Using Supercritical CO2 in the Preparation of Metal-Organic Frameworks: Investigating Effects on Crystallisation

In this report, we explore the use of supercritical CO2 (scCO2) in the synthesis of well-known metal-organic frameworks (MOFs) including Zn-MOF-74 and UiO-66, as well as on the preparation of [Cu24(OH-mBDC)24]n metal-organic polyhedra (MOPs) and two new MOF structures {[Zn2(L1)(DPE)]∙4H2O}n and {[Zn3(L1)3(4,4′-azopy)]∙7.5H2O}n, where BTC = benzene-1,3,5-tricarboxylate, BDC = benzene-1,4-dicarboxylate, L1 = 4-carboxy-phenylene-methyleneamino-4-benzoate, DPE = 1,2-di(4-pyridyl)ethylene, 4.4′-azopy = 4,4′- azopyridine, and compare the results versus traditional solvothermal preparations at low temperatures (i.e., 40 °C). The objective of the work was to see if the same or different products would result from the scCO2 route versus the solvothermal method. We were interested to see which method produced the highest yield, the cleanest product and what types of morphology resulted. While there was no evidence of additional meso- or macroporosity in these MOFs/MOPs nor any significant improvements in product yields through the addition of scCO2 to these systems, it was shown that the use of scCO2 can have an effect on crystallinity, crystal size and morphology.


Introduction
Use of supercritical carbon dioxide (scCO 2 ) has been extensively studied as a way to incorporate stable permanent porosity in materials during various processing steps including crystallisation [1][2][3], impregnation [4], dispersion [5], drying [6], and activation [7,8]. The study of porous materials, which can be either ordered structures (including zeolites, metal-organic frameworks (MOFs) and silicates), or disordered materials (including activated carbons, ceramics, metals, and polymers), has been listed among the fastest growing research areas of recent years. In this field, MOFs, normally microporous or mesoporous crystallites constructed by the coordination of transition-metal nodes and organic linkers, have emerged as enabling materials for a wide variety of potential applications, showing promise for  MOF syntheses in scCO2 are normally carried out in the presence of organic liquids such as DMF and DMSO to increase the solubility of both polar and nonpolar compounds in the reaction. It has been shown that using ligands such as tert-butylpyridine (t-bpy) and 4,4′-bipyridine can increase the solubility of the reagents in the reaction media, resulting in more efficient use of scCO2 [41,42,55]. Here, we employ the flexible reduced Schiff-base compound 4-carboxy-phenylene-methyleneamino-4-benzoic acid as a linker; we note that the biphenyl-4,4′-dicarboxylate linkers have been used with great success in MOF synthesis [56][57][58][59].

Materials and Methods
All reagents were purchased from commercial sources and used without further purification. Due to constraints on CO2 pressure, MOF synthesis at low temperature was preferred. In this study, 40 °C   . Precursor solution where MOFs/MOPs were formed without additional antisolvents is referred to as "precursor solution type I" and precursor solution where MOFs/MOPs crystallised only in the presence of an antisolvent is referred to as "precursor solution type II".
MOF syntheses in scCO 2 are normally carried out in the presence of organic liquids such as DMF and DMSO to increase the solubility of both polar and nonpolar compounds in the reaction. It has been shown that using ligands such as tert-butylpyridine (t-bpy) and 4,4 -bipyridine can increase the solubility of the reagents in the reaction media, resulting in more efficient use of scCO 2 [41,42,55]. Here, we employ the flexible reduced Schiff-base compound 4-carboxy-phenylene-methyleneamino-4-benzoic acid as a linker; we note that the biphenyl-4,4 -dicarboxylate linkers have been used with great success in MOF synthesis [56][57][58][59].

Materials and Methods
All reagents were purchased from commercial sources and used without further purification. Due to constraints on CO 2 pressure, MOF synthesis at low temperature was preferred. In this study, 40 • C was chosen as the synthesis temperature for Zn-MOF-74, Zn-BTC, [Cu 24 (OH-mBDC) 24 ] n metal-organic polyhedron, {[Zn 2 (L 1 )(DPE)]·4H 2 O} n and {[Zn 3 (L 1 ) 3 (4,4 -azopy)]·7.5H 2 O} n . Structures of linkers and pyridine derivatives used in these MOF syntheses are given in Table A1 in Appendix B.4. The conventional syntheses of these materials were also reported to allow comparison with the scCO 2 method.

Synthesis of Zn-MOF-74
Zn-MOF-74 (i.e., Zn 2 (DOBDC), where DOBDC is 2,5-dioxido-1,4-benzenedicarboxylate) was synthesised at low temperature following the method reported by Yaghi et al. [64]. Typically, 0.24 g 2,5-dihydroxyterephthalic acid (H 4 DOBDC, 2.4 mmol) and 0.69 g Zn(OAc) 2 ·2H 2 O (6.24 mmol) were dissolved in 40 mL of DMF in a 200 mL glass vial, stirring magnetically for 30 min until a clear solution formed. This vial was sealed with parafilm before being placed in an oven for crystallisation at 40 • C for 24 h. The product was separated by filtration and repeatedly washed with methanol, before drying at room temperature to obtain 0.69 g MOF-74 solid (giving a yield of 34.0% based on Zn). The reaction scheme for formation of this MOF at low temperature is shown in Scheme 1.
Crystals 2019, 9, x FOR PEER REVIEW 4 of 29 linkers and pyridine derivatives used in these MOF syntheses are given in Table A1 in Appendix B1.
The conventional syntheses of these materials were also reported to allow comparison with the scCO2 method.

Synthesis of Zn-MOF-74
Zn-MOF-74 (i.e., Zn2(DOBDC), where DOBDC is 2,5-dioxido-1,4-benzenedicarboxylate) was synthesised at low temperature following the method reported by Yaghi et al. [64]. Typically, 0.24 g 2,5-dihydroxyterephthalic acid (H4DOBDC, 2.4 mmol) and 0.69 g Zn(OAc)2·2H2O (6.24 mmol) were dissolved in 40 mL of DMF in a 200 mL glass vial, stirring magnetically for 30 min until a clear solution formed. This vial was sealed with parafilm before being placed in an oven for crystallisation at 40 °Ϲ for 24 h. The product was separated by filtration and repeatedly washed with methanol, before drying at room temperature to obtain 0.69 g MOF-74 solid (giving a yield of 34.0% based on Zn). The reaction scheme for formation of this MOF at low temperature is shown in Scheme 1.

Synthesis of [Cu24(OH-mBDC)24]n Metal-Organic Polyhedra
The synthesis of [Cu24(OH-mBDC)24]n MOP was followed by the procedure reported for the synthesis of copper-based cuboctahedron metal-organic polyhedra by Lee et al. [65]. The synthesis of this MOF is illustrated in Scheme 2. Typically, 40 mL of 1.46 g OH-mBDC in MeOH was mixed with 120 mL of 1.60 g Cu(OAc)2•H2O in MeOH (MeOH was used as a solvent in this system) to form a stock solution. No precipitation occurred after 10 days. After that, 12.5 mL N,N'-dimethylacetamide and 7.5 mL of MeOH were added to 80 mL of the stock solution, stirring at 200 rpm at 40 °Ϲ for 3 days. The blue solid was separated by filtration, repeatedly washed with methanol, yielding 1.9 g of solid after drying at room temperature.  24 ] n MOP was followed by the procedure reported for the synthesis of copper-based cuboctahedron metal-organic polyhedra by Lee et al. [65]. The synthesis of this MOF is illustrated in Scheme 2. Typically, 40 mL of 1.46 g OH-mBDC in MeOH was mixed with 120 mL of 1.60 g Cu(OAc) 2 ·H 2 O in MeOH (MeOH was used as a solvent in this system) to form a stock solution. No precipitation occurred after 10 days. After that, 12.5 mL N,N'-dimethylacetamide and 7.5 mL of MeOH were added to 80 mL of the stock solution, stirring at 200 rpm at 40 • C for 3 days. The blue solid was separated by filtration, repeatedly washed with methanol, yielding 1.9 g of solid after drying at room temperature. this MOF is illustrated in Scheme 2. Typically, 40 mL of 1.46 g OH-mBDC in MeOH was mixed with 120 mL of 1.60 g Cu(OAc)2•H2O in MeOH (MeOH was used as a solvent in this system) to form a stock solution. No precipitation occurred after 10 days. After that, 12.5 mL N,N'-dimethylacetamide and 7.5 mL of MeOH were added to 80 mL of the stock solution, stirring at 200 rpm at 40 °Ϲ for 3 days. The blue solid was separated by filtration, repeatedly washed with methanol, yielding 1.9 g of solid after drying at room temperature.  24 ] n (copper clusters represented as blue polyhedra, C-C bonds are represented as dark grey sticks, C-O bonds are represented as orange sticks; all hydrogen atoms are omitted for clarity).

