Micro-nanoarchitectonic of aluminum-hydrogel propellant with static stability and dynamic rheology

Abstract

The aluminum-water system is a promising propellant due to high energy and low signal characteristics, and the gel form is easier to store and utilize. In this work, hydrogels of water and aluminum particles were prepared using the low-molecular-weight gellant agarose. The various physical properties of gel systems, including the water loss rate, phase transition temperature, and centrifugal stability at different gellant and aluminum contents, were examined. Rheological properties were assessed through shear thinning tests, thixotropy tests, strain sweep analysis, and frequency sweep experiments. The microstructure of the gel was obtained through scanning electron microscopy images. The results show that the aluminum-hydrogel network structure is composed of micron-scale aluminum and agarose nanosheets, and the unique micro-nanostructure endows the gel with excellent mechanical strength and thermal stability, which improve with increasing gellant and aluminum contents. Notably, the gel with 2% agarose and 20% aluminum had the best performance; the storage modulus reached 90647 Pa, which was within the linear viscoelastic region, and the maximum withstand pressure was 111.2 kPa, which was 118.8% greater than that of the pure hydrogel. Additionally, the gel demonstrates remarkable shear thinning behavior and can undergo gel-sol transformation upon shearing or heating to exceeding 114 °C.

Graphical abstract

Keywords

aluminum-water propellant / gelled propellant / rheological property / mechanical strength

1 Introduction

Given that the historical evolution manifests the blooming population, speedy civilization, and increased fossil fuel combustion, massive of greenhouse gases (e.g., CO₂, methane, nitrous oxide) are being emitted due to anthropogenic activities and the progressive manufacturing industry [1,2]. Among the greenhouse gases mentioned above, CO₂ accounts for the highest proportion (80%). It was arguable that the global CO₂ emission in 2022 was expected to reach 36.1 Gt (Fig.1) [3]. As claimed by the universal emission trend, a cursory extrapolation was drawn that CO₂ will surmount 450 parts per million before 2050 and go beyond the threshold value that is customarily deemed as the ceiling [4]. The immense increment in CO₂ reserve will ineluctably trigger global warming, in turn bringing about a series of climate and environmental issues [2,5]. When it comes to the enduring geochemical and climate risks derived from high-growing atmospheric CO₂ contents, climate engineering manifests fatigue to the fullest. It is a matter of considerable interest to cope with the semi-permanent climate variation hazard, especially decarbonizing quicker than nature does, to consign Earth a warmer future for millennia or commit ourselves to a sustainable paradigm. Nevertheless, humanity’s carbon emission halt is essentially impossible, and the climate risks they posed up rise monotonically even at on-going augmented CO₂ content. CO₂ capture can mitigate global warming, help protect the environment and the Earth’s climate, promote sustainable economic growth and create a cleaner living environment for future generations. Therefore, carbon capture project is a practical and indispensable system to manage the disturbance of the carbon cycle, and it is one of the key technologies to realize sustainable development and build a low-carbon economy.

Fig.1 Global CO2 emissions 1970–2022. Reprinted with permission from Ref. [3], copyright 2023, Springer Nature.

Full size|PPT slide

Carbon capture and storage and carbon capture and utilization are required to limit the burgeoning Earth’s temperature originating from the soar rising CO₂ [1,4,6−8]. New techniques and materials for such deep reduction in carbon content are still sought after. Presently, post-combustion capture, pre-combustion capture, chemical looping combustion, oxyfuel combustion, direct air capture and capture from fermentation routines are the prominent technologies for carbon capture [8,9]. Of these options, post-combustion CO₂ capture is the most commonly utilized in terms of its high maturity and readiness. This technique comprises amine scrubbing, membrane CO₂ capture and cryogenic CO₂ capture system. Membrane separation, as an environmental-friendly technology, has been certified to be the most encouraging scheme for CO₂ capture and recovery [10−13]. Polymeric membranes featured with lower energy consumption, reduced investment costs and simplified handling requirements have attracted increasing attention, nonetheless, suffered from permeance-selectivity trade-off. Meanwhile, homogeneous polymeric membranes, such as polysulfone (PSf) [14], polyimides (PIs) [15] and polydimethylsiloxane (PDMS) [16] could not fulfill industrial efficiency due to thick membrane bulk compositions (usually larger than 10 μm) [17−19]. Recently, thin film composite (TFC) membranes made of polymerization chemistry extend the membrane applicability and showcase favorable perspectives for meticulous majorization of CO₂ capture performance and membrane durableness [20,21].

