Introduction
A vigorous search for novel membrane materials has been stimulated by the growing demand of energy-efficient separations [
1–
3]. Polymer membranes have been extensively investigated and applied in industrial gas separation [
4], reverse osmosis [
5], etc. due to their easy processing and mechanical strength. Inorganic membranes, for instance, zeolite membranes, have been successfully used for organic dehydration [
6]. However, polymer membranes always suffer from low chemical and thermal stability, while zeolite membranes possess issues of brittleness, limiting their applications. Metal-organic frameworks (MOFs) [
7], a class of hybrid materials constructed by coordinating metal-containing units with organic ligands, have received tremendous attention from membrane scientists in virtue of their versatile topologies and customizable chemistry. The teams of Caro [
8] and Kapteijn [
9] reported the earliest Metal-organic frameworks (MOF) films in 2007 and 2008, respectively, and later on in 2009, a few MOF membranes were explored for gas separation [
10–
15]. In the past 10 years, targeting to various separations, a booming development of MOF membranes has taken place. MOF-5, HKUST-1, IRMOF, ZIF, MIL and CAU membranes have been studied extensively [
6,
16–
22]. However, they were always accompanied by concerns about the hydrothermal and chemical stability [
23], which ultimately limited their further application in industries.
Recently, zirconium(IV)-carboxylate MOFs (Zr-MOFs) [
24] have emerged as promising membrane materials due to their exceptionally high stability. According to Pearson’s hard/soft acid/base principle [
25], strong coordination bonds are expected by joining Zr
4+ (hard Lewis acid) and carboxylate based ligands (hard Lewis base) to determine the thermodynamic stability of Zr-MOFs. Besides, tetravalent Zr ions require more ligands to balance their charge and thus highly connected frameworks are formed with sufficient steric hindrance against attacks, which guarantees the structural stability from the kinetic aspect [
26].
UiO-66 (UiO stands for University of Oslo) [
27] is a prototypical and pioneered Zr-MOF, with the formula Zr
6O
4(OH)
4(BDC)
6 (BDC= 1,4-benzene-dicarboxylate) (Fig. 1(a)). It was firstly reported by Lillerud’s group [
27]. The inorganic brick of UiO-66 consists of a secondary building unit (SBU) Zr
6O
4(OH)
4 core, bridged by 12 carboxylates (–CO
2) originating from the dicarboxylic acids forming a face-centred cubic lattice (fcu topology). The diameters of the octahedral and tetrahedral cavities are ~1.1 and ~0.9 nm, respectively, and the triangular aperture size is
~0.6 nm as estimated from crystallographic data (Fig. 1(b)) [
27,
28]. Since its discovery in 2008, UiO-66 has been extensively investigated in the synthesis of its analogues [
29]. Tetravalent metal ions (Fig. 1(c)) and BDC type ligands (Fig. 1(d)) were attempted to add to the family members of UiO-66. More than 40 of UiO-66 analogues have been confirmed [
29]. Over the last few years, UiO-66 has almost dethroned MOF-5, HKUST-1, ZIF-8 and MIL-101 as a benchmark MOF material.
There are two main types of UiO-66 membranes: supported polycrystalline membranes and mixed matrix membranes (MMMs) [
30–
84]. This review focuses on the study of polycrystalline UiO-66 membranes (shortened to UiO-66 membranes unless otherwise stated). However, their development was hindered. During the growth of UiO-66, the high charge density Zr
4+ polarizes the Zr-O bond to present covalent character, slowing down the ligand exchange rate [
24]. In this case, it is unfavourable for defect repair during the crystallization process. Consequently, UiO-66 powders with poor crystallinity are harvested after a relatively long reaction time. Low nucleation density and poor intergrowth were reported in the fabrication of UiO-66 membranes.
The silence of the absence of dense UiO-66 membranes was broken by Liu et al. in 2015 [
28]. After a thorough optimization of the preparation parameters (composition of mother solution, duration of synthesis and substrates), well-intergrown UiO-66 membrane was fabricated on
α-alumina hollow fibers and applied for water desalination. As stated by Liu et al. [
28], high nucleation density of UiO-66 and satisfactory intergrowth could be achieved by adjusting the afore-mentioned preparation parameters. Water in the mother solution played an essential role [
85], because the SBU of UiO-66 contains OH
– ions in addition to O
2– ions. Subsequently, a few continuous UiO-66 membranes supported on varying substrates were reported [
86–
89].