Synthesis of ZnAzopy-MOF
The reduced Schiff-base linker, 4-carboxy-phenylene-methylene-4-benzoic acid (named L 1 H 2 ), was synthesised following the method reported previously by Liu et al. [56]. In brief, 3.0 g 4-carboxybenzoic acid (20 mmol) and 2.7 g 4-aminobenzoic acid (20 mmol) were dissolved in MeOH and the formation of a brownish solid was observed, indicating the formation of the intermediate Schiff-base product.
After stirring for 2 h, NaBH 4 (2.00 g) was added to reduce the intermediate Schiff-base, resulting in the formation of the intermediate Schiff-base product. The colour faded, and the precipitate dissolved. The solvent was removed from the filtrate by rotary evaporation, and then 100 mL of distilled water was added. The mixture was adjusted to pH~5 with HCl (1.0 mol l −1 ), and an off-white solid of H 2 L 1 formed immediately. The solid was filtered off, washed with Et 2 O, dried, and recrystallized from MeOH/H 2 O to afford the diacid, which was used without further purification. The synthetic route for the preparation of this linker is given in Scheme 3. The reduced Schiff-base linker, 4-carboxy-phenylene-methylene-4-benzoic acid (named L 1 H2), was synthesised following the method reported previously by Liu et al. [56]. In brief, 3.0 g 4-carboxybenzoic acid (20 mmol) and 2.7 g 4-aminobenzoic acid (20 mmol) were dissolved in MeOH and the formation of a brownish solid was observed, indicating the formation of the intermediate Schiff-base product. After stirring for 2 h, NaBH4 (2.00 g) was added to reduce the intermediate Schiff-base, resulting in the formation of the intermediate Schiff-base product. The colour faded, and the precipitate dissolved. The solvent was removed from the filtrate by rotary evaporation, and then 100 mL of distilled water was added. The mixture was adjusted to pH ~ 5 with HCl (1.0 mol l −1 ), and an off-white solid of H2L 1 formed immediately. The solid was filtered off, washed with Et2O, dried, and recrystallized from MeOH/H2O to afford the diacid, which was used without further purification. The synthetic route for the preparation of this linker is given in Scheme 3. In the synthesis of the ZnAzopy-MOF, 0.1 g of prepared L 1 H2 (0.4 mmol), 0.04 g Zn(OH)2 (0.4 mmol) and 0.04 g 4,4′-azopyridine (0.2 mmol) were added to 20 mL of distilled water. The suspension was sonicated for 10 min., and then transferred to a 23 mL Teflon-lined steel reaction vessel, which was subsequently sealed and heated to 40 °Ϲ for 72 h. The product was centrifuged and washed with distilled water, repeating 3 times, before finally drying at room temperature to obtain 0.06 g of red ZnAzopy-MOF prisms (giving a yield of 37.9% based on Zn).

Synthesis of ZnDPE-MOF
ZnDPE-MOF was synthesised in a similar manner as the ZnAzopy-MOF, but using 1,2-di(4-pyridyl)ethylene (DPE, 0.2 mmol) as a pillaring linker instead of 4,4′-azopyridine. All of the In the synthesis of the ZnAzopy-MOF, 0.1 g of prepared L 1 H 2 (0.4 mmol), 0.04 g Zn(OH) 2 (0.4 mmol) and 0.04 g 4,4 -azopyridine (0.2 mmol) were added to 20 mL of distilled water. The suspension was sonicated for 10 min., and then transferred to a 23 mL Teflon-lined steel reaction vessel, which was subsequently sealed and heated to 40 • C for 72 h. The product was centrifuged and washed with distilled water, repeating 3 times, before finally drying at room temperature to obtain 0.06 g of red ZnAzopy-MOF prisms (giving a yield of 37.9% based on Zn).

Synthesis of ZnDPE-MOF
ZnDPE-MOF was synthesised in a similar manner as the ZnAzopy-MOF, but using 1,2-di(4-pyridyl)ethylene (DPE, 0.2 mmol) as a pillaring linker instead of 4,4 -azopyridine. All of the subsequent steps were maintained to obtain 0.14 g of yellow ZnDPE-MOF prisms (giving a yield of 40.6% based on Zn).

Synthesis of MOFs Using Supercritical CO 2
In the scCO 2 routes (Scheme 4), MOF precursor solutions (a mixture of metal salt and acid linker dissolved in a solvent) were produced with the same concentrations as the conventional methods above and were placed in a 200 mL glass vial, then sealed inside a 250 mL cylindrical steel reactor pressure vessel, magnetically stirred at 200 rpm and heated to 40 • C in an oven. The vessel was connected to a scCO 2 rig equipped with a flow-controllable liquid pump. The vessel was pressurised to 75 bar at a flowrate of 5 g min −1 , keeping the reaction time the same as for the conventional synthesis, i.e., without scCO 2 . After reaction, the reactor was depressurised slowly to atmospheric pressure. The resulting solution was centrifuged (10,000 rpm for 10 min), washed with methanol 3 times, and dried in air at room temperature to obtain a solid product.
As a comparative control experiment in the absence of scCO 2 , a mixture of 0.1 g L 1 H 2 , 0.04 g Zn(OH) 2 , 0.04 g 1,2-di(4-pyridyl)ethylene and 20 mL distilled water was pressurised with N 2 at 75 bar and 40 • C for 72 h to confirm the effect of the different the gases under supercritical conditions.

Results and Discussions
The first part of this study focuses on some well-known MOFs, namely: Zn-MOF-74, UiO-66, Zn-BTC, as well as the [Cu24(OH-mBDC)24]n MOP (see Table 1). The second part of this study focuses on new MOF systems (ZnAzopy-MOF and ZnDPE-MOF) using flexible reduced Schiff-base linkers and pyridine derivatives as ancillary ligands.
Synthetic procedures of the well-known MOFs mentioned above with and without scCO2 are given either in the Materials and Methods section or in the Appendix B. In the conventional method in the absence of scCO2, Zn-MOF-74 was synthesised by using DMF and taken to the crystallisation step without the use of antisolvents. Using scCO2 in these MOF syntheses was expected to form the same meso-or macroporous structures as seen in HKUST-1 and Zn-BTC reported by Peng et al. [38,39]. [Cu24(OH-mBDC)24]n metal-organic polyhedron with nanoscale polyhedral structure was synthesised in MeOH in which a precursor solution of OH-mBDC and Cu(OAc)2 was formed. The results of these syntheses are further discussed below.