TFC membrane, in which an ultrathin-dense selective layer deposited onto a porous support, has garnered increasing attention [22,23]. The porous support serves as a fundamental structure without imparting additional transport resistance (generally with an ultrahigh gas permeance as 10⁴–10⁵ GPU), while selective layer is the critical part of a TFC membrane, which is usually synthesized via interfacial polymerization (IP) process, granting TFC membrane with concentration and separation properties [24]. The term “IP” was put forward in 1955 and subsequently, was motivated and ushered to maturity rapidly. The specific spatial boundary, interface between two distinct matters (generally liquid-liquid interface), undertakes prominent responsibilities for chemical reaction and material synthesis. For customary IP reactions, monomers in immiscible bulk phases undergo mutual diffusion at the constrained interface to induce the production of films with dissimilar propensities. The films originating from IP are normally homogeneous and defects-free because the monomer diffusion without dense boundaries is relatively easier, permitting the successive expansion of film. Besides, IP is a self-terminating process as steric hindrance imposed by the incipient layer blocks reactive monomers’ further diffusion, allowing the manufacturing highly permeable barrier with ultrathin thickness (less than 50 nm) [25]. The rising of innovative superspreading interface engineering and polymer chemistry (supramolecular polymers, metal-organic frameworks (MOFs) and covalent-organic frameworks (COFs)) also implant fresh vitality into IP. By altering the IP reaction parameters and monomers/solvents/additives/categories, thin film’s morphologies and chemical properties could be controllably adjusted for precisely matching specified separation systems. The advantages of the IP process, such as adjustable membrane thickness and structure, low cost, simple operation, low energy consumption and easy industrialization, are of great importance in realizing an efficient and cost-effective CO₂ capture process.

IP has manifested progressive power for membrane separation, particularly CO₂ capture/separation. Despite its late start, various IP membranes have been designed for targeted gas pairs (e.g., CO₂/N₂, CO₂/H₂, CO₂/CH₄, H₂/CO₂, H₂/N₂, N₂/CH₄ and N₂/O₂) separation. The progress of CO₂ separation membrane synthesized by IP reaction has been reviewed from previous work by giving equal considerable attention to the mixed gases system, including water vapers and other acid gases [23]. Nevertheless, these papers rarely mentioned the molecular dynamics (MD) simulation and machine learning (ML), as a powerful tool to predict membrane structure and performance for CO₂ separation, giving less attention than it deserves. It is more promising and our interest to expect exceptional agreement by coupling MD simulation and ML with experiment measurements. In the light of dilemmas mentioned above, this review concentrates on the recent progress of IP process for CO₂ separation membranes construction, to compensate for the scope gap with a comparative copious summary and discussion and attempt to look forward to the future of IP for carbon capture. This review is organized as follows: (1) first, we introduced the fabrication of IP membranes with alternated aqueous phase monomer (amine or hydroxyl monomers) for CO₂ separation (Section 2); (2) the thin film nanocomposite (TFN) membrane was illustrated and the corresponding coherence between membrane structure and nanofiller nature was analyzed in detail. Besides, general strategies for improving the nanofillers compatibility in selective layers of TFN membranes were provided (Section 3); (3) then, IP regulation strategies were described type by type (Section 4); (4) through an unprecedented integration of MD simulation and ML with membranes, accomplishing the membrane compositions designation and further enlightening researches predicts IP membrane performance for CO₂ separation (Section 5); (5) finally, challenges and prospects of integrating IP membranes for CO₂ separation were also demonstrated along with some conclusions (Section 6). This review aims to reveal the enormous potential of IP membranes for CO₂ separation to encourage relevant researchers and provide valuable guidance for future research endeavors.

2 IP of TFC membranes for CO₂ separation

Membrane separation performance is dominated by its microstructure and physicochemical properties [26]. Because the selective layer of IP membrane is formed by the polycondensation of the monomers from aqueous phase and organic phase on the substrate surface (PSf or polyethersulfone (PES)), these monomers can fundamentally tailor the microstructure and physicochemical properties of the obtained membranes, thereby indicating a significant influence on their separation performances (Fig.2). Nowadays, most of the research has focused on the aqueous phase monomers, because trimethylbenzoyl chloride (TMC) with low solubility in the aqueous phase, low price and excellent reaction activity seems to be the most favored organic phase monomer [23]. Over the previous decades, researchers have made a lot of unremitting efforts to explore aqueous phase monomers for CO₂ separation. In this section, some novel aqueous phase monomers applied for developing a high-performance IP membrane for CO₂ separation will be reviewed.

Fig.2 Synthesis of IP membrane and amine and hydroxyl group aqueous phase monomers.

Full size|PPT slide

2.1 Amine monomers for membranes synthesized

As the most widely used aqueous phase monomers for preparing IP membranes, amine monomers can easily react with acyl chloride to form an ultra-thin selective layer. It has been proved that the reversible reaction between the amine group and CO₂ can effectively enhance the performance of membrane for CO₂ separation [27]. In this section, IP membranes prepared by different amine monomers for CO₂ separation are reviewed. The commonly used amine monomers and the performances of IP membranes for CO₂ separation synthesized by these monomers are listed in Tab.1.