UiO-66 membranes were further developed by using modulated synthesis [
59,
90–
96]. The so-called coordination modulation method was initially proposed by Tsuruoka et al. [
97] and employed in UiO-66 crystal preparation by Schaate et al. [
98]. Modulated synthesis of UiO-66 refers to regulating the coordination equilibrium by introducing modulators (e.g., formic or acetic or benzoic acid) as the organic ligands used to hinder the coordination interaction between Zr
4+ and BDC ligands [
98]. As a result, the competitive reaction can adjust the rate of nucleation and crystal growth, improve the reproducibility of synthesis procedures and tune crystal features such as size, morphology, and crystallinity. That is in essence the reason why the fabrication of UiO-66 membranes benefited from modulated synthesis.
In this article, we review five approaches for preparing UiO-66 membranes and films, discuss their applications in gas, liquid and ion separations, and provide future perspectives on the development of their preparation methods and potential applications. Such a review about the specific MOF UiO-66 membranes aims to provide guidance for their in-depth investigation from basic research to practical application.
Approaches for fabricating UiO-66 membranes
Continuous growth of UiO-66 results in either a polycrystalline or an epitaxial film. As free-standing films were not mechanically robust, porous and nonporous substrates were employed to support UiO-66 membranes [
28,
59,
86–
96] and films [
96,
99–
112], respectively. Porous metal and ceramic substrates with minimal permeation resistance in the configuration of flat sheets and tubes were adopted for supporting UiO-66 membranes. The main task for the synthesis of high-quality UiO-66 membranes is to control heterogeneous nucleation, crystallization and intergrowth on the substrate surface, and minimize the nonselective intercrystal pinholes. The quality of UiO-66 membranes is assessed in terms of crystal structure and morphology and separation capacities. A variety of synthesis methods have been explored for obtaining UiO-66 membranes and films, such as
in situ synthesis, secondary synthesis, biphase synthesis, gas-phase deposition and electrochemical deposition (Tables 1 and 2). The synthesis of UiO-66 membranes can be analogous to that of other MOF membranes [
16,
17,
19] and zeolite membranes [
6,
113], as they are all crystalline porous materials.
In situ synthesis
In situ synthesis is defined as when a porous substrate is immersed in the mother solution without any UiO-66 crystals previously attached to the surface. The nucleation, growth and intergrowth of UiO-66 crystals on the substrate all take place during a single fabrication step.
As exemplified by Liu et al. [
86], UiO-66 polycrystalline membranes were fabricated on the prestructured yttria-stabilized zirconia hollow fibers (YSZ HF) by an
in situ solvothermal approach via a thorough optimization of the heating duration, composition, and temperature of the synthetic mother solutions. As depicted in Fig. 2(a), after 2 h of heating, a very thin amorphous gel layer was formed on the top of the substrate. This was possibly caused by the aggregation of gel particles originating from the mother solution, which were transported to the substrate due to chemical interaction between the ligands and substrate, and Brownian motion. During the consequent synthesis, heterogeneous nucleation occurred probably at the interface of the gel and the solution (Fig. 2(a)), the only place where both the metal and ligand source were present in abundance. In parallel, further gel settlement could still be proceeding, which buried and disturbed the UiO-66 nuclei. Afterwards, crystals propagated through the gel network and then sank to the substrate by consuming the gel around them. Meanwhile, the aggregation and densification of nanocrystals occurred. With prolonged heating, crystal growth occurred (Fig. 2(a)) by acquisition of nutrients from the bulk solution, from nearby unreacted amorphous gel and small UiO-66 crystals (Ostwald ripening). A well-intergrown membrane layer (Fig. 2(a) and (c)) was finally fabricated after continuous heating for 48 h.