Results and Discussions
The first part of this study focuses on some well-known MOFs, namely: Zn-MOF-74, UiO-66, Zn-BTC, as well as the [Cu 24 (OH-mBDC) 24 ] n MOP (see Table 1). The second part of this study focuses on new MOF systems (ZnAzopy-MOF and ZnDPE-MOF) using flexible reduced Schiff-base linkers and pyridine derivatives as ancillary ligands.
Synthetic procedures of the well-known MOFs mentioned above with and without scCO 2 are given either in the Materials and Methods section or in the Appendix B. In the conventional method in the absence of scCO 2 , Zn-MOF-74 was synthesised by using DMF and taken to the crystallisation step without the use of antisolvents. Using scCO 2 in these MOF syntheses was expected to form the same meso-or macroporous structures as seen in HKUST-1 and Zn-BTC reported by Peng et al. [38,39].
[Cu 24 (OH-mBDC) 24 ] n metal-organic polyhedron with nanoscale polyhedral structure was synthesised in MeOH in which a precursor solution of OH-mBDC and Cu(OAc) 2 was formed. The results of these syntheses are further discussed below.

MOFs Metal Nodes Linkers Comments on the Formation Using Conventional Methods
Zn-MOF-74 Zn given either in the Materials and Methods section or in the Appendix B. In the conventional method in the absence of scCO2, Zn-MOF-74 was synthesised by using DMF and taken to the crystallisation step without the use of antisolvents. Using scCO2 in these MOF syntheses was expected to form the same meso-or macroporous structures as seen in HKUST-1 and Zn-BTC reported by Peng et al. [38,39]. [Cu24(OH-mBDC)24]n metal-organic polyhedron with nanoscale polyhedral structure was synthesised in MeOH in which a precursor solution of OH-mBDC and Cu(OAc)2 was formed. The results of these syntheses are further discussed below. in the absence of scCO2, Zn-MOF-74 was synthesised by using DMF and taken to the crystallisation step without the use of antisolvents. Using scCO2 in these MOF syntheses was expected to form the same meso-or macroporous structures as seen in HKUST-1 and Zn-BTC reported by Peng et al. [38,39]. [Cu24(OH-mBDC)24]n metal-organic polyhedron with nanoscale polyhedral structure was synthesised in MeOH in which a precursor solution of OH-mBDC and Cu(OAc)2 was formed. The results of these syntheses are further discussed below. in the absence of scCO2, Zn-MOF-74 was synthesised by using DMF and taken to the crystallisation step without the use of antisolvents. Using scCO2 in these MOF syntheses was expected to form the same meso-or macroporous structures as seen in HKUST-1 and Zn-BTC reported by Peng et al. [38,39]. [Cu24(OH-mBDC)24]n metal-organic polyhedron with nanoscale polyhedral structure was synthesised in MeOH in which a precursor solution of OH-mBDC and Cu(OAc)2 was formed. The results of these syntheses are further discussed below. The use of DMSO in BTC to form hydrogen-bonded complexes as reported in the HKUST-1 synthesis [47] was extended to the Zn-BTC. However, this method did not seem to be successful for this MOF, with no precipitation occurring from the stock solution after adding up to ten volume equivalents of MeOH and introducing scCO2 into the system after 3 days. This might be due to the metal-ligand coordination forces between Zn and BTC not being able to overcome the H-bonding, even in the presence of CO2-expanded MeOH as an antisolvent at 40 °Ϲ. In the synthesis of UiO-66, 40 °Ϲ was chosen as a technically achievable temperature for the scCO2 method. However, UiO-66 synthesised at this temperature via either conventional or scCO2 synthesis, while still crystalline, showed much broader PXRD peaks than normal UiO-66, indicating small crystallite sizes which were not noticeably affected by the presence of scCO2. Syntheses of Zn-BTC (in DMSO) and UiO-66 are described and further discussed in the Appendix B.

Using scCO2 in Precursor Solution Type I
Zn-MOF-74 was synthesised via the methods given in the Materials and Methods section, enabling comparison of the MOFs synthesised using conventional solvothermal synthesis, without CO2 (here called "normal Zn-MOF-74"), to the MOF synthesised with scCO2 (here called "Zn-MOF-74 in scCO2"). As seen in Figure 2a, normal Zn-MOF-74 samples show a similar powder X-ray diffraction (PXRD) pattern to the simulated Zn-MOF-74 with all the main peaks at 7, 12, 22, 25 and 32 2 preserved, confirming that this MOF can be successfully synthesised at low temperature (40 C), in addition to the higher temperatures of 110 or 100 C as reported by others [66,67]. The PXRD pattern of Zn-MOF-74 in scCO2 is almost identical to the normal Zn-MOF-74 sample (see Figure 2b), showing that the scCO2 did not overly affect the MOF crystal structure during the synthesis. Yields in both syntheses were also comparable (34% for conventional synthesis and 36% in scCO2). SEM analysis of these samples revealed that the typical morphology of normal Zn-MOF-74 is octahedral with smooth faces, while Zn-MOF-74 in scCO2 shows a similar morphology but with very rough faces (Figure 2b), which might be due to an acidic etching effect of the CO2 at 75 bar and 40 °Ϲ for 72 h. Effect of the acid etching on surface roughness have been previously investigated on amorphous silica and ceramic [68,69]. In addition, etching of MOFs in aqueous acidic solutions such as a mixture of hydroquinone and MeOH (up to 180 °Ϲ for 72 h) or a mixture of phosphoric acid, MeOH and DMSO (at 40 °Ϲ for up to 10 days) has been shown to result in remarkable effects such as interconnected geometrical macropores and etched hole features [37,51,70]. The use of DMSO in BTC to form hydrogen-bonded complexes as reported in the HKUST-1 synthesis [47] was extended to the Zn-BTC. However, this method did not seem to be successful for this MOF, with no precipitation occurring from the stock solution after adding up to ten volume equivalents of MeOH and introducing scCO 2 into the system after 3 days. This might be due to the metal-ligand coordination forces between Zn and BTC not being able to overcome the H-bonding, even in the presence of CO 2 -expanded MeOH as an antisolvent at 40 • C. In the synthesis of UiO-66, 40 • C was chosen as a technically achievable temperature for the scCO 2 method. However, UiO-66 synthesised at this temperature via either conventional or scCO 2 synthesis, while still crystalline, showed much broader PXRD peaks than normal UiO-66, indicating small crystallite sizes which were not noticeably affected by the presence of scCO 2 . Syntheses of Zn-BTC (in DMSO) and UiO-66 are described and further discussed in the Appendix B.