Tab.1 Membranes synthesized by amine monomers and their performancesa)

Support	Aqueous phase monomers	Organic phase monomers	CO2 permeance/GPU	CO2/N2 selectivity(gas ratio)	CO2/CH4 selectivity(gas ratio)	Ref.
PES	TETAC6H18N4, MW 146.23	TMC	13.3	–	94.1 (10/90)	[ 28]
PSf	MPDC6H8N2, MW: 108.14	IPC	15.2	–	14.4 (pure)	[ 29]
PSf	DNMDAMC7H19N3, MW: 145.25	TMC	173	70 (20/80)	37 (10/90)	[ 30]
PSf/PDMS	DGBAMEC8H20N2O2, MW: 176.26	TMC	973	84 (15/85)	31 (10/90)	[ 31]
PSf/PDMS	PEA	TMC	360	67.2 (20/80)	31.5 (10/90)	[ 32]
PSf/PDMS	DAPPC10H24N4, MW: 200.32	TMC	420	85 (15/85)	40 (10/90)	[ 33]
PSf	CHMAC8H18N2, MW: 114.19	TMC	25	–	28 (30/70)	[ 34]
PSf	PIPC4H10N2, MW: 86.14	IPC	2.4	4.32 (pure)	22 (pure)	[ 35]
PSf/PDMS	PI	TMC	330	30	32	[ 36]
PSf/PDMS	DGBAME&DNMDAM	TMC	1612	138 (15/85)	52 (10/90)	[ 37]
PSf/PDMS	DGBAME&DAMBSC7H7N2NaO2, MW: 174.13	TMC	5831	86 (15/85)	–	[ 38]

a) MW: molecular weight; TETA: triethylenetetramine; MPD: m -phenylenediamine; DGBAME: bis(3-aminopropyl) ether; DNMDAM: 3,3’-diamino- N -methyldipropylamine; PEA: polyetheramine; DAPP: 1,4-bis(3-aminopropyl)piperazine; CHMA: 1,3-cyclohexanebis-methylamine; DAMBS: sodium 3,5-diaminobenzoate; PIP: piperazine; IPC: isophthaloyl chloride.

PIP and MPD are the most common amine monomers for an IP reaction. Most of the commercial IP membranes are also prepared by PIP and MPD. In 2013, for the first time, Andrew et al. [39] tested the CO₂ separation performance of 4 kinds of commercial nanofiltration membranes formed by cross-linking PIP and TMC. Although the GE HL and Trisep TS40 only showed a CO₂/N₂ selectivity below 3, the Trisep XN45 and Dow NF3838/30FF exhibited a CO₂/N₂ selectivity of 14 and 19, respectively. Besides, Sridhar et al. [29] studied the performance of the IP membranes for CO₂ separation using MPD as the aqueous monomer. The IP membrane obtained a CO₂ permeance of 15.2 GPU and a CO₂/CH₄ selectivity of 14.4. Awad et al. [35] prepared an IP membrane with a CO₂ permeance of 2.4 GPU and a CO₂/N₂ selectivity of 22 through an IP reaction between PIP and IPC. Due to fewer acyl chloride groups on IPC, the IP membrane synthesized by PIP and IPC possessed more unreacted amine groups than the membrane synthesized by PIP and TMC, thus showing a higher selectivity. The above studies confirmed that the IP membranes synthesized from amine monomers are feasible for CO₂ separation. This can be rationalized by the dense and defect-free selective layer and the reactivity of amine groups with CO₂.

