Incorporation of CO₂ philic moieties into a TiO₂ nanomembrane for preferential CO₂ separation

Abstract

Here we report a preferential CO₂ separation membrane consisting of a nanometer-thick TiO₂ layer incorporated with phtalic acid (PA) molecules on polydimethylsiloxane (PDMS) (PA@TiO₂/PDMS). Incorporated PAs in TiO₂ act as CO₂-philic pores for preferential CO-₂ permeation over nitrogen. CO₂ binding to the PA incorporated in TiO₂ is confirmed by the density functional theory calculation (DFT). As a result, membranes with of PA@TiO₂ layer demonstrated much higher selectivity to CO₂ for mixed CO₂/N₂ gas separation compared to a conventional PDMS membrane. The exceptional selectivity of the composite layer alone (>150) was estimated by a resistance model.

---

Intensive emission of CO₂ into the atmosphere due to combustion of fossil fuels causes global warming.¹ To prevent the emission of carbon dioxide into the atmosphere, carbon capture and storage should be introduced in mass emission sites such as coal-fired power plants.

While CO₂ separation by membranes is a promising method to reduce capture costs,^2,3 it has not been widely used, because it still requires higher gas flux, durability and selectivity. Organic polymer membranes have been mostly investigated, however despite the good results achieved in the separation of CO₂, polymers are lacking long-term usability and chemical stability for industrial use.⁴ In contrast, inorganic membranes, for instance made of zeolites, have better chemical and physical stability, at the same time have less flexibility in molecular design of membrane materials and membrane fragility limits further developments. Composites of organic and inorganic materials would have potentials to integrate useful characteristics of both components in membrane material. For example, Lau et al. reported that the addition of inorganic fillers into the organic polymer matrix enabled not only to prevent aging in super glassy polymers but in some cases increase the selectivity of the materials⁵ to overcome limits set by Robeson upper bound.^6,7 One of the approaches in designing gas separation membrane is to control the CO₂ affinity of membrane materials. Recent computational works has revealed such molecular interactions between CO₂ and sorption material. Torrisi et al. showed that the use of polar functional groups in benzene ring has a major impact in improving in the CO₂-ligand affinity and the best –H substituents are found to be –NH₂, –SO₃H, and –COOH.^8,9 Similar finding was reported by Dasgupta et al.¹⁰ who have assessed the influence of specific functional groups on the adsorption selectivity of CO₂/N₂ mixtures with model system consisting of a bilayer graphene nanoribbon that has been edge functionalized with –OH, –NH₂, –NO₂, –CH₃ or –COOH. Their studies also revealed that functionalization always leads to an increase in the adsorption of both CO₂ and N₂. However, enhancement in selectivity was detected only for –COOH edge functionalized graphene nanoribbons so the authors concluded that specific functionalization with –COOH groups can provide material's design strategy to improve CO₂ selectivity in microporous adsorbents. These results inspire that inorganic membranes possessing abundance of carboxylic groups would have potential to develop CO₂ separation membrane.

In our research, we design materials based on titanium dioxide composited with aromatic carboxylic acids resulting in the abundance of free –COOH groups in the membrane matrix. Phthalic acid (PA) was employed as a source of –COOH moieties incorporated into the metal oxide. It has been reported that aromatic carboxylate derivatives can be easily incorporated into a titanium matrix via the molecule complexation with titanium alkoxides.¹¹

In detail, a glass substrate was coated with a layer of poly (styrenesulfonic acid) sodium salt (PSS) with the thickness of ca. 1 μm by spin-coating. Then polydimethylsiloxane (PDMS) support film with the thickness of ca. 2 μm was prepared on the substrate by spin-coating of 20% PDMS solution in hexane. In order to make the PDMS surface hydrophilic for the formation of uniform TiO₂ layer on it, PDMS film was treated by O₂ plasma. Separately, PA (10 mM) and titanium n-butoxide (100 mM) were dissolved in n-butanol and stirred for 12 hours at room temperature for their complexation. This mixture solution was then spin-coated on the plasma-treated PDMS (p-PDMS) with the rotation speed of 3000 rpm for 1 min to form the composite layer of PA@TiO₂ (PA@TiO₂/p-PDMS). As a reference sample, a TiO₂ layer without PA (TiO₂/p-PDMS) was also prepared by the same procedure with the use of same concentration titania precursor solution. The formation of the TiO₂ layer by spin coating of its corresponding precursor solution is a well-known phenomenon. Due to the relatively quick hydrolysis of TiO₂ precursors (in our case Ti(O-ⁿBu)₄) a film is readily formed under ambient temperature and humidity.¹¹ Subsequently the substrate was immersed in water to detach the films from the substrate by dissolving the PSS layer. Detached membrane floating on water surface was transferred onto a porous support for gas permeation experiments.