Using scCO 2 in Precursor Solution Type I
Zn-MOF-74 was synthesised via the methods given in the Materials and Methods section, enabling comparison of the MOFs synthesised using conventional solvothermal synthesis, without CO 2 (here called "normal Zn-MOF-74"), to the MOF synthesised with scCO 2 (here called "Zn-MOF-74 in scCO 2 "). As seen in Figure 2a, normal Zn-MOF-74 samples show a similar powder X-ray diffraction (PXRD) pattern to the simulated Zn-MOF-74 with all the main peaks at 7 • , 12 • , 22 • , 25 • and 32 • 2θ preserved, confirming that this MOF can be successfully synthesised at low temperature (40 • C), in addition to the higher temperatures of 110 or 100 • C as reported by others [66,67]. The PXRD pattern of Zn-MOF-74 in scCO 2 is almost identical to the normal Zn-MOF-74 sample (see Figure 2b), showing that the scCO 2 did not overly affect the MOF crystal structure during the synthesis. Yields in both syntheses were also comparable (34% for conventional synthesis and 36% in scCO 2 ). SEM analysis of these samples revealed that the typical morphology of normal Zn-MOF-74 is octahedral with smooth faces, while Zn-MOF-74 in scCO 2 shows a similar morphology but with very rough faces (Figure 2b), which might be due to an acidic etching effect of the CO 2 at 75 bar and 40 • C for 72 h. Effect of the acid etching on surface roughness have been previously investigated on amorphous silica and ceramic [68,69]. In addition, etching of MOFs in aqueous acidic solutions such as a mixture of hydroquinone and MeOH (up to 180 • C for 72 h) or a mixture of phosphoric acid, MeOH and DMSO (at 40 • C for up to 10 days) has been shown to result in remarkable effects such as interconnected geometrical macropores and etched hole features [37,51,70]. Gas sorption experiments were carried out on these samples to investigate the etching effect on the micro/mesoporosity. The surface area observed in the normal Zn-MOF-74 in this study (BET surface area of normal Zn-MOF-74 is 201 m 2 g −1 , see Figure 3a) was lower than those reported in the literature, and the type III isotherm indicated a lower level of microporosity [72]. This may be due to incomplete solvent removal as solvent removal was achieved through scCO2 depressurisation rather than the lengthy high-temperature activation used in previous studies [72]. Further drying these samples under scCO2 flow at 40 °Ϲ, 120 bar and 12 h showed improved microporous area (see Appendix C4). It should be noted that the scCO2 route also provides additional features on the surface of the sample (presumably due to an etching effect), which had the effect of increasing the BET surface area of this MOF (BET surface area of Zn-MOF-74 in scCO2 is ~350 m 2 g −1 ) compared to the normal Zn-MOF-74 synthesised at low temperature (~200 m 2 g −1 , see Figure 3a).
The pore size distributions by the BJH method, however, do not show any significant differences between these two samples, meaning that there were no additional meso-and macropores formed in Zn-MOF-74 through exposure to scCO2, indicating that addition of scCO2 during this MOF synthesis did not help to enlarge the pores as was the case for HKUST-1 [39,47]. Note that the HKUST-1 precursor solution was stabilised by strong O-H O hydrogen bonds, Gas sorption experiments (Appendix A) were carried out on these samples to investigate the etching effect on the micro/mesoporosity. The surface area observed in the normal Zn-MOF-74 in this study (BET surface area of normal Zn-MOF-74 is 201 m 2 g −1 , see Figure 3a) was lower than those reported in the literature, and the type III isotherm indicated a lower level of microporosity [72]. This may be due to incomplete solvent removal as solvent removal was achieved through scCO 2 depressurisation rather than the lengthy high-temperature activation used in previous studies [72]. Further drying these samples under scCO 2 flow at 40 • C, 120 bar and 12 h showed improved microporous area (see Appendix C.4). It should be noted that the scCO 2 route also provides additional features on the surface of the sample (presumably due to an etching effect), which had the effect of increasing the BET surface area of this MOF (BET surface area of Zn-MOF-74 in scCO 2 is~350 m 2 g −1 ) compared to the normal Zn-MOF-74 synthesised at low temperature (~200 m 2 g −1 , see Figure 3a). The pore size distributions by the BJH method, however, do not show any significant differences between these two samples, meaning that there were no additional meso-and macropores formed in Zn-MOF-74 through exposure to scCO 2 , indicating that addition of scCO 2 during this MOF synthesis did not help to enlarge the pores as was the case for HKUST-1 [39,47]. Note that the HKUST-1 precursor solution was stabilised by strong O-H . . . O hydrogen bonds, forming a two-dimensional supramolecular network within each layer (see Appendix B.1). The donor groups are the hydroxyls of the trimesic acid molecules, while the acceptors are the carbonyl or the sulfoxide O atoms [73].

Using scCO2 in Precursor Solution Type II
Synthesis of the [Cu24(OH-mBDC)24]n MOP is based on self-assembly of Cu2(COO)4 paddlewheels serving as square secondary building units (SBUs) bonded at a 120° angle to OH-mBDC and N,N'-dimethylacetamide serving as the ligands [54]. It was shown that the metal nodes and linkers in this metal-organic polyhedron remained unreacted in MeOH for up to 10 days, resulting in a stable precursor solution before DMSO is introduced. Stable precursors of this sort are promising for investigation of the effect of scCO2 on this metal-organic polyhedron because the aggregation can happen at the same time the solvent is expanded. The crystallinity of [Cu24(OH-mBDC)24]n MOP synthesised via the conventional method was confirmed by the similarity between the PXRD pattern of this sample to the simulated pattern reported by Li et al. [74]. In comparing the two [Cu24(OH-mBDC)24]n MOP samples synthesised from the same starting materials and at the same temperature but with and without scCO2 pressure (i.e., at 0 and 75 bar), some distortion occurred in the sample treated with scCO2, as evidenced by differences in the intensities and shapes of the PXRD peaks at 6, 7, 11, 14 and 17 2 (see Figure 4a). These two samples were analysed under inelastic neutron scattering (INS) spectroscopy (see Appendix C2), showing that all bond vibrations within these two structures are identical. In the SEM results (Figure 4b), it can be seen that they have identical morphologies. However, the crystallite size of the samples produced in scCO2 is remarkably larger than the conventionally synthesised sample (~2 µ m compared with ~1 µ m), indicating the scCO2 would lower the deprotonation rate of OH-mBDC, thus decreasing the growth rates of this MOP [75]. Dissolving 1.6 g Cu(OAc)2•H2O and 1.5 g OH-mBDC in 100 mL DMSO resulted in precipitation after 2 h (see Appendix B1).  24 ] n MOP is based on self-assembly of Cu 2 (COO) 4 paddlewheels serving as square secondary building units (SBUs) bonded at a 120 • angle to OH-mBDC and N,N'-dimethylacetamide serving as the ligands [54]. It was shown that the metal nodes and linkers in this metal-organic polyhedron remained unreacted in MeOH for up to 10 days, resulting in a stable precursor solution before DMSO is introduced. Stable precursors of this sort are promising for investigation of the effect of scCO 2 on this metal-organic polyhedron because the aggregation can happen at the same time the solvent is expanded. The crystallinity of [Cu 24 (OH-mBDC) 24 ] n MOP synthesised via the conventional method was confirmed by the similarity between the PXRD pattern of this sample to the simulated pattern reported by Li et al. [74]. In comparing the two [Cu 24 (OH-mBDC) 24 ] n MOP samples synthesised from the same starting materials and at the same temperature but with and without scCO 2 pressure (i.e., at 0 and 75 bar), some distortion occurred in the sample treated with scCO 2 , as evidenced by differences in the intensities and shapes of the PXRD peaks at 6 • , 7 • , 11 • , 14 • and 17 • 2θ (see Figure 4a). These two samples were analysed under inelastic neutron scattering (INS) spectroscopy (see Appendix C.2), showing that all bond vibrations within these two structures are identical. In the SEM results (Figure 4b), it can be seen that they have identical morphologies. However, the crystallite size of the samples produced in scCO 2 is remarkably larger than the conventionally synthesised sample (~2 µm compared with~1 µm), indicating the scCO 2 would lower the deprotonation rate of OH-mBDC, thus decreasing the growth rates of this MOP [75]. Dissolving 1.6 g Cu(OAc) 2 ·H 2 O and 1.5 g OH-mBDC in 100 mL DMSO resulted in precipitation after 2 h (see Appendix B.1).  Table 2.

Using scCO2 in the Presence of Pyridine Derivatives as Ancillary Ligands to Synthesise Pillared MOFs
Synthesis of MOF systems using flexible reduced Schiff base linkers and pyridine derivatives as ancillary ligands, including ZnAzopy-MOF and ZnDPE-MOF were also performed with and without scCO2, to determine the effect of these differing conditions on the nature of the final product.  Table 2.

