Type 3 porous liquids based on non-ionic liquid phases – a broad and tailorable platform of selective, fluid gas sorbents

We describe a series of Type 3 porous liquids, denoted “T3PLs”, based on a wide range of microporous solids including MOFs, zeolites and a porous organic polymer (PAF-1). These solids are dispersed in various non-ionic liquid phases (including silicone oils, triglyceride oils, and polyethylene glycols) which have a range of structures and properties, and that are in many cases sterically excluded from the pores of the solids. Several stable dispersions with high gas uptakes are obtained. We show how these dispersions can be tailored toward important gas separation processes (CO2/CH4, C2H4/C2H6) and applications that require biocompatibility.


S.I. 1 Materials and measurements
The MOFs and PAF-1 were synthesized according to literature. Zeolites (including Zeolite Sigma -a zeolite branded by Sigma Aldrich, product no. 96096) were purchased from Sigma Aldrich in > 98% purity. All silicone oils, triglyceride oils, halogenated oils and polyethylene glycol derivatives were obtained from Sigma Aldrich UK in >98% purity or Alfa Aesar in >95% purity and were used as obtained without further purification.
Genosorb ® 1753 was a gift from Clariant International Ltd. PXRD measurements were carried out on a PANanalytical X'Pert Pro X-ray diffractometer. Copper was used as the X-ray source with a wavelength of 1.5405 Å. All experiments were carried out ex-situ using a spinning stage. Diffractograms were typically obtained from 5-50° with a step size of 0.0167°. Infrared spectra were obtained with aPerkin Elmer Spectrum 1. Thermogravimetric Analysis (TGA) were measured by the Analytical Service department of the School of Chemistry and Chemical Engineering (ASEP) using a Mettler Toledo DSC/TGA 1 Star instrument.
The solid was filtered and washed with distilled water (3× 30 mL). The remaining solid was filtered and dried in air at room temperature.

S.I. 4 Characterization of porous liquids
All porous liquids were characterized by Powder X-Ray Diffractometry (PXRD) and IR spectroscopy (Figures S1 and S2). All PXRD patterns show peaks corresponding to the crystalline solid component, indicating that the solid components remain intact and crystalline, superimposed on broad features due to the liquid components.   Figure S4: TGA and DSC curves for HKUST-1/Olive oil.

HKUST-1/Olive oil
Step    Dispersion stability is expected to be greater for smaller particles due to the greater surface area for interaction between the particles and liquid. Correspondingly, reducing the particle size of HKUST-1 particle size by brief ball milling from ca. 400 nm to ca. 50nm, the stability of HKUST-1/PDMS dispersions increased from ca. 1 day to ca. 3 days as shown by PXRD analysis of the upper layer ( Figure S7). Similarly, Al(fum)(OH) was synthesised at three different particle sizes by varying the solution concentration during synthesis ( Figure S8).

Al(fum)(OH)/paraffin oil
This provided Al(fum)(OH) particles with three different size ranges (as shown by PXRD and SEM, Figure S9). The dispersion with the largest particles sedimented in less than one day, whilst that with the smallest particles was stable for more than 1 month. Figure   S7: HKUST-1/PDMS dispersion analysis with particle sizes 400nm and 50nm). 10

