Concentration‐Dependent Seeding as a Strategy for Fabrication of Densely Packed Surface‐Mounted Metal–Organic Frameworks (SURMOF) Layers

Abstract The layer‐by‐layer (LbL) method is a well‐established method for the growth of surface‐attached metal–organic frameworks (SURMOFs). Various experimental parameters, such as surface functionalization or temperature, have been identified as essential in the past. In this study, inspired by these recent insights regarding the LbL SURMOF growth mechanism, the impact of reactant solutions concentration on LbL growth of the Cu2(F4bdc)2(dabco) SURMOF (F4bdc2−=tetrafluorobenzene‐1,4‐dicarboxylate and dabco=1,4‐diazabicyclo‐[2.2.2]octane) in situ by using quartz‐crystal microbalance and ex situ with a combination of spectroscopic, diffraction and microscopy techniques was investigated. It was found that number, size, and morphology of MOF crystallites are strongly influenced by the reagent concentration. By adjusting the interplay of nucleation and growth, we were able to produce densely packed, yet thin films, which are highly desired for a variety of SURMOF applications.

As ag eneral remark, the resultso ft he characterization of all samples by IR spectroscopy and X-ray diffraction are in line with formationo ft he desired MOF system. In some of the IR spectra,t here are hints on amorphous or even non-MOF material, as indicatedb y, for example, bands above 1700 cm À1 (compareF igure 1, left side). However,t he amount of this material is much lower than ten percent.V arying the concentrations of the pillar and linker molecules and the copper source, however, resulted in different amounts of depositedm aterial as was indicated by the QCM resultsa nd ar emarkable variety of crystallite shapes,sizes (compare the respective micrographs in Figure 2a nd in the SupportingI nformation), surface area densities (see Figure S42 in the Supporting Information) and roughnessvalue (see Figure S44).
To efficiently explore the concentration range, we went to the solubilityl imits of the respective reagents (pillar,l inker: 3mm,C u 2 + :2 0mm). As was demonstrated by the in situ QCM measurements, under these conditions more than five times of the material became deposited compared to the standard conditions (pillar,l inker:0 .1 mm,C u 2 + :1mm;see Figure S35 in the Supporting Information); [17] but the QCM sensorgram of this experiment does not show regularL bL steps [16, 29, 31-33, 35, 40, 41] (Figures S19 and S20 in the Supporting Information). Morphology and crystallinity of the deposit were unsatisfactory (compare SEM data in Figure S31). Surprisingly,b oth diffraction (Figure S33) and spectroscopic ( Figure S34) data neverthelesss uggest the formation of the desired MOF system.T oe lucidate at which concentration the transition of the ordered growth of crystallites into the observed disorder takes place, we varied the concentration of the ligand mixture (pillar,l inker), and the one of the Cu 2 + independently from each other.I nafirst LbL series,w ek ept the Cu 2 + concentration at 1mm and applied  pillar,l inker concentrations of 0.1 to 3mm.A ll resulting QCM curves are on first sight compliant with the assumption of a regularL bL process (black and gray curves in Figure 1, Figures S3-S12 in the Supporting Information);t he amount of depositedm aterial slightly increased with increasing pillar,l inker concentration ( FigureS35). Note, though, that the amount of deposited materialp er surface area is markedlyh igher than would be expected from the previously favored liquid-phase epitaxy model (addition of one MOF layer per cycle), by about one order of magnitude. Wei nterpret this outcomea sa nother indication that this model is not applicable. Note that also other studies on LbL growth of pillar-layer SURMOFsr eport deposition of more material than expected. [42] Nevertheless, at the lowest concentrations (pillar,l inker:b oth 0.1 mm,C u 2 + : 1mm)o ur previousr esults [17] could be reproduced (see SEM data in Figures 2a and S23, gray XRD and IR curves in Figure 3, as well as the XRD and IR data in Figures S33 and S34, respectively).T he analysis of the XRD data of these samples, based on ab imodal model (assuming that all crystallites are either (100) or (001) oriented) [17] suggestst hat the orientational quality increasesw ithi ncreasing pillar,l inker concentration ( Figure S40). But under these conditions, ad iscrepancy with the orientational data obtained by analysis of the IR spectra arises (also Figure S40). The SEM images reveal the likely reason for this:W hen the pillar and linker concentration is varied from 0.1 to 1mm,t he deposited crystals becomes maller,s tubbier and more numerous, and apparently the growth perpendicular to the (001) direction becamel ess dominant (Figures S24 and S25 in the SupportingI nformation). At ap illar,l inker concentration of 1.5 mm,t he crystals change from am ore plate-like into aconicalform (FiguresS26 and S27). Orientationsdifferent from (100) and (001) appear,m any of which are not detectable by XRD due to their high index nature,w hich leads to the discrepancy between the IR and the XRD results. We interpret the presence of such spiky structures to be correlated with a change in deposition mechanism, whichi sd ue to an increased supersaturationt hat induces ad iffusion-limited adhesive growth rather than ab irth-and-spread growth (Figure 4, center). [19] As additional orientations appear at higherp illar, linker concentrations, the simple binarym odel to evaluate the proportion of (001) orientation is not applicable anymore. To take this into account, we derived af ormula to obtain an average tilt angle from the IR data (see the Supporting Information). This tilt angle is indeed affected by the pillar,l inker concentration (see Figure S41), but the concentration effect on the orientation is not as high as the one of temperature. [17] To examinet he impact of the Cu 2 + concentration on SURMOF formation, we also performed LbL experimentsa t constant pillar,l inker concentration and varying Cu 2 + concentrations.A taCu 2 + concentration of 3mm,t he QCM sensorgrams becamea typical,a sc an be seen in the higherc ycle regime in Figure S13, in which the step height continuously decreases and the steps almost vanish. At c(pillar, linker) = 0.1 mm and c(Cu 2 + ) = 20 mm,d uring the exposure to copper solution,e ven al oss of materialw as observed (Figures S17 and S18 in the Supporting Information), whichc annot be explained within the liquid-phase epitaxy model. Instead, this finding is an indication of dissolving reactants and ap rocess of crystallization of the MOF out of as olutiont hat contains all reactants at the same time (Figure 4), am echanism that has already been proposed for other SURMOFs ystems. [18,24] Although both the XRD and the IR data of theses amples do not show as ignificant deviation from the classical c(Cu 2 + ) = 1mm,c (pillar, linker) = 0.1 mm experiment (see FiguresS33 andS 34 in the Supporting Information), the SEM images ( Figures S28 and  S30) show ac onsiderable twinningo ft he deposited crystallites at the higherC u 2 + concentrations.A lthough at 3mm Cu 2 + , the amount of material determined by QCM was similart ot he 1mm Cu 2 + case, twice as much of the SURMOF material becamed epositeda t2 0mm Cu 2 + concentration ( Figure S35). As was shown by the micrographs, this is caused by as ignificantly higher density of crystallites at the latter case, although the size of the crystalsw as similarf or both concentrations (compareF igures S28 and S30). Although no spikes were found in these cases,f or 3mm Cu 2 + and the highest pillar, linker concentration of 3mm,apronouncedf ormation of spikes was observed ( Figure S29).
Obviously,t his strong dependence of the number,s ize, and morphology of the MOF crystallites on reactantc oncentrations is ac onsequence of the competition of nucleationa nd growth of the MOF crystallites, combined with the transition between different growth modes,t hat is, birth-and-spread versus diffusion limited growth. [19] On the one hand, higherr eactant concentrationsc learly prohibit aS URMOF buildup following the liquid phase epitaxy model. On the other hand, even at the lowest concentrationsa tw hich LbL can be executed, there is lots of evidencet hat all reactants are presenti nt he solution at the same time and that MOF growth actually is a( pseudo-) equilibrium crystallization process out of this solution rather than liquid-phasee pitaxy [17,[22][23][24] (see Figure 4). In line with this model,p rolonging the deposition time does not increaset he amount of deposited material( see the results of an according experiment in FiguresS45-S48 in the Supporting Information).