However, it seems that PIP and MPD are not competent for preparing a high-performance CO₂ separation membrane, thereby other amine monomers have also been investigated to prepare an advanced membrane. For example, TETA was applied by Zhao et al. [28] to react with TMC for preparing a novel IP membrane. This membrane exhibited a CO₂ permeance of 13.3 GPU and a CO₂/CH₄ selectivity of 94.1. Such a high selectivity may stem from the abundant amine groups in TETA (containing both primary and secondary amines). Jo et al. [34] successfully developed a novel inside-coated hollow fiber membrane by IP process between CHMA and TMC [34]. This membrane presented a CO₂ permeance of 25 GPU and a CO₂/CH₄ selectivity of 28. To further increase the permeance of the IP membrane, Xu et al. [36] synthesized a water-soluble PI-based amine monomer and successfully prepared a novel IP membrane by reacting it with TMC (Fig.3). The membrane exhibited a high CO₂ permeance of 330 GPU while the CO₂/N₂ and CO₂/CH₄ ideal selectivity are 30 and 32, respectively. This obviously improved CO₂ permeance can be ascribed to the rigid micropores (3–9 Å) of the PI and the thin selective layer (104 nm). It has been well known that the tertiary amine groups have higher CO₂ stoichiometric loadings and faster amine-CO₂ reaction rates than primary and secondary amines [40], thereby endowing the membrane with superior CO₂ separation abilities. To take the advantages of tertiary amine groups, Yu et al. [30] first prepared a novel IP membrane by crosslinking DNMDAM and TMC. The obtained membrane exhibited a CO₂ permeance of 173 GPU along with a CO₂/N₂ selectivity of 70. Similarly, DAPP with tertiary amine groups was also applied to react with TMC [33]; the optimal membrane exhibited a CO₂ permeance of 420 GPU and a CO₂/N₂ selectivity of 84. It should be noted that although the tertiary amine groups in IP membrane can significantly improve the CO₂ separation performance, it can barely react with CO₂ in the absence of water. Therefore, suitable humidification was required to guarantee the excellent CO₂ affinity of the tertiary amine. And the transport of CO₂ and N₂ in the membrane is improved when the membrane is properly swollen by water, but it is more beneficial to CO₂ [41]. Therefore, the CO₂ separation performance of IP membrane can be enhanced under the condition of wet gas. In addition, the poor stability of the swollen membrane cannot be ignored.

Fig.3 (a) Synthesis of membrane from amine-end-capped PI oligomers and TMC and the three-dimensional (3D) view of an amorphous cell containing the cross-linked network; (b) schematic diagram of preparing the selective layer by the IP process. Reprinted with permission from Ref. [36], copyright 2021, American Chemical Society.

Full size|PPT slide

Due to the strong polar interaction with CO₂, ether oxygen (EO) groups are also promising to enhance the performance of the IP membrane [42]. By merging the EO group into amine monomers, DGBAME and diamino polyethylene glycol have been explored to synthesize IP membrane with TMC (named EO-3 and EO-21 respectively, based on the length of the EO units) [31]. EO-3 and EO-21 exhibited a CO₂ permeance of 973 and 1310 GPU along with a CO₂/N₂ selectivity of 84 and 33, respectively. The result showed that the CO₂ permeation augmented with increasing EO chain length in monomer, while selectivity tended to decrease. Different from polyamide (PA) IP membrane, Ding et al. [43] prepared a PI IP membrane containing EO groups by cross-linking commercial diamines, o,o’-bis(2-aminopropyl) polypropylene glycol-block-polyethylene glycol-block-polypropylene glycol (Jeffamine) with 1,2,4,5-benzenetetracarbonyl tetrachloride (BTAC) (Fig.4). Due to the existence of many high polarity EO groups, the membrane had a super high CO₂/N₂ selectivity of 429. However, the thick selective layer (> 1 μm) and moderate CO₂ permeance (150 GPU) limited its application. There was still a requirement to develop an ultra-thin selective layer to increase the permeance.

Fig.4 Scheme of the PI separation layer fabricated by the IP process (PEI: polyetherimide). Reprinted with permission from Ref. [43], copyright 2024, Elsevier.

Full size|PPT slide

Except for selecting/synthesizing new monomers, the combination of different monomers is also a feasible strategy to enhance the CO₂ separation performance. For example, based on Yu’s work, Li et al. [37] prepared an IP membrane containing tertiary amine groups and EO groups by adding DNMDAM and DGBAME into aqueous phase. Compared with the membrane only using DNMDAM as the monomer, it showed better performance with a CO₂ permeance of 1612 GPU and a CO₂/N₂ selectivity of 138. In another work of this group, a novel membrane was prepared by mixing DAMBS and DGBAME in aqueous phase [38]. The carboxylate group in DAMBS and the EO group in DGBAME improved the separation performance of CO₂ together. Carboxylate groups also provided excellent antioxidizability. The as-prepared membrane displayed a high CO₂ separation performance with a CO₂ permeance of 5831 GPU and a CO₂/N₂ selectivity of 86, showing great potential in CO₂ separation from flue gas.

2.2 Hydroxyl monomers for membranes synthesized

As promising alternative monomers, hydroxyl monomers, such as polyphenols, polyols, and polymers with higher molecular weights containing hydroxyl groups, have received increasing attention during the past decades [44]. Different with the PA (–CONH–) chains, the ester bonds (–COO–) in the polyester molecule usually contribute to a low hydrophilicity, thus showcasing inefficiency for CO₂ capture and separation when compared to PA membrane. Notwithstanding, the vast majority of hydroxyl monomers with non-planar rigid structures or internal cavity structures have endowed the membrane with interconnected micropore and high free volume, still enabling its applicability in some certain circumstances (chemical corrosion resistance, high-pressure stability, etc.) [26]. Moreover, the ether-oxygen bond formed by cross-linking can also enhance the membrane performance for CO₂ separation [45]. Tab.2 summarizes these monomers and their membrane performances.