Using scCO 2 in the Presence of Pyridine Derivatives as Ancillary Ligands to Synthesise Pillared MOFs
Synthesis of MOF systems using flexible reduced Schiff base linkers and pyridine derivatives as ancillary ligands, including ZnAzopy-MOF and ZnDPE-MOF were also performed with and without scCO 2 , to determine the effect of these differing conditions on the nature of the final product.

In ZnAzopy-MOF Synthesis
ZnAzopy-MOF is conventionally synthesised by hydrothermal reaction between the flexible reduced Schiff base linker 4-carboxy-phenylene-methyleneamino-4-benzoic acid (L 1 H 2 ) and zinc hydroxide and the rigid pillaring linker 4,4 -azopy. The inclusion of an element of rigidity provides a backbone around which the flexible linkers can bind, thereby allowing the assembly of 3D structures bearing novel network topologies. The molecular structure has been determined by single crystal X-ray diffraction (see Figure 5). The asymmetric unit comprises one Zn ion in a general position {Zn (1) (2) is 6-coordinate, distorted octahedral, binding to six carboxylate oxygens. Three Zn ions form a double lantern arrangement with carboxylates bridging pairs of Zn ions and pyridyl groups capping both ends. The coordination environment around the metal centres in this structure is reminiscent of the paddlewheel SBU, which comprises three zinc centres forming an hour-glass shape by the coordination of carboxylate moieties from six L 1 units in the di-mono-dentate bridging mode in the equatorial positions, with 4,4′-azopy units coordinated to the axial positions of the hour glass, (Figure 5a). The L 1 units extend in six directions in the a/b plane, adopting a bent conformation through the rotation of the amine and methylene bridges, whilst the 4,4′-azopy ligands extend along the c axis, linking hour-glass SBUs in to an extended 3D framework, bearing roughly cylindrical 1D channels of 4.75 Å diameter observed parallel to the c axis (Figure 5b). Elemental analysis of this MOF is given in Appendix C3, together with alternative views of the structure. This MOF was synthesised without scCO2 ("normal ZnAzopy-MOF") and with scCO2 ("ZnAzopy-MOF in scCO2") using the same concentration of starting materials, reaction temperature and reaction time. It can be seen that the experiment using scCO2 formed a slightly greater yield than without scCO2 at the same temperature and for the same reaction time (46.7% compared with 37.9%). This might be due to the increased solubility of the metal salt in the mixture of L 1 H2 linker, 4,4′-azopyridine and scCO2. The crystallinity of both samples was examined using PXRD, as shown in Figure 6, with PXRD patterns showing the ZnAzopy-MOF in scCO2 retained the main features (including peaks at 12, 15, 18, 20, 22, 23, 25 and 28 2) when compared to the This MOF was synthesised without scCO 2 ("normal ZnAzopy-MOF") and with scCO 2 ("ZnAzopy-MOF in scCO 2 ") using the same concentration of starting materials, reaction temperature and reaction time. It can be seen that the experiment using scCO 2 formed a slightly greater yield than without scCO 2 at the same temperature and for the same reaction time (46.7% compared with 37.9%). This might be due to the increased solubility of the metal salt in the mixture of L 1 H 2 linker, 4,4 -azopyridine and scCO 2 . The crystallinity of both samples was examined using PXRD, as shown in Figure 6, with PXRD patterns showing the ZnAzopy-MOF in scCO 2 retained the main features (including peaks at 12 • , 15 • , 18 • , 20 • , 22 • , 23 • , 25 • and 28 • 2θ) when compared to the PXRD patterns of the conventional sample. However, there were some key differences in the pattern of the MOF synthesised in scCO 2 , including the disappearance of low-angle peaks at 7 • , 8 • and 13 • 2θ, the appearance of a new peak at 17 • 2θ and a shift in the peak at 33 • 2θ, showing that the addition of scCO 2 during the synthesis had a small but noticeable effect on the crystal structure. This might be due to the change in dilution of 4,4 -azopyridine, which was highly soluble in scCO 2 [76]. Note that this ancillary ligand is crucial for formation of this MOF. In general, large crystals were precipitated from both syntheses, as can be seen in Figure 7. The increase in the z-dimensional diameter of ZnAzopy-MOF in scCO 2 can be referred back to the higher solubility of the reagents in the media with 4,4 -azopyridine used as a co-solvent and a pillaring linker with respect to scCO 2 , thus supporting crystal growth and nucleation. Gas sorption results (Appendix C.4), however, showed that these samples are non-porous with surface areas less than 20 m 2 g −1 .
Crystals 2019, 9, x FOR PEER REVIEW 12 of 29 of the MOF synthesised in scCO2, including the disappearance of low-angle peaks at 7, 8 and 13 2, the appearance of a new peak at 17 2 and a shift in the peak at 33 2, showing that the addition of scCO2 during the synthesis had a small but noticeable effect on the crystal structure. This might be due to the change in dilution of 4,4′-azopyridine, which was highly soluble in scCO2 [76]. Note that this ancillary ligand is crucial for formation of this MOF. In general, large crystals were precipitated from both syntheses, as can be seen in Figure 7. The increase in the z-dimensional diameter of ZnAzopy-MOF in scCO2 can be referred back to the higher solubility of the reagents in the media with 4,4′-azopyridine used as a co-solvent and a pillaring linker with respect to scCO2, thus supporting crystal growth and nucleation. Gas sorption results (Appendix C4), however, showed that these samples are non-porous with surface areas less than 20 m 2 g −1 .   of the MOF synthesised in scCO2, including the disappearance of low-angle peaks at 7, 8 and 13 2, the appearance of a new peak at 17 2 and a shift in the peak at 33 2, showing that the addition of scCO2 during the synthesis had a small but noticeable effect on the crystal structure. This might be due to the change in dilution of 4,4′-azopyridine, which was highly soluble in scCO2 [76]. Note that this ancillary ligand is crucial for formation of this MOF. In general, large crystals were precipitated from both syntheses, as can be seen in Figure 7. The increase in the z-dimensional diameter of ZnAzopy-MOF in scCO2 can be referred back to the higher solubility of the reagents in the media with 4,4′-azopyridine used as a co-solvent and a pillaring linker with respect to scCO2, thus supporting crystal growth and nucleation. Gas sorption results (Appendix C4), however, showed that these samples are non-porous with surface areas less than 20 m 2 g −1 .

In ZnDPE-MOF Synthesis
ZnDPE-MOF is conventionally formed by the hydrothermal reaction of equimolar quantities of Zn(OH) 2 and L 1 H 2 with half an equivalent of DPE, to produce the 3D structure{[Zn 2 (L 1 )(DPE)]·4H 2 O} n . The coordination environment around the metal centres in this structure, as determined by single crystal X-ray diffraction, is distorted octahedral with two chelating carboxylates and two monodentate DPE pyridinyl nitrogens in a cis conformation (see Figure 8a). Coordination of the pyridinyl nitrogens of the neutral DPE linkers in the cis conformation gives rise to zig-zag chains and considering also the L 1 connections an overall 3D network (see Figure 8b). Elemental analysis of this MOF is given in the Appendix C.4.