Fig. 1(a and b) shows the optical image of the p-PDMS, TiO₂/p-PDMS (both used as references) and PA@TiO₂/p-PDMS membranes surfaces respectively.† Fig. 1d shows the typical cross-section SEM image of the inorganic (or composite) film deposited to the p-PDMS surface confirming the formation of TiO₂ layer with about 100 nm thickness. After the deposition of TiO₂ and PA@TiO₂ layers on p-PDMS, cracks appear on the surface as shown in Fig. 1(b and c). Before detaching the film from a glass substrate, no cracks were observed. During the detachment, a membrane undergoes flexible movements while floating on water and thus, this is the step when cracks are introduced, since a ceramic film is not so elastic like organic polymer membranes. Defects in the gas separation membrane are a considerable drawback, since defects result in simple gas leak. Despite cracks in the composite layer, PA@TiO₂/p-PDMS membrane showed preferential CO₂ permeation over nitrogen as shown in Fig. 2.† Namely, the pristine p-PDMS membrane showed CO₂ permeance higher than other membranes and CO₂ selectivity over N₂ (α(CO₂/N₂) ≃ 4.3). Deposition of pure TiO₂ on a p-PDMS membrane resulted in suppressing the both gases permeance and gave the similar gas selectivity (α(CO₂/N₂) ≃ 3.6). However, with PA incorporation into TiO₂ layer much higher decrease was observed in nitrogen permeation, resulting in the enhancement of CO₂ selectivity (α(CO₂/N₂) ≃ 16.1). This selectivity enhancement compared to the pristine p-PDMS is a clear indication that PA@TiO₂ contributed significantly to the selectivity enhancement, despite the number of cracks observed on the membrane surface.

Fig. 1 Optical microscopy images of the surface of (a) PDMS, (b) TiO₂/p-PDMS and (c) PA@TIO₂/p-PDMS membranes; (d) cross-section SEM image of the membrane top demonstrating the attachment of the composite TiO₂ layer to the surface of the p-PDMS support.

Fig. 2 Permeance of CO₂ and N₂ through the composite PA@TiO₂/p-PDMS membrane compared to plasma treated PMDS (p-PDMS) and TiO₂/p-PDMS membranes. O₂ plasma treatment time (a) 30 s, (b) 5 s (feed gas – 1 : 1 mixed N₂ and CO₂).

In the process of the membrane fabrication, oxygen plasma treatment is needed to prepare a well-attached and uniform a TiO₂ layer on a PDMS. But this surface oxidation results in the reduction of both gas permeance and selectivity. Earlier works suggested that this is caused by the formation of the thin SiO₂ layer on the PDMS due to treatment.¹² In order to minimize the effect of a SiO₂ layer, the plasma time was reduced to only 5 s. The gas permeance obtained from the membranes with shorter plasma treatment is higher than that of the membranes with longer plasma treatment (Fig. 2b). Long term performance of the optimized PA@TiO₂/p-PDMS (5 s plasma treatment) for the CO₂/N₂ (1 : 1) dry mixture gas separation is shown in Fig. 3. It is seen that the membrane performance is stable over 25 hours (the observed selectivity α(CO₂/N₂) of PA@TiO₂/p-PDMS is higher than 30). Although further improvements of a TiO₂/p-PDMS membrane morphology is necessary, we believe that evaluating the potential of this composite membrane is worthy to prove our design concept. In order to evaluate the selectivity and permeability of the PA@TiO₂ layer alone, the resistance model was employed.^13,14

Fig. 3 (a) Long time performance of the PA@TiO₂/PDMS membrane for the mixed CO₂ and N₂ separation and (b) permeability–selectivity relation for PDMS, composite membrane PA@TiO₂/PDMS and selective layer PA@TiO₂ (calculated by resistance model).