Based on these considerations,w ef iguredt hat we could achieve the deposition of al ow amount of MOF materialt hat yet form ac losely packed layer on the substrate surfaceb y making use of varying concentrationsd uring the LbL procedure. Ah igh reactant concentration during earlyL bL cycles providesahigh number of crystallization nuclei, whereas low concentrations later in the LbL procedure limits the amount of deposited MOF material ( Figure 4, right column). We kept the Cu 2 + concentration constanta t1 mm and applied c(pillar, linker) = 3mm in the first cycle and 0.1 mm in the subsequent cycles ("high-concentrations eeding"). IR and XRD (Figure 3, red curves and Figures S33 and S34) suggest that the Cu 2 (F 4 bdc) 2 (dabco) MOF was formed in this experiment.S EM reveals that after high-concentration seeding, the substrate is completely covered with crystallites smaller and more numerous than after LbL with constantr eactantconcentrations (compare Figure 2c with a, also Figures S32 with S23i nt he Supporting Information). We also found the desired regular behavior of the QCM sensorgram-with as light decrease in the frequencys teps at higherc ycles (red curve in Figure 1, Figures S21 and S22). As was assumed,t he quantity of material deposited by high-concentrations eeding turned out to be lower than in the case of the conventionalL bL experiments at pillar,l inker concentrations of 3.0 mm ande ven 0.1 mm (Figures 1and S35 in the Supporting Information). From astatistics of crystallites formed during the LbL procedure ( Figure S42), we estimate the number of initial nuclei to be higherb ya factor of approximately 40 in comparison to the c(pillar, linker) = 0.1 mm case. The large initial number of nuclei in the high-concentration seeding experiment is the reasonf or both the great number and the small size of the crystallites formed and for the lower amount of deposited material, because the coalescence of the crystallites limits their respective growth during the LbL experiment (transition from 3D to 2D growth). Note that in the high-concentration seedingc ase, the crystallites showaslightly broader size distribution and are less oriented, butare still plate-like (Figure 2c)asinthe standardconcentrationc ase (Figure 2a)a nd in contrast to the spike-like morphologya th ighp illar,l inker concentration ( Figure 2b). Moreover,t he high concentration seeding sample exhibited ar elatively low roughnessc ompared to all other samples (see Figure S44). This appearst ob ei nl ine with the fact that the crystallites produced by high concentrations eeding are markedly smalleri nc omparison to the ones from all otherL bL experiments (compare Figure S43;n ote that these data should be taken with ag rain of salt due to the high variation of crystallite shapes that limits the comparability between the samples).
In conclusion, by systematically varying the reactant concentrations in LbL SURMOF deposition experiments, we found a strong impact on the growth mechanism resulting in different density,s ize, and morphology of the MOF crystallites, which adds on top of the Volmer-Weber growth mechanism. Consequently,w ee mployed this concentration dependence as a new tool to control the SURMOFL bL growth that joins surface functionality, [13][14][15][16][17] temperature, [17,20] ands urface energy. [18] We found that exploiting the competitionb etween seedinga nd growth in SURMOF formation [19] by highc oncentration seeding during the first step of the LbL procedure opens the possibility Figure 4. Scheme of the mechanism of the concentration-dependent SURMOF deposition. Left:Atl ow concentrations, equilibria are established within arelatively wide diffusion layer (light blue background). This results in alow density of nuclei, but in very defined crystallites at laterstages. Closed layers are only obtained whenthe size of the crystals exceeds their average distance. Middle:Ath igh concentrations, nucleation is very efficient, but due to av ery narrow diffusion zone the crystals become dendritic. Right:Byc ombiningthe processes, dense layers of nuclei can be obtained, which become closely packeda fter only few depositionsteps. The crystalsare well defineddue to the equilibrium formation,b ut growth is slow due to 2D transport. Other phenomena, such as Ostwald ripening, material storage,a nd the orientational disorder are omitted for clarity.
for the fabrication of closely packed SURMOFsy et at low depositeda mounts and at am arkedlyl ower number of LbL cycles than otherwise necessary to fully cover the substrate surface. This new approach further expands the control of the growth of tailor-made SURMOF systemsf or specific applications and in addition offers valuable time and material savings in the SURMOF production process.