Tab.2 Membranes synthesized by hydroxyl monomers and their performancea)

Support	Aqueous phase monomers	Organic phase monomers	CO2 permeance/GPU	CO2/N2 selectivity(gas ratio)	CO2/CH4 selectivity(gas ratio)	Ref.
PSf/PDMS	MEDAC5H13NO2, MW: 119.16	TMC	2905	64 (15/85)	–	[ 46]
PSf/PDMS	TTSBIC21H24O4, MW: 340.42	TMC	870	43 (15/85)	–	[ 47]
PES	β -CDC42H70O35, MW: 1134.98	TMC	200	10.53 (pure)	–	[ 48]
PES/N-GQD	β -CD	TMC	174.5	23.3 (50/50)	–	[ 49]
PSf/PDMS	TTSBI&MEDA	TMC	1800	370 (15/85)	–	[ 50]
PSf/PDMS	MED β CD&MPDC44H76N2O34, MW: 1177.07	TMC	180	40.5 (pure)	69 (pure)	[ 51]
PSf/PDMS	H β -CD&DNMDAMC45H76O36, MW: 1193.07	TMC	2792	171 (15/85)	–	[ 52]

a) MEDA: N -methyldiethanolamine; N-GQD: amino-functionalized graphene quantum dot; β -CD: β -cyclodextrin; TTSBI: 5,5′,6,6’-tetrahydroxy-3,3,3′3’-tetramethyl-1,1’-spiralbisindane; MED β CD: mono-(6-ethanediamine-6-deoxy)- β -cyclodextrin; H β -CD: hydroxypropyl- β -CD.

The non-planar rigid monomers can be cross-linked with acyl chloride to form a selective layer with sub-nanometer scale micropores. Such a microporous structure not only has an insignificant effect on the transport of smaller gases, but also can prevent the penetration of larger molecules, which effectively enhances the performance of the membrane via size sieve effect. By using TTSBI as aqueous phase monomer, Yu et al. [47] synthesized a novel ultra-microporous IP membrane (Fig.5(a)). The obtained membrane showed a CO₂ permeance of 870 GPU and a CO₂/N₂ selectivity of 43. Ma et al. [50] prepared an IP membrane by mixing TTSBI and MEDA as aqueous phase monomers. The addition of long-chain MEDA can enhance the flexibility of the polymer chain, thereby providing more transport paths for CO₂. The optimal IP membrane exhibited a CO₂ permeance of about 1800 GPU and a CO₂/N₂ selectivity of 370. Meanwhile, carboxymethyl chitosan (CMC) also can be used as aqueous phase monomer. An IP membrane with excellent performance was prepared by blending PIP and CMC [53], and different CMC loading showed different Turing nanostructures. Except the formation of a loose selective layer, the CMC monomer can introduce more amine and carboxyl groups, thus facilitating the CO₂ separation.

Fig.5 IP membranes synthesized by TMC with hydroxyl monomers. (a) TTSBI. Reprinted with permission from Ref. [47], copyright 2019, Elsevier. (b) CDs. Reprinted with permission from Ref. [54], copyright 2018, Wiley-Blackwell. (c) MEDβCD. Reprinted with permission from Ref. [51], copyright 2023, Elsevier.

Full size|PPT slide

In addition to the microporous structure formed, the use of macrocyclic monomers with inner cavities can also provide mass transfer channels, thereby enhancing the separation performance of the membrane. Cyclodextrin is a low-cost and nontoxic macromolecule with a unique hollow truncated cone structure that is a unique molecule well-suited for gas separation (Fig.5(b)) [54]. Sun et al. [48] and Niu et al. [49] both prepared IP membranes with circular micropores by cross-linking β-CD and TMC. The resultant membranes showed a CO₂ permeance of 200 and 174.5 GPU, while the CO₂/N₂ selectivity were 10.53 and 23.3, respectively. Accurate CO₂ separation was not achieved due to the uncertain arrangement of β-CDs in the selective layer [44]. More interestingly, membranes prepared with the same monomers exhibited different membrane performances, which could be caused by the different IP interfaces. Niu’s group used N-GQDs as the interlayer, in which the CO₂-philic groups promoted the penetration of CO₂ and improved the selectivity. Some researchers also obtained good results by modifying β-CD and mixing it with an amine monomer. Li et al. [51] prepared an IP membrane for CO₂ separation by blending MPD and MEDβCD (Fig.5(c)). And the introduction of CO₂-philic groups in the modified β-CD can improve the CO₂ separation performance. Li et al. [52] prepared an IP membrane by mixing Hβ-CD and DNMDAM. The inner cavities of Hβ-CD and its synergistic effect with tertiary amine groups endow the membranes with a high performance with a CO₂ permeance of 2792 GPU and a CO₂/N₂ selectivity of 171 (Fig.6). However, the effect of low water solubility and diffusivity of hydroxyl monomer as well as its low polymerization reactivity cannot be ignored when preparing such a CO₂ separation membrane [55].