According to the resistance model, total permeance P_AB of the simple double layer defect-free membrane schematically depicted below can be calculated using eqn (1).¹³

where P_A and P_B are the gas permeances of the layer A (a composite TiO₂ in our case) and B (PDMS in our case) respectively, shown as the schematic illustration in the eqn (1). In order to account the influence of the defects in the composite layer as schematically shown below, the eqn (2) combined with eqn (1) should be used.^13,14where P_M is the experimentally measured permeance of composite membrane, a_A and a_B are the surface area of TiO₂ coated PDMS (represented as a Layer A) and PDMS support layer (represented as a Layer B), respectively. The intact area of the TiO₂ layer (a_A) was estimated to be about 96–98% by analysing the SEM images of the membrane surface with the image analysis software.† The calculated selectivity vs. permeability for PA@TiO₂, PA@TiO₂/p-PDMS and ideal PDMS alone¹⁷ are plotted in Fig. 3b. These are lying close to the Robeson upper bound.⁷ Noteworthy, PA@TiO₂ layer shows exceptionally high selectivity (α > 150). This results implies that CO₂ affinity of PA is integrated in a composite membrane.

In order to explain the observed selectivity of PA@TiO₂/p-PDMS membrane, gas affinities of PA derivatives to CO₂ were calculated by using DFT method.† The results of the geometry optimization and binding energies calculation of CO₂ to phthalic acid, phthalic acid methyl ester and phthalic acid dimethyl ester are summarized in Fig. 4. As a result of the esterification, i.e., a model that represents the binding of the phthalic acid to the TiO₂ framework, the binding energy of CO₂ gradually decreases from −5.674 kcal mol⁻¹ for the free acid, to −3.765 kcal mol⁻¹ for the methyl ester, and finally to −2.510 kcal mol⁻¹ for the dimethyl ester. The binding energy of N₂ to phthalic acid and its esters is calculated to be always negligible.

Fig. 4 Binding energies of CO₂ to the phthalic acid in various model states (free, partially bound, and completely bound) obtained using DFT calculations.

The high binding energy of CO₂ to PA compared to that of nitrogen, is caused mainly by hydrogen bonding interaction between CO₂ and carboxyl group of PA similarly to that reported by Torrisi et al. for benzoic acid⁹ and Dasgupta et al. for –COOH functionalized graphene.¹⁰ In our case, even monovalent coordination of PA to TiO₂ decreases the binding energy, it is still high enough compared to nitrogen. However, full loss of free carboxylic acid group likely decrease binding energy leading to losing the gas separation selectivity.

Considering these results, possible separation mechanism can be discussed. In the case of only TiO₂ coating, unreacted butoxide groups remains in a TiO₂ matrix, since condensation reaction between titanium precursors may not be proceeded sufficiently under current mild condition. These unreacted moieties may act as micropores for gas permeation. However, the number of micropores may not be sufficient to maintain the gas permeance of a pristine PDMS membrane, resulting in the decrease of the permeance of CO₂ and N₂ equally. Incorporation of PA molecules into a TiO₂ matrix enhances the CO₂ affinity and therefore TiO₂ matrix would uptake CO₂ preferentially over N₂. As a result, the space of micropores in TiO₂ would become smaller due to limited physical space for simple gas diffusion without any chemical interaction. Thus, N₂ permeation is suppressed and the gas selectivity of CO₂ over N₂ becomes high. This description has not been fully confirmed experimentally and further investigation of gas permeation such as comparison of mixed gas vs. pure, solubility and diffusion tests, feed pressure influence etc. are currently under way, and the comprehensive discussion will be reported elsewhere.

In conclusion, we successfully incorporated the CO₂ philic molecules into the TiO₂ membrane. As a result, the enhancement of selectivity to CO₂ for the N₂/CO₂ mixed gas separation was observed. Application of resistance model for membrane parameters calculation revealed that the PA@TiO₂ layer has unprecedented selectivity (>150) towards the CO₂/N₂-separation. DFT calculations demonstrated that the preferential binding of CO₂ to free carboxylic groups of the phthalic acid plays an important role to enhance gas selectivity. Our current approach is very simple and versatile. The material design used in this work was only predicted by computation in earlier works.^8–10 To the best of our knowledge our work is a first experimental demonstration of attributed selectivity in titania membrane due to incorporation of small molecules with CO₂ affinity. We believe that this approach can be further sophisticated to fabricate better gas separation membranes. Improved gas separation performance may be possible by selection of molecules with higher CO₂ affinity and incorporating them into a nanometer-thick membranes.

Acknowledgements

This work was supported by World Premier International Research Center Initiative (WPI), MEXT, Japan. Also work was financially supported by the Japanese Society for the Promotion of Science (JSPS Grant-in-aid for Research Activity Start-up, No. 26889045).