In ZnDPE-MOF Synthesis
ZnDPE-MOF is conventionally formed by the hydrothermal reaction of equimolar quantities of Zn(OH)2 and L 1 H2 with half an equivalent of DPE, to produce the 3D structure{[Zn2(L 1 )(DPE)]·4H2O}n. The coordination environment around the metal centres in this structure, as determined by single crystal X-ray diffraction, is distorted octahedral with two chelating carboxylates and two monodentate DPE pyridinyl nitrogens in a cis conformation (see Figure 8a). Coordination of the pyridinyl nitrogens of the neutral DPE linkers in the cis conformation gives rise to zig-zag chains and considering also the L 1 connections an overall 3D network (see Figure 8b). Elemental analysis of this MOF is given in the Appendix C4. Different gases (CO2 with Tc 31.1 °Ϲ and Pc 73.9 bar; and N2 with Tc −147 °Ϲ and Pc 34 bar) were introduced to the precursor solution to synthesise this MOF at 40 °Ϲ and 75 bar, forming supercritical systems (scCO2 and scN2) with increased solubility and decreased polarity. It can be seen that the yield of this MOF in scCO2 is considerably higher than in scN2 (an increase of 29.8 wt%). The crystallinity of samples was examined using PXRD (see Figure A6 in Appendix C1 and Figure 9). Most of the main peaks at 8°, 20°, 26°, 27° and 29 2 in the simulated ZnDPE-MOF were present in both samples ( Figure A6), showing that the crystalline structure had formed and was preserved under high-pressure syntheses. There was a slight difference between the samples synthesised in scCO2 and in scN2. In the sample synthesised in scCO2, there is an extra peak at 12 2, and the peak appearing at 14 2 is more clearly a doublet in comparison to the sample in scN2. The transformation in peak shape was extended to those at 15°, 17° and 24 2. These differences show the evidence of the effect of CO2 on the MOF crystal structure. At the same time, the difference in the nature of the supercritical fluids introduced had an effect on the morphology of this MOF during the crystallisation (Figure 10). In general, both samples appear to not have uniform shape, with the size varying between 3 and 15 µm. However, in more detail, the sample synthesised in scCO2 is Different gases (CO 2 with T c 31.1 • C and P c 73.9 bar; and N 2 with T c −147 • C and P c 34 bar) were introduced to the precursor solution to synthesise this MOF at 40 • C and 75 bar, forming supercritical systems (scCO 2 and scN 2 ) with increased solubility and decreased polarity. It can be seen that the yield of this MOF in scCO 2 is considerably higher than in scN 2 (an increase of 29.8 wt%). The crystallinity of samples was examined using PXRD (see Figure A6 in Appendix C.1 and Figure 9). Most of the main peaks at 8 • , 20 • , 26 • , 27 • and 29 • 2θ in the simulated ZnDPE-MOF were present in both samples ( Figure A6), showing that the crystalline structure had formed and was preserved under high-pressure syntheses. There was a slight difference between the samples synthesised in scCO 2 and in scN 2 . In the sample synthesised in scCO 2 , there is an extra peak at 12 • 2θ, and the peak appearing at 14 • 2θ is more clearly a doublet in comparison to the sample in scN 2 . The transformation in peak shape was extended to those at 15 • , 17 • and 24 • 2θ. These differences show the evidence of the effect of CO 2 on the MOF crystal structure. At the same time, the difference in the nature of the supercritical fluids introduced had an effect on the morphology of this MOF during the crystallisation (Figure 10). In general, both samples appear to not have uniform shape, with the size varying between 3 and 15 µm. However, in more detail, the sample synthesised in scCO 2 is represented by bulky crystal agglomerates compared with the smooth surfaces observed for the crystals of the MOF synthesised in N 2 . This may be due to the etching effect of CO 2 , which was shown to result in rougher, pitted crystal surfaces in the Zn-MOF-74 synthesis. represented by bulky crystal agglomerates compared with the smooth surfaces observed for the crystals of the MOF synthesised in N2. This may be due to the etching effect of CO2, which was shown to result in rougher, pitted crystal surfaces in the Zn-MOF-74 synthesis.

Conclusions
In this study, scCO2 was introduced to different precursor complexes to synthesise various MOF/MOP systems. There was substantial evidence that scCO2 had an effect on surface texturing and crystal growth in these MOF syntheses. Zn-MOF-74 was formed in scCO2 and precursor solution type I, showing some changes in surface texture and surface area. [Cu24(OH-mBDC)24]n MOP with increased crystal size was also formed in scCO2 and precursor solution type II, indicating the growth rates of this MOF increased in the CO2-expanded solvent system. However, there was no evidence showing that additional porosity appeared in these MOFs in scCO2. Zn-BTC (in DMSO) and UiO-66 were formed but contained smaller crystallites in the scCO2 case. The introduction of scCO2 to the synthesis of two new MOF structures {[Zn2(L 1 )(DPE)]•4H2O}n and {[Zn3(L 1 )3(4,4′-azopy)]•7.5H2O}n using pyridine derivatives as ancillary ligands additionally showed that changes in crystallinity and morphology could result. This was shown (through analogous experiments with N2 under the same conditions) to be an effect of the presence of scCO2, rather than simply a direct result of high-pressure synthesis. However, again this supercritical fluid did not represented by bulky crystal agglomerates compared with the smooth surfaces observed for the crystals of the MOF synthesised in N2. This may be due to the etching effect of CO2, which was shown to result in rougher, pitted crystal surfaces in the Zn-MOF-74 synthesis.

Conclusions
In this study, scCO2 was introduced to different precursor complexes to synthesise various MOF/MOP systems. There was substantial evidence that scCO2 had an effect on surface texturing and crystal growth in these MOF syntheses. Zn-MOF-74 was formed in scCO2 and precursor solution type I, showing some changes in surface texture and surface area. [Cu24(OH-mBDC)24]n MOP with increased crystal size was also formed in scCO2 and precursor solution type II, indicating the growth rates of this MOF increased in the CO2-expanded solvent system. However, there was no evidence showing that additional porosity appeared in these MOFs in scCO2. Zn-BTC (in DMSO) and UiO-66 were formed but contained smaller crystallites in the scCO2 case. The introduction of scCO2 to the synthesis of two new MOF structures {[Zn2(L 1 )(DPE)]•4H2O}n and {[Zn3(L 1 )3(4,4′-azopy)]•7.5H2O}n using pyridine derivatives as ancillary ligands additionally showed that changes in crystallinity and morphology could result. This was shown (through analogous experiments with N2 under the same conditions) to be an effect of the presence of scCO2, rather than simply a direct result of high-pressure synthesis. However, again this supercritical fluid did not

Conclusions
In this study, scCO 2 was introduced to different precursor complexes to synthesise various MOF/MOP systems. There was substantial evidence that scCO 2 had an effect on surface texturing and crystal growth in these MOF syntheses. Zn-MOF-74 was formed in scCO 2 and precursor solution type I, showing some changes in surface texture and surface area. [Cu 24 (OH-mBDC) 24 ] n MOP with increased crystal size was also formed in scCO 2 and precursor solution type II, indicating the growth rates of this MOF increased in the CO 2 -expanded solvent system. However, there was no evidence showing that additional porosity appeared in these MOFs in scCO 2 . Zn-BTC (in DMSO) and UiO-66 were formed but contained smaller crystallites in the scCO 2 case. The introduction of scCO 2 to the synthesis of two new MOF structures {[Zn 2 (L 1 )(DPE)]·4H 2 O} n and {[Zn 3 (L 1 ) 3 (4,4 -azopy)]·7.5H 2 O} n using pyridine derivatives as ancillary ligands additionally showed that changes in crystallinity and morphology could result. This was shown (through analogous experiments with N 2 under the same conditions) to be an effect of the presence of scCO 2 , rather than simply a direct result of high-pressure synthesis. However, again this supercritical fluid did not result in any further meso-or macroporosity in the resulting samples. While scCO 2 indeed showed some positive effects on the synthesis of Zn-BTC and HKUST-1 in introducing additional porosity, the mechanism of these effects needs to be further studied to extend the method to other MOF systems.
These findings provide new information on the effects of introducing scCO 2 into the synthesis of a broad range of different MOFs, and provides information to direct further application of scCO 2 as an approach for morphological control.