Fig.6 Gas transport mechanisms in Hβ-CD membrane. Reprinted with permission from Ref. [52], copyright 2023, Elsevier.

Full size|PPT slide

Although less popular than amine monomer, hydroxyl monomer also has great application potentials. In practical applications, the appropriate aqueous monomers should be selected based on specific separation requirements and conditions. Nonetheless, the specific application scenarios still need to be evaluated according to the actual situation, including the composition of gas mixture, operating conditions, and separation requirements. It is usually necessary to comprehensively consider membrane performance, reliability, cost, and other factors, meanwhile, performing sufficient experiments and tests before choosing CO₂ separation membrane.

3 IP of TFN membranes for CO₂ separation

With the progress of advanced materials, nanoparticles have attracted extensive attention in membrane separation technology [56]. Generally speaking, doping nanoparticles into the skin layer of IP membranes can maximize their potential [57], and a small amount of nanoparticles can significantly enhance the membrane separation performance. Extensive studies have been conducted to prepare high-performance nanofilms by incorporating diverse nanoparticles. In this section, the commonly used nanoparticles are reviewed according to their porosity, i.e., non-porous and porous nanoparticles. Meanwhile, how to alleviate the compatibility between polymer matrix and nanoparticles is also discussed. Tab.3 summarizes the nanoparticles utilized in preparing IP membranes for CO₂ separation along with their membrane performances.

Tab.3 Commonly used nanoparticles in IP membranes for CO2 separation and their membrane performancea)

Support	Aqueous phase monomers	Organic phase monomers	Nanoparticles	CO2 permeance/GPU	CO2/N2 selectivity (gas ratio)	CO2/CH4 selectivity (gas ratio)	Ref.
PSf	DNMDAM	TMC	Silica	59.4	85.4 (20/80)	–	[58]
PSf/PDMS	DGBAME	TMC	PMMA-MWCNT	70.54	67.18 (pure)	29.03 (pure)	[59]
PSf	DABA	TMC	C-MWCNTs	21.02	24.74 (pure)	18.45 (pure)	[60]
PSf	DNMDAM&DGBAME	TMC	CNT&GO	66.3	47.1 (pure)	36.5 (pure)	[57]
PSf	PIP	IPC	CDC	2.32	–	24.08 (pure)	[61]
PSf	DNMDAM&TOTDAM	TMC	BN	46.04	43.61 (pure)	–	[62]
PSf	PEA	TMC	PG	70	130 (pure)	–	[63]
PSf	PEA	TMC	MMT	95.03	37 (pure)	–	[64]
PES	Polyethyleneimine	TMC	mGO	73	60 (pure)	–	[65]
PSf	TETA	TMC	UiO-66-NH2	27.1	–	58.3 (30/70)	[66]
PSf	TETA	TMC	Zn ion	128.5	106.7 (50/50)	55.2 (50/50)	[67]
PSf/PDMS	DNMDAM	TMC	ZIF-8	2740	104 (15/85)	–	[68]
PSf/PDMS	DNMDAM	TMC	NH2-ZIF-8	1572	230 (15/85)	–	[69]
PSf/PDMS	PIP	TMC	TpPa-1	854	148 (15/85)	–	[70]

a) DABA: diaminobenzoic acid; PMMA-MWCNT: polymethyl methacrylate grafted MWCNT; C-MWCNTs: carboxylated multi-walled carbon nanotubes; CNT: carbon nanotube; GO: graphene oxide; CDC: carbide-derived-carbon; BN: boron nitride; PG: porous graphene; MMT: Na-montmorillonite; mGO: modified GO; UiO-66-NH2: zirconium elemental organic framework.

3.1 The applications of nanoparticles in IP membranes for CO₂ separation

The embedding of nanoparticles into the selective layer can create additional paths in the membrane and lead to an improved membrane separation performance. For example, the honeycomb structure of BN can be used as a perfect barrier for molecular diffusion [71]. By utilizing this barrier property, the performances of the BN-embedded membrane were successfully enhanced [62] (Fig.7). Also, the doped BN reduced the cross-linking degree of the selective layer, thereby increasing the CO₂ permeance. Compared with pure membrane, the BN-added membrane demonstrated a 37% and 20% augment in CO₂ permeance and selectivity, respectively. Furthermore, the potential application of CDC in gas separation also has gained significant attention. The involved CDC (pore size (> 1.5 nm) hardly used for gas separation) reduced the PA matrix cross-linking degree and thickness and made the selective layer have a higher free volume to transport gas. The selective layer doped CDC enhanced membrane performance, which showed that the CO₂ permeance and selectivity were improved by 88% and 49% compared with the pure membrane [61].