Notes and references

1. D. F. Sanders, Z. P. Smith, R. Guo, L. M. Robeson, J. E. McGrath, D. R. Paul and B. D. Freeman, Polymer, 2013, 54, 4729–4761.
2. J. Hansen, M. Sato, R. Ruedy, K. Lo, D. W. Lea and M. Medina-Elizade, Proc. Natl. Acad. Sci. U. S. A., 2006, 103, 14288–14293.
3. Basic research needs for carbon capture: beyond 2020, US Department of Energy Basic Energy Sciences Workshop for Carbon Capture, 2010.
4. M. Pera-Titus, Chem. Rev., 2014, 114, 1413–1492.
5. C. H. Lau, P. T. Nguyen, M. R. Hill, A. W. Thornton, K. Konstas, C. M. Doherty, R. J. Mulder, L. Bourgeois, A. C. Y. Liu, D. J. Sprouster, J. P. Sullivan, T. J. Bastow, A. J. Hill, D. L. Gin and R. D. Noble, Angew. Chem., Int. Ed., 2014, 53, 5322–5326.
6. L. M. Robeson, J. Membr. Sci., 1991, 62, 165–185.
7. L. M. Robeson, J. Membr. Sci., 2008, 320, 390–400.
8. A. Torrisi, C. Mellot-Draznieks and R. G. Bell, J. Chem. Phys., 2009, 130, 194703.
9. A. Torrisi, C. Mellot-Draznieks and R. G. Bell, J. Chem. Phys., 2010, 132, 044705.
10. T. Dasgupta, S. N. Punnathanam and K. G. Ayappa, Chem. Eng. Sci., 2015, 121, 279–291.
11. S. W. Lee, S. Korposh, R. Selyanchyn and T. Kunitake, in Handbook of Molecular Imprinting: Advanced Sensor Applications, Pan Stanford Publishing Pte. Ltd, 2012, pp. 3–62.
12. D. A. Markov, E. M. Lillie, S. P. Garbett and L. J. McCawley, Biomed. Microdevices, 2014, 16, 91–96.
13. J. M. S. Henis and M. K. Tripodi, J. Membr. Sci., 1981, 8, 233–246.
14. M. Ai, S. Shishatskiy, J. Wind, X. Zhang, C. T. Nottbohm, N. Mellech, A. Winter, H. Vieker, J. Qiu, K.-J. Dietz, A. Gölzhäuser and A. Beyer, Adv. Mater., 2014, 26, 3421–3426.
15. TURBOMOLE V6.2 2010, a development of University of Karlsruhe and Forschungszentrum Karlsruhe GmbH, TURBOMOLE GmbH, since 2007, 1989–2007, available from http://www.turbomole.com.
16. S. Grimme, J. Antony, S. Ehrlich and H. Krieg, J. Chem. Phys., 2010, 132, 154104.
17. W. L. Robb, Ann. N. Y. Acad. Sci., 1968, 146, 119–137.

---

Footnote

† The cross-section and surface morphologies of the membranes were observed by FE-SEM. The membrane samples were obtained by breaking the membranes in liquid nitrogen and coating them with thin layer of platinum before measurement. Optical images were obtained using Keyence VHX-500 digital microscope. In order to estimate the area of PDMS exposed due to the cracks formations in the composite layer, image analysis software (Image-Pro Plus v.4.5) was used to analyse the surface SEM images.

For the gas permeance measurements – membrane area was masked with Kapton and alumina tapes to provide the circle of desired diameter and area (d = 1 cm, a = 0.785 cm²). Additionally, to prevent bending upon vacuuming membrane was placed on the porous polycarbonate support filter (1.2 μm pore size). The dry 1 : 1 mixed N₂/CO₂ permeation through the membranes at 25 °C was measured using the GTR-11A/31A gas barrier testing system (GTR Tec Corp., Japan). The part of the machine where test gas is sampled (GTR-31AKU) uses differential-pressure method for film permeability testing i.e. gas permeation is induced by the vacuum on the permeate side and extra pressure applied at the feed side. Total pressure difference was set to 200 kPa. The gas volume was measured by gas chromatograph with a thermal conductivity detector (TCD) (Yanaco G3700T, Japan). Gas permeance and permeability were calculated and represented in GPU and barrer units, where GPU = 7.5 × 10⁻¹² m³(STP) m⁻² s⁻¹ Pa⁻¹ and barrer = 7.5 × 10⁻¹⁸ m³(STP)·mm⁻² s⁻¹ Pa⁻¹. Selectivity α(CO₂/N₂) was calculated as ratio of the CO₂ and N₂ permeances through the membrane.

Density functional theory calculations were performed using the TURBOMOLE program¹⁵ and B3LYP level of theory with TZPP basis set and DFT-D3 dispersion energy correction¹⁶ in order to estimate the binding energy of phthalic acid and its esters to CO₂ and N₂ molecules.

---