynamic products in the self-assembly of MOFs.