Fig.7 Separation mechanism of the IP membranes containing BN. Reprinted with permission from Ref. [62], copyright 2021, Academic Press Inc.

Full size|PPT slide

In addition to add additional transport paths by free voids, porous nanoparticles with finely controlled pore structure can provide more transport channels for CO₂. Yu et al. [58] doped commercial silica into IP membrane for CO₂ separation. The excellent CO₂ adsorption capacity and CO₂/N₂ adsorption ratio of silica improved the CO₂ separation performance. Zhang et al. [64] and Li et al. [63] embedded clay minerals and modified PG nanosheets, respectively, as fillers to prepare CO₂ separation membranes (Fig.8(a)). The enhancement of CO₂ separation performance could be due to the formed molecular sieving paths from doped nanosheets in the selective layer. Choi et al. [60] developed a thin selective layer with C-MWCNTs (Fig.8(b)). The membranes doped with 0.04% C-MWCNTs exhibited a CO₂ permeance of ~20 GPU and a CO₂/CH₄ selectivity of 24.74. In recent years, MOFs [72−74] and COFs [75,76] have been increasingly used as fillers to improve the separation performance of membranes due to their advantages in large specific surface area, high porosity, and tunable pore size. Jiao et al. [66] embedded MOF, UiO-66-NH₂, into the selective layer. The CO₂ permeation path was significantly increased by utilizing the 3D structure of added UiO-66-NH₂. In addition, the strong interaction between Zr center of UiO-66-NH₂ and amine groups of amine monomer can affect the diffusion of monomer, thereby tailoring the specific surface morphology (Fig.8(c)). Xu et al. [70] doped amine-rich COFs, TpPa-1, into PA selective layer. It disturbed the structure of the selective layer and provided a fast CO₂ transmission channel. Moreover, the introduction of TpPa-1 increased the number of amine groups in the membrane, thus enhancing the promotion of CO₂ transport. Consequently, the optimal membrane showed excellent performance with a CO₂ permeance of 854 GPU and a CO₂/N₂ selectivity of 148.

Fig.8 Schematic of fabrication process of TFN membranes by embedding (a) PG. Reprinted with permission from Ref. [63], copyright 2017, Elsevier. (b) C-MWCNTs. Reprinted with permission from Ref. [60], copyright 2021, Elsevier. (c) UiO-66-NH2. Reprinted with permission from Ref. [66], copyright 2021, American Chemical Society.

Full size|PPT slide

3.2 Modification of nanoparticles to relieve filler-polymer compatibility

Filler-polymer compatibility is also a key topic that should be considered for the application of nanoparticles. Ideally, the advantages of the nanoparticles and the polymer matrix can be fully exploited in the preparation of membranes due to the perfect compatibility between the nanoparticles and polymer matrix. However, in the actual scenario, because of the large differences in chemical composition and polarity, the doped nanoparticles may agglomerate in the polymer matrix or form interface defects, which compromise the performance of the membrane.

To approach the ideal state, researchers have made great efforts to improve the filler-polymer compatibility. Amination or grafting functional chains is a widely used method to modify nanoparticles. This method can enhance the compatibility of nanoparticles by creating covalent bonds between the polymer matrix and nanoparticles. For example, Yu et al. [69] modified the ZIF-8 by involving amine groups to synthesize NH₂-ZIF-8 for enhancing ZIF-8 compatibility in PA layer (Fig.9(a)). The membrane showed a high CO₂ permeance and selectivity of 1572 GPU and 230, respectively. Elsewhere, to achieve good dispersion between CNT and polymer matrix, Wong et al. [59] successfully incorporated PMMA-MWCNTs into the selective layer. The PMMA grafting ensured the desired MWCNTs dispersion and good filler-polymer compatibility. Compared to the pure membrane, the membrane doped with PMMA-MWCNTs showed that the CO₂ permeance and selectivity were enhanced by 29% and 47% compared to the neat membrane. In addition, the use of nanofillers that are inherently compatible with the polymer matrix is also a strategy. For example, amine-rich TpPa-1 could exhibit high compatibility by establishing covalent bonds with the polymer matrix (Fig.9(b)).

Fig.9 Schematic illustration of covalent bonds between PA and (a) NH2-ZIF-8. Reprinted with permission from Ref. [69], copyright 2017, Elsevier. (b) TpPa-1. Reprinted with permission from Ref. [70], copyright 2022, Chemical Industry Press. (c) The process of forming IP membranes via the swelling-controlled nanofiller positioning method. Reprinted with permission from Ref. [68], copyright 2021, Elsevier. (d) The homogeneous distribution of GO/CNT within the PA layer, Reprinted with permission from Ref. [57], copyright 2019, Elsevier.