Acknowledgments:
The X-ray Crystallographic Service at Southampton is thanked for data collection for ZnAzopy-MOF and ZnDPE-MOF. We thank the British Council Newton Fund for funding for a materials workshop at Northwest University in Xi'an. The INS spectra for [Cu 24 (OH-mBDC) 24 ] n MOP herein, assisted by Svemir Rudic, were collected on the TOSCA instrument at ISIS Neutron and Muon Source, Didcot, UK.

Conflicts of Interest:
The authors declare no conflict of interest.

Appendix A. Characterisation Methods
Materials after synthesis were tested by powder X-ray diffraction (PXRD) on a BRUKER AXS D8-Advance instrument with Vantec-1 detector using Cu Kα (λ = 1.5418 Å) as the source of X-ray radiation, in flat plate geometry, spinner speed 15 rpm, at 21 • C. The 2θ range, between 2-60 • was used with a 0.02 • interval for collection and 20 min scans.
Scanning electron microscopy (SEM) images were taken using JEOL IT300 SEM with a SED detector at School of Chemistry, University of Bristol with the following settings, objective apeture: 1, acceleration voltage: 5 kV, probe current: 10.3 m, working distance: 10.8 mm and magnifications from 350 to 40,000 times. The samples were coated with 15 nm silver before conducting the experiment.
Gas sorption isotherms were determined using nitrogen sorption at 77 K with a Micromeritics 3-Flex volumetric gas sorption analysis system (nitrogen with purity of 99.9999% was purchased from Air Products). Samples were degassed at 120 • C under dynamic high vacuum (10 −6 mbar) over 6 h prior to analysis. The total pore volume was taken at the end of the filling of the pore. Surface area was determined by the BET method according to British Standards, with relative pressure (P/P o ) selected considering the Rouquerol consistency criterion between values of 0 and 0.3.
Single crystal diffraction data for ZnAzopy-MOF and ZnDPE-MOF were collected on a Rigaku Saturn 724+ CCD diffractometer using a rotating anode X-ray source and 10 cm confocal mirrors monochromator at 100 K. Data were corrected for absorption and Lp effects. Structures were solved by direct methods and refined by full-matrix least squares on F 2 [77]. Further details are given in Table 1. H atoms were included in a constrained riding model except for those on water where coordinates were refined. In ZnAzopy-MOF the badly disordered solvent water in the void spaces was modelled as diffuse electron density by the Platon Squeeze procedure which recovered 2 × 73 and 2 × 78 electrons over four voids in the unit cell [78]. This equates to approx. 30 water molecules per unit cell or approx. 7.5 per void or Zn 3 unit. The CH 2 NH linkers C(13)/N(3) and C(28)/N(4) in the dicarboxylate ligands could not be clearly distinguished and were each modelled as 50/50 C/N. For ZnAzopy-MOF, where disorder was modelled, restraints on both geometry and displacement parameters were applied. CCDC 935555 and 1919724 contain the supplementary crystallographic data for this paper. These data can be obtained free of charge from The Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/data_request/cif. 24 ] n MOP with inelastic neutron scattering (INS) spectroscopy was achieved on the TOSCA indirect geometry spectrometer at ISIS Neutron and Muon Source, Didcot, UK [79]. The dehydrated sample was prepared within a glove box under an inert argon atmosphere. The sample was first loaded into aluminium foil sachets, with the loaded sample mass recorded. The sachets were then sealed between two aluminium plates, using indium wire as a seal to keep the sample under an inert atmosphere. INS spectra were recorded within the energy transfer range −80-8050 cm −1 , at a temperature of 10 K for 5-7 h. Subsequent data analysis, visualisation and normalisation was achieved using the Mantid software [80]. UK [79]. The dehydrated sample was prepared within a glove box under an inert argon atmosphere. The sample was first loaded into aluminium foil sachets, with the loaded sample mass recorded. The sachets were then sealed between two aluminium plates, using indium wire as a seal to keep the sample under an inert atmosphere. INS spectra were recorded within the energy transfer range −80-8050 cm −1 , at a temperature of 10 K for 5-7 h. Subsequent data analysis, visualisation and normalisation was achieved using the Mantid software [80].

Appendix B.2. Synthesis of Zn-BTC in DMSO and MeOH
The coordination of Zn(II) metal nodes and benzene-1,3,5-tricarboxylic acid (BTC) linkers, resulting in a 3D open network as found in Zn-BTC MOF. This MOF was successfully synthesised by solvothermal reactions between BTC and Zn(NO3)2•6H2O in a mixture of DMF and DMAC without any additives [81]. In fact, Peng et al. used this MOF to test the effect of CO2 in the presence of N-EtFOSA/TMGT solution [38]. By changing the pressure of the formed ionic liquid system from 10 UK [79]. The dehydrated sample was prepared within a glove box under an inert argon atmosphere. The sample was first loaded into aluminium foil sachets, with the loaded sample mass recorded. The sachets were then sealed between two aluminium plates, using indium wire as a seal to keep the sample under an inert atmosphere. INS spectra were recorded within the energy transfer range −80-8050 cm −1 , at a temperature of 10 K for 5-7 h. Subsequent data analysis, visualisation and normalisation was achieved using the Mantid software [80].

Appendix B.2. Synthesis of Zn-BTC in DMSO and MeOH
The coordination of Zn(II) metal nodes and benzene-1,3,5-tricarboxylic acid (BTC) linkers, resulting in a 3D open network as found in Zn-BTC MOF. This MOF was successfully synthesised by solvothermal reactions between BTC and Zn(NO3)2•6H2O in a mixture of DMF and DMAC without any additives [81]. In fact, Peng et al. used this MOF to test the effect of CO2 in the presence of N-EtFOSA/TMGT solution [38]. By changing the pressure of the formed ionic liquid system from 10

Appendix B.2. Synthesis of Zn-BTC in DMSO and MeOH
The coordination of Zn(II) metal nodes and benzene-1,3,5-tricarboxylic acid (BTC) linkers, resulting in a 3D open network as found in Zn-BTC MOF. This MOF was successfully synthesised by solvothermal reactions between BTC and Zn(NO 3 ) 2 ·6H 2 O in a mixture of DMF and DMAC without any additives [81]. In fact, Peng et al. used this MOF to test the effect of CO 2 in the presence of N-EtFOSA/TMGT solution [38]. By changing the pressure of the formed ionic liquid system from 10 to 63 bar, tetrahedron-like Zn-BTC particles with some evidence of mesopores were observed. In some cases where a high number of bulky molecules get involved, macropores were more favoured to minimise diffusion barrier, hence were managed to achieve in this study. With the success of the creation of macroporous HKUST-1 (or Cu-BTC) in scCO 2 [47], Zn-BTC synthesis was repeated in the same manner. In a typical experiment, 3.1 mg Zn(NO 3 ) 2 ·6H 2 O (10 mmol) and 1.5 mg BTC (5 mmol) were dissolved in 100 mL DMSO, resulting a clear precursor solution. As demonstrated previously [82], H-bonding between DMSO and BTC dominated the metal-ligand coordination forces and hindered the aggregation process, resulting in a stable stock solution. This solution (1-5 mL) was added in MeOH at varied volume ratio between 1:1 and 1:10, and left for up to 10 days at 40 • C. However, no precipitation was observed, consistent with the precursor solution of HKUST-1 [47]. A mixture of 4 mL stock solution and 40 mL MeOH was pressurised with CO 2 at 40 • C and 75 bar; however, no solid was formed after 3 days. This might be due to the metal-ligand coordination forces between Zn and BTC not being able to overcome the H-bonding even in the presence of CO 2 -expanded MeOH as an antisolvent at 40 • C. to 63 bar, tetrahedron-like Zn-BTC particles with some evidence of mesopores were observed. In some cases where a high number of bulky molecules get involved, macropores were more favoured to minimise diffusion barrier, hence were managed to achieve in this study. With the success of the creation of macroporous HKUST-1 (or Cu-BTC) in scCO2 [47], Zn-BTC synthesis was repeated in the same manner. In a typical experiment, 3.1 mg Zn(NO3)2•6H2O (10 mmol) and 1.5 mg BTC (5 mmol) were dissolved in 100 mL DMSO, resulting a clear precursor solution. As demonstrated previously [82], H-bonding between DMSO and BTC dominated the metal-ligand coordination forces and hindered the aggregation process, resulting in a stable stock solution. This solution (1-5 mL) was added in MeOH at varied volume ratio between 1:1 and 1:10, and left for up to 10 days at 40 °Ϲ. However, no precipitation was observed, consistent with the precursor solution of HKUST-1 [47]. A mixture of 4 mL stock solution and 40 mL MeOH was pressurised with CO2 at 40 °Ϲ and 75 bar; however, no solid was formed after 3 days. This might be due to the metal-ligand coordination forces between Zn and BTC not being able to overcome the H-bonding even in the presence of CO2-expanded MeOH as an antisolvent at 40 °Ϲ.