As stated by Liu et al. [
86], since this membrane was fabricated with simultaneous growth and nucleation, UiO-66 crystals emerging on the surface of the membrane layer were identified in the EDX mapping image (Fig. 2(c)). FTIR-ATR characterization indicated that chemical bonds were established between the UiO-66 ligands and substrate, probably between the carboxyl and zirconium. This chemical interaction provides evidence for disclosing the energy-dispersive X-ray spectroscopy (EDXS) mapping (Fig. 2(c)). Although no visible UiO-66 crystals were found in the bulk substrate (Fig. 2(j)), slight intrusion of the C signal (yellow) into the substrate (green) was observed. This might be because the substrate was chemically modified by the BDC ligands during membrane preparation. The chemical interaction could boost the adhesion of the membrane layer to the substrate to a large extent, improving membrane stability.
Viability of the
in situ synthesis of UiO-66 membranes was tested on varying substrates with different shape and roughness, for example,
α-alumina hollow fibers (Fig. 2(b)) [
28], micropatterned YSZ sheets (Fig. 2(d)) [
87], and nanochanneled polyethylene terephthalate (PET) films (Fig. 2(e)) [
88]. Apart from fabrication on bare substrates, modified substrates were adopted to facilitate membrane growth, such as ZrO
2 [
90] and 3-aminopropy-ltriethoxysilane (APTES) modified
α-alumina tubes [
89]. In addition, attempts were made to obtain UiO-66 films by
in situ synthesis using bare ZrO
2 fibrous mats [
101], and modified substrates including fluorine-doped tin oxide (FTO) glasses [
99], polyurethane (PU) foams [
100], polyacrylonitrile (PAN) fibers [
102], silanized
α-alumina and ob-SiC foams [
103].
Secondary growth
Secondary growth is defined as when a porous substrate is immersed in the mother solution with UiO-66 crystals previously attached to the surface. In comparison with the in situ synthesis, the nucleation and growth of polycrystalline membranes can be well-balanced by the secondary growth method.
Modulated synthesis was used in the case of secondary growth of UiO-66 membranes and films. Larger and isolated UiO-66 crystals were always produced with the addition of a modulator, whereas microsized intergrown UiO-66 crystals were yielded without modulation. The addition of monocarboxylic acid modulator could probably form complexes with zirconium cations [
98]. Molecular zirconium complexes with different monocarboxylic acids (HO
2CR, R=
t-Bu, C(CH
3)
2Et, etc.) [
114,
115] and structures similar to that of the SBU in UiO-66 have been described. Such complexes could act as intermediates and the framework construction would then proceed through an exchange between modulator and linker molecules at the coordination sites of the zirconium ion [
98]. The application of modulators would decrease the possibility that the linker is connected to the SBU. Therefore, the formation of framework nuclei is disfavored, thus promoting the incubation of larger UiO-66 crystals. Furthermore, modulators can inhibit UiO-66 crystal growth in the (111) direction, leading to the formation of octahedral crystalline configurations rather than the cubic lattices generated from the original synthesis [
27].
Friebe et al. [
95] reported (002)-oriented UiO-66 membranes employing secondary growth with benzoic acid as a modulator. The growth started from randomly oriented seed crystals until they contacted each other. Afterwards, the crystals grew along the direction with the highest growth rate (i.e., (002)), thus building the top layer of the membrane. The SEM top view in Fig. 3(a) shows the tips of the UiO-66 octahedrons, in good accordance with the model of the van der Drift growth [
116]. The cross-section images reveal a 5
mm thick layer with a high orientation and the tilting angle of the octahedrons is around 15° (Fig. 3(a)).
Taking advantage of the uniform size and shape of the octahedral UiO-66 crystals, Lu et al. [
117] produced large-area 2D oriented monolayers on a water surface through a liquid-air interfacial assembly technique (Fig. 3(b)). The obtained monolayers can be further transferred easily to a silicon platform and (111)-oriented UiO-66 films with long-range 3D superlattices can be formed (Fig. 3(b)). Furthermore, UiO-66 films with preferred (111) orientation were fabricated by repeated solvothermal synthesis (Fig. 3(c)) [
105].
Biphase synthesis
As claimed in the modulated synthesis, isolated UiO-66 crystals were always produced instead of intergrown ones. An interpretation was provided by Shan et al. [
96] that the partially deprotonated ligand caused by the accumulated protons in the reaction solution is the key factor preventing the intergrowth of the UiO-66 crystals (Fig. 4). With the addition of a deprotonating agent, trimethylamine (TEA), in an
in situ biphase solvothermal reaction (Fig. 4), well-intergrown UiO-66 membranes and films were fabricated with tunable (200) and (111) orientations. As shown in Fig. 4(b), a hexane-dimethylformamide (DMF) biphase system was designed. TEA was initially dissolved in the hexane phase, and metal and ligand sources were charged in the DMF phase together with the modulator. Since the TEA could diffuse from the hexane phase to DMF phase and act as a deprotonating agent, the quantity of partially deprotonated ligands were efficiently reduced. Finally, the intergrowth between UiO-66 crystals was facilitated, affording dense membranes.
Gas-phase deposition
Atomic layer deposition (ALD) [
118–
122] in a mode also known as molecular layer deposition (MLD) is a technique where two or more precursors are individually pulsed into a reaction chamber in the gas phase and left to react with and saturate the surface of a substrate. When the surface is saturated by the first precursor, excess precursor is carried away by purging with an inert gas, and then the second precursor is applied in the same way. A thin-film is constructed with a thickness of one atomic layer or one molecular layer at a time by reiterating these steps in a cyclic process.
Lausund et al. [
106] deposited UiO-66 thin films in an all-gas-phase process by the aid of ALD. Sequential reactions of ZrCl
4 and 1,4-benzenedicarboxylic acid formed amorphous organic–inorganic hybrid films that are crystallized to the UiO-66 structure after the treatment in acetic acid vapour (Fig. 5(a) and Table 2). The stoichiometry between metal clusters and organic linkers was well controlled by modulation of the ALD growth with additional acetic acid pulses. Unlike other fabrication methods, which rely on solvothermal nucleation and growth, the all-gas-phase method is based on scalable, solvent-free, seed-free, thickness-controllable, a well-established material processing technology to coat irregular substrates.
By applying vapor-assisted conversion (VAC) [
108], highly oriented thin films of UiO-66 and UiO-66(NH
2), were produced on a variety of surfaces—bare gold, gold surfaces modified with thiol SAMs, and bare silicon (Fig. 5(b)). The obtained MOF films are well intergrown and possess a high degree of crystallinity and crystal orientation extending to large areas. The relationship between the rate of crystallization and formation of the oriented MOF film was revealed by adjusting the parameters including modulator equivalents, precursor concentration, temperature, and reaction duration.
Electrochemical deposition
Electrochemical MOF deposition [
123] has been proposed as a promising method for
in situ deposition and patterning on conductive surfaces on the basis of two different mechanisms corresponding to anodic and cathodic deposition. In anodic deposition, MOF film formation occurs on a metal anode in contact with a ligand solution in virtue of the release of a critical concentration of metal ions by anodic dissolution [
124]. On the other side (cathodic deposition), a solution containing metal ions, ligands, and a so-called probase is put in contact with a cathodic surface. Film deposition in this case relies on an increase in pH near the cathodic surface, where electrochemical reduction of the probase leads to local base generation and subsequent ligand deprotonation, enabling MOF formation [
125].
As demonstrated by Stassen et al. [
110], electrochemical deposition of the UiO-66 and its isoreticular analogues has been identified and elucidated. The crystallite size, film morphology, together with the deposition mechanism were rationalized through synthesis modulation. Whereas anodic deposition results in superior adhesion of the MOF layer onto the metallic zirconium substrate, which is attributed to the formation of an oxide bridging layer (Fig. 6), cathodic deposition has the merit of broad substrate flexibility.
Electrophoretic deposition (EPD) was used for the patterned growth of UiO-66 thin films on conductive glasses [
109]. EPD is a well-established technique for depositing thin films, especially from nanoparticulate building blocks. The application of a DC electric field to a suspension composed of charged particles and nonpolar solvent can result in particle transport and deposition onto a conductive substrate [
109]. During the synthesis of UiO-66, some surface defects are present (possibly due to missing metal nodes), which will give rise to partially negative charges on its surface. During the EPD process, those negative charges drive the particles toward the positively charged electrode and fabricate films.