Full size|PPT slide

Besides the general treatment methods, Li et al. [68] adopted a method called swelling-controlled nanofiller positioning to improve the compatibility of ZIF-8. The ZIF-8-doped organic phase was first incorporated into the PSf/PDMS substrate, followed by the pre-positioning of ZIF-8 via the swelling of PDMS in the organic solvent. In addition, the inhibition of ZIF-8 fluidity and the improvement of the interfacial compatibility between ZIF-8 and PDMS can be achieved by the post-crosslinking reaction of unreacted PDMS oligomer and the swelling of PDMS segments (Fig.9(c)). The CO₂ permeance and CO₂/N₂ selectivity of the obtained membrane were 2740 GPU and 104, respectively. Wong et al. [57] improved the dispersion of CNTs in IP membranes by adding another amphiphilic GO nanosheets. As a superior dispersant for nanotubes, GO nanosheets doped in aqueous solution effectively inhibit nanotube aggregation. The tendency of CNTs agglomeration was also decreased by depositing nanotubes on GO (Fig.9(d)). Incorporating amine acid-modified CNTs and GO together can improve the CO₂ permeance and selectivity by approximately 30% and 60%, respectively. Furthermore, the synergistic effects of doping these two carbon-based nanomaterials with distinct geometries offered additional flexibility in controlling the formation of selective layers as compared to single-filler incorporation. The size of nanoparticles also needs attention. Appropriate particle size can create a thin selective layer, and the pores of nanoparticles can serve as gas transport channels [77]. Too large or too small nanoparticles may form a thick selective layer, thus reducing the penetration of CO₂.

4 Optimizing IP process for CO₂ separation

Although the selection of monomers or nanoparticles is vital to the CO₂ separation performance of IP membranes, it’s also important to consider IP reaction parameters on the performance of IP membranes. This section focuses on the process optimization methods (e.g., interface, solvents, additives) aimed at further improving the CO₂ separation performance of IP membranes.

4.1 Reaction interface

4.1.1 Substrates

Currently, PSf and PES are the most commonly used substrate materials for preparing composite membranes. The substrates, as a physical template for IP membrane growth, not only support the generation of selective layers and provide sufficient mechanical strength, but also play a role in storing and transporting monomers. The performance of IP membranes is greatly influenced by their pore structure parameters. The substrate with a larger pore size inevitably allows the selective layer to permeate, which may form larger defects and compromise the selectivity. And it stores more residual aqueous solution, which leads to hydrolysis of acyl chloride groups and destroys the cross-linking degree of the formed PA layer, thus damaging the selectivity of IP membrane [78]. In addition, the combination between the pore wall and amine monomer in the substrate with small pore size is very strong, which inhibits the migration of amine monomer, forming a defective selective layer and damaging the membrane performance [79]. And the smaller pore sizes are not conducive to the penetration of separated substances. For substances passing through the selective layer, molecules in the non-porous region of the substrate surface must diffuse an additional distance to reach the porous region and pass through the substrate. Therefore, higher porosity can significantly shorten the transport path and reduce the transmembrane resistance, thereby improving membrane permeance [80]. And higher porosity is beneficial for storing more aqueous phase monomers, which makes the IP reaction more vigorous, forming a relatively thin separation layer and improving the permeance of the membrane [81]. In addition, the hydrophilicity of the membrane surface also affects membrane performance. The hydrophilic substrate can adsorb amine monomer more uniformly, and its strong affinity with amine monomer further limits the IP reaction area, which helps to form a thin and dense selective layer, thus improving the membrane separation performance. IP membranes prepared with more hydrophilic substrates show enhanced CO₂ separation performance [65]. In the later research, pore size, porosity and hydrophilicity of the substrate should be considered together to meet the requirements of CO₂ separation in further development.

Besides PSf and PES, developing new materials that can be designed as the substrate of IP process is beneficial to the preparation of IP membranes with superior performances. Polymers with ease of processing, low cost and wide variety are still the first choice. Some new polymers (PEI, poly(ether ether ketone), polyester, polythiosemicarbazide, polypropylene, etc.) have been successfully used as substrates to adapt to the harsh conditions [82]. Similarly, (nano)fibrous materials (polyacrylonitrile nanofibrous, aramid fibers, etc.) that can reduce the concentration polarization of membranes, and inorganic materials (alumina, etc.) with strong mechanical properties are also optional substrate materials [82]. However, the suitability of these substrates for CO₂ separation needs to be further explored.

4.1