Appendix B.3. Synthesis of UiO-66 in DMF at Low Temperature
Normal UiO-66 was synthesised by dissolving 3.18 g ZrCl4 (0.014 mol) and 2.04 g benzene-1,4-dicarboxylic acid (H2BDC) (0.014 mol) in 80 mL anhydrous dimethylformamide (DMF) at room temperature. The reaction of this step is given in Scheme A1. The reaction mixture was heated in an autoclave at 120 °Ϲ for 24 h. After cooling in air to room temperature the white solid was separated by filtration, repeatedly washed with DMF and dried at room temperature. In the modified synthetic procedure, the autoclave was heated at 40 °Ϲ instead of 120 °Ϲ, with all other steps remaining the same. In the synthesis with scCO2, the stock solution was added in a glass vial and placed in a 250 mL cylindrical steel reactor pressure vessel and heated to 40 °Ϲ in an oven. The vessel was connected to a scCO2 rig equipped with a flow controllable liquid pump. The vessel was pressurised to 75 bar at a flowrate of 5 g min −1 for 24 h. After that, the reactor was depressurised slowly to atmospheric pressure. The white solid was separated by filtration, repeatedly washed with DMF and dried at room temperature.  Normal UiO-66 was synthesised by dissolving 3.18 g ZrCl 4 (0.014 mol) and 2.04 g benzene-1,4-dicarboxylic acid (H 2 BDC) (0.014 mol) in 80 mL anhydrous dimethylformamide (DMF) at room temperature. The reaction of this step is given in Scheme A1. The reaction mixture was heated in an autoclave at 120 • C for 24 h. After cooling in air to room temperature the white solid was separated by filtration, repeatedly washed with DMF and dried at room temperature. In the modified synthetic procedure, the autoclave was heated at 40 • C instead of 120 • C, with all other steps remaining the same. In the synthesis with scCO 2 , the stock solution was added in a glass vial and placed in a 250 mL cylindrical steel reactor pressure vessel and heated to 40 • C in an oven. The vessel was connected to a scCO 2 rig equipped with a flow controllable liquid pump. The vessel was pressurised to 75 bar at a flowrate of 5 g min −1 for 24 h. After that, the reactor was depressurised slowly to atmospheric pressure. The white solid was separated by filtration, repeatedly washed with DMF and dried at room temperature. modified synthetic procedure, the autoclave was heated at 40 °Ϲ instead of 120 °Ϲ, with all other steps remaining the same. In the synthesis with scCO2, the stock solution was added in a glass vial and placed in a 250 mL cylindrical steel reactor pressure vessel and heated to 40 °Ϲ in an oven. The vessel was connected to a scCO2 rig equipped with a flow controllable liquid pump. The vessel was pressurised to 75 bar at a flowrate of 5 g min −1 for 24 h. After that, the reactor was depressurised slowly to atmospheric pressure. The white solid was separated by filtration, repeatedly washed with DMF and dried at room temperature. The sample synthesised at 40 • C and dried at room temperature appeared to be gel-like light yellow crystals which are very different from the solids in the sample synthesised at 120 • C. Tian et al. also reported that a gel like UiO-66 monolith could be formed by a sol-gel process and mild drying conditions [83]. In the PXRD results, it can be seen that the low temperature synthesised sample has preserved peaks at 7 • , 12 • , 22 • , 26 • , 31 • , 43 • and 50 • 2θ (see Figure A4), confirming the UiO-66 structure in this sample. However, these peaks appeared much broader than the peaks in normal UiO-66, showing that the sample synthesised at 40 • C has smaller crystallite size than those synthesised at 120 • C. It has been demonstrated that UiO-66 obtained by room-temperature synthesis had defect sites with the maximal number achieved (~1.3 missing linker per SBU) at a temperature of 45 • C [84]. In this study, highly-crystalline MOFs were the principal focus. The scCO 2 experiment was applied in this MOF synthesis, however, the PXRD patterns of both samples synthesised at 40 • C are almost identical, showing that scCO 2 did not enhance the crystallite size in this MOF. The sample synthesised at 40 °Ϲ and dried at room temperature appeared to be gel-like light yellow crystals which are very different from the solids in the sample synthesised at 120 °Ϲ. Tian et al. also reported that a gel like UiO-66 monolith could be formed by a sol-gel process and mild drying conditions [83]. In the PXRD results, it can be seen that the low temperature synthesised sample has preserved peaks at 7°, 12°, 22°, 26°, 31°, 43° and 50° 2 (see Figure A4), confirming the UiO-66 structure in this sample. However, these peaks appeared much broader than the peaks in normal UiO-66, showing that the sample synthesised at 40 °Ϲ has smaller crystallite size than those synthesised at 120 °Ϲ. It has been demonstrated that UiO-66 obtained by room-temperature synthesis had defect sites with the maximal number achieved (~1.3 missing linker per SBU) at a temperature of 45 °Ϲ [84]. In this study, highly-crystalline MOFs were the principal focus. The scCO2 experiment was applied in this MOF synthesis, however, the PXRD patterns of both samples synthesised at 40 °Ϲ are almost identical, showing that scCO2 did not enhance the crystallite size in this MOF.

Appendix C.3. Method and Elemental Analysis of ZnAzopy-MOF and ZnDPE-MOF
Ligand L 1 H2 (0.2 mmol, 0.05 g), Zn(OH)2 (0.2 mmol, 0.02 g) and 4,4′-azopy (0.1 mmol, 0.02 g) were added to 10 mL distilled water. The suspension was transferred to a 23 mL Teflon-lined steel reaction vessel which was subsequently sealed and heated to 110 °Ϲ for 72 h. The vessel was then cooled at a rate of 2 °Ϲ h -1 to room temperature, yielding 0.04 g red prisms (50.5% based on Zn). Elem  Ligand L 1 H 2 (0.2 mmol, 0.05 g), Zn(OH) 2 (0.2 mmol, 0.02 g), DPE (0.1 mmol, 0.018 g) and one drop of 6 M of NaOH solution were added to 10 mL distilled water. The suspension was then transferred to a 23 mL Teflon-lined steel reaction vessel which was subsequently sealed and heated to 110 • C for 72 h. The vessel was then cooled at a rate of 6 • C h −1 to room temperature, yielding 0.045 g yellow prisms (68.9% based on Zn). Elem. anal. calcd. for ZnDPE-MOF (C 27