Applications of UiO-66 membranes
Applications of UiO-66 membranes were predominately located in the separation field. The effective aperture size and functional groups of the UiO-66 type MOFs determine the membrane separation capability as predicted by molecular sieving and adsorption-diffusion mechanism. The flexibility of framework, missing ligand defects and substitutes on the ligands redefine the aperture size of UiO-66 rather than the 0.6 nm as estimated from crystallographic data. The functional groups of UiO-66 type MOFs are abundant. The OH groups from SBU, as well as the phenyl and the substituent groups from the ligands provide versatile adsorption sites. Herein, we discuss four categories of application based on membrane processes (Tables 1 and 2): gas separation, pervaporation, nanofiltration and electrochemical ion separation.
Gas separation
Liu et al. [
28] applied the UiO-66 membranes constructed by
in situ synthesis to gas separation. The gas permeance did not follow the kinetic diameters of the gases because of the larger aperture size of UiO-66 (
~6.0 Å) (Fig. 7(a)). Figure 7(b) shows the kinetic diameters of the studied gases. The H
2 permeance was ca. 7.2 × 10
-7 mol·m
-2·s
-1·Pa
-1, with a high H
2/N
2 ideal selectivity of 22.4. Owing to the effect of preferential CO
2 adsorption, the permeance of CO
2 (9.5 × 10
-7 mol·m
-2·s
-1·Pa
-1) is higher than that of all the other studied gases, leading to a satisfactory CO
2/N
2 separating selectivity (29.7). As claimed, UiO-66 was a good membrane material for the purpose of H
2 purification and CO
2 capture. The similar order of gas permeation was recently confirmed by Wu et al. [
126].
Gas separation was also performed by Friebe et al. [
95] using (002) orientated UiO-66 membranes fabricated by secondary growth with modulated synthesis. Different from the above observation, the permeance of CO
2 was lower than that of H
2 and N
2 (Fig. 7(b)). The permeance of H
2 is the highest compared with that of the other gases (N
2, CO
2, CH
4, C
2H
6, C
3H
8), and the permeance decreased significantly with kinetic gas diameter, which seems to be in good accordance with the concept of molecular sieving. The selectivity of H
2/C
3H
8 reached the highest, being 28.5. Furthermore, the gas separation performance of UiO-66 type membranes was investigated by Liu et al. [
92], Miyamoto et al. [
91] and Shan et al. [
96].
Pervaporation
In 2017, Liu et al. [
28] reported UiO-66 membranes for organic dehydration. The membrane was activated on-stream and remained robust after being treated with boiling benzene and water. No discernible degradation of membrane performance was recognized in the following 200 hours’ stability test for water/
n-butanol and water/furfural separation even sulfuric acid was introduced (Fig. 8(a)). At higher temperature (e.g., 80°C), the membranes exhibited a very high flux of up to ca. 6.0 kg·m
-2·h
-1 and great separation factor (>45000). This performance, in terms of separation factor, is 10–100 times of commercially available silica and polymeric membranes with equivalent flux. The resistance against harsh environments was superior to commercial zeolite NaA membranes. Under the same
in situ synthesis condition, Huang et al. [
87] fabricated UiO-66 membranes on micropatterned YSZ substrates. The resulting membranes displayed a 100% improvement in the apparent water permeation flux over conventional flat UiO-66 membranes without compromising the separation factor of water over
n-butanol.
Wu et al. [
90] synthesized UiO-66-NH
2 membranes for pervaporative desulfurization with an optimum permeation flux of 2.16 kg·m
-2·h
-1 and a separation factor of 17.86 under 40°C for 1312 ppm thiophene/
n-octane mixtures. The separation factor is higher than polymer-based membranes in the literature (Fig. 8(b)). As evidenced, the preferential adsorption of thiophene is an important contribution to the selectivity. Moreover, the studies from Wan et al. [
89], Miyamoto et al. [
91] and Wu et al. [
93] extended the applications of pervaporation to other systems.