iv) Matching the densities of solid and liquid phases
Dispersions of the highly fluorinated MOF SIFSIX-3-Zn in liquids such as silicone oils or triglyceride oils were found to be unstable, settling out within an hour. This can be ascribed to its high crystallographic density of 1.57 g/cm 3 [13] . Use of an oil with high density, specifically the fluorinated oil Fomblin ® Y (density: 1.88 g/cm 3 ; cf. silicone oil density: 0.96 g/cm 3 ) was found to give far more stable dispersions.  Figure S14. Integrals of the trace for the respective components of the porous liquid were obtained by a variation of Rietveld refinement previously reported. 17 The ratios of the integrals were calculated by simple division and normalised to 12.5% (in relation to the theoretical amount of crystalline material present in the upper layer for a stable dispersion). The calculation was performed each day for nine days and results are plotted graphically in Figure   2b. Gas uptake studies were carried out using an isochoric method described elsewhere [18] . The measurements were carried out at 298K and the average equilibrium pressure was ca. 0.85 bar ( Figure S15). Total internal volume of the apparatus was calibrated with corresponding gases before measurement. Equilibrium was deemed to be reached when the system pressure was unchanged for at least 2 hours. Figure S15: Apparatus for isochoric gas solubility measurement.
Gas uptakes of the pure porous solids were measured to compare with literature values (Table S3). Gas uptakes of pure liquid media were also measured for comparison with the gas uptakes of porous liquids. All porous liquids consisted of 12.5 wt% porous solid content, unless otherwise stated.  All solid content in porous liquids are measured at w solid = 12.5 wt% loading, unless specified.
A direct comparison of the measured and predicted (eq. 1) gas uptake values (experimental -predicted uptake values), shown in Table S4a -c below, allows us to infer whether or not the liquid phase has entered the pores of the solid.
As shown in Tables 1a and 1b, most of the compositions show enhanced CO 2 uptake compared to the pure oils and the experimental values are comparable to the values predicted from the uptakes and proportions of the components using eq. 1, indicating that CO 2 uptake behaviour of porous liquids seems to be nearly ideal and thus predictable.
In cases where highly thermally stable liquid media were used (e.g. silicone oils), T3PLs could be activated directly without pre-activation of porous solids before forming T3PLs. Additionally, higher (25wt%) loadings were analysed for some solid/liquid combinations and good agreement between experimental and predicted uptakes were again seen ( Table S4e).         Representative examples for CO 2 selectivity over N 2 (CO 2 /N 2 ) and over CH 4 (CO 2 /CH 4 ) were calculated for the porous liquids and the results are shown in Tables S7a and S7b.  Most of the compositions exhibit greater selectivity for CO 2 over N 2 and CH 4 than do the pure oils, indicating the addition of porous solid enhances the overall CO 2 selectivity over to N 2 and CH 4 .
S.I. 7: CO 2 uptake kinetic data for selected T3PLs ns compared to vacuum lifetime of ~142 ns) [30][31][32] . The magnitude of this reduction in lifetime is governed by the number of collisions of the o-Ps with the 'pore-walls', thus providing a relatively accurate correlation to the pore sizes and size distributions [30][31][32] . In the same fashion, the presence of any ingressing molecules into the pores, the collisions of the o-Ps with molecules within the pore would also reduce the o-Ps lifetime providing a signature of the changed pore geometry. [30][31][32] A measured positron lifetime spectrum would thus consist of multiple components (as can be seen in Figure S17) Therefore, in the analysis of the lifetime spectra, we use fitting of both discrete lifetimes of each annihilation modes as well as lifetime distributions around the individual discrete values. [30][31][32] In The lifetime distribution mode, the 'peak' of the distribution gives a measure of the average lifetime obtained from the discrete mode. In the context of this paper, it is the o-Ps lifetimes that provide the information regarding the pore sizes. In complex situations, as is the case here, evaluating a combination of the average lifetimes and their spread are essential to arrive at a comprehensive picture of the underlying physics/chemistry.  To illustrate the principles, we first look at a detailed lifetime analysis for pure ZIF-8 in Figure   S17. Here, the samples were measured as received (not degassed) and then the gas was

S.I. 9 Regeneration
Since they are physical sorbents these porous liquids are expected to be easily regenerated by applying mild heating or vacuum. As shown in Figure S19, two porous liquids (12.5 wt% HKUST-1/PDMS and 12.5 wt% Al(fum)(OH)/PMDS were found to recover at least 75% of their CO 2 uptake capacity by applying vacuum for 2h. However, the conventional aminebased solution (12.5 wt% MEA/H 2 O) shows around 5% recovery. A regeneration study with HKUST-1/PDMS was also conducted. Preliminary tests using c.a.
13 wt% HKUST-1 in silicone oil shows that almost all of the porous liquid capacity can be regenerated by applying vacuum (up to 8.5 x 10 -2 bar) to remove captured CO 2 and the CO 2 uptake capacity maintains for at least 5 cycles (Figure S20).
12.5wt% Al(fum)(OH)/PDMS Figure S22: Reproducibility of CO 2 uptake for two Type 3 porous liquids over three cycles using brief evacuation to regenerate the porous liquid between measurements, compared to that of an aqueous amine solution (red) which shows greater initial uptake but is not regenerated under the same conditions S.I. 10 High pressure gas solubility measurements (1-5 bar, 25°C-75°C) High pressure gas uptake studies were carried out using a Parr reactor based on mass flow. Data points were taken for 60 minutes at each pressure ( Figure S23). As shown in the data, at low pressure (1 bar  All measurements were carried out between 1 -5 bar at 298 K, 323 K and 348 K. Data for Al(fum)(OH) are given in Table S8. As with the low-pressure measurements, the measured CO 2 uptake of porous liquids at high pressure is comparable to the predicted values based on the uptakes of the individual solid and liquid components.  (Table S9 and S10) for selected porous liquids.   S.I. 12 Biocompatible CD-MOF-1/Olive oil T3PL Figure S24: CO 2 uptake of 12.5 wt% CD-MOF-1/olive oil porous liquid versus olive oil.