Nanofiltration
UiO-66 membranes were applied in water-softening for multivalent and trivalent cations rejection (86.3%, 98.0%, and 99.3% for Ca
2+ (0.82 nm), Mg
2+ (0.86 nm), and Al
3+ (0.95 nm), respectively) in light of the size exclusion effect (Fig. 9(a)) [
28]. Although the diameter of hydrated monovalent ions (Cl
-: 0.66 nm, K
+: 0.66 nm, and Na
+: 0.72 nm) exceeded the aperture size of UiO-66 (
~6.0 Å), the rejections were moderate (i.e., 45.7% and 47.0% for K
+ and Na
+, respectively). Two possible reasons were proposed: one is the ligand dynamics of UiO-66 [
127] because its carboxylate groups can change their coordination mode from edge-bridging to monodentate; the other is the missing-ligand defects [
128–
130] in the UiO-66 crystals.
Wang et al. [
94] mitigated the ligand-missing defects in UiO-66(Zr)-(OH)
2 membranes by postsynthetic defect healing (PSDH), boosting the Na
+ rejection rate by 74.9% (from 26% to 45%), and achieved a perfect block of methyl blue (from 98.7% to 99.8%) (Figs. 9(b) and 9(c)). As anticipated, the membranes display excellent hydrothermal stability in aqueous solutions (>600 h).
Electrochemical ion separation
Zhang et al. [
88] reported UiO-66 membranes for ultrafast selective transport of alkali metal ions. The resulting membranes can preferentially transport Li
+ over other alkali metal ions following unhydrated size exclusion mechanism, with the ion transport rate order of Li
+>Na
+>K
+>Rb
+. The LiCl/RbCl selectivity is of ~1.8, which outperforms the LiCl/RbCl selectivity of 0.6–0.8 evaluated in traditional membranes (Fig. 10(a)). This study may potentially open up a new avenue for efficient ion separations in the future.
Cyclic voltammetry (CV) experiments were conducted to assess the molecular sieving capability of UiO-66 films supported on FTO using redox-active species (including Ru(NH
3)
63+ (diameter ca. 0.55 nm) and Fe(phen)
32+ (diameter ca. 1.3 nm)) as probes [
99]. The UiO-66 coated electrodes showed moderate CV signals for Ru(NH
3)
63+ but were not responsive to Fe(phen)
32+, verifying their size-selective accessibility to these two species, which is in line with the fact that the pore aperture of UiO-66 (0.60 nm, estimated from crystallographic data) is between the diameter of Ru(NH
3)
63+ and Fe(phen)
32+. The ion discrimination of UiO-66 film (healing with polydimethylsiloxane (PDMS)) was further evidenced by the electrochemical study in a mixed solution of Ru(NH
3)
63+ and Fe(phen)
32+, where well-defined redox peaks were observed only for the former species (Fig. 10(b)).
Conclusions, remarks and perspectives
With adequate members of the UiO-66 family and exceptionally high stability, UiO-66 based membranes stand out from MOF membranes as well as novel porous material membranes for organic purification under harsh conditions.
Regarding synthetic protocols,
in situ synthesis is a facile method for fabricating UiO-66 membranes. Electrochemical deposition will be a promising method for coating the membranes on devices. Although secondary growth is the benchmark method for large-scale production of polycrystalline zeolite membranes, gas assistant deposition [
119] and interfacial synthesis [
131] may have opportunities in scaled-up synthesis of UiO-66 membranes.
Precise separation is one of the future directions for membrane-based separation. The author speculates that there would be some optimal preparation conditions where the UiO-66 membranes have the opportunity for (i) separating isomers of hydrocarbons; furthermore, (ii) purification of organics under harsh conditions may offer the other position for UiO-66 membranes.
New membrane materials are always accompanied by challenges. (i) In line with the principle of green chemistry, water [
132,
133] is more welcome than DMF as an alternative solvent, which reduces the cost of UiO-66 membranes. Consequently, systemic optimization of synthetic variables is required. (ii) Novel zirconium sources are desired because the usual metal source ZrCl
4 requires careful storage to avoid deliquescence. (iii) Scalable synthesis requires a clear understanding of the membrane reproducibility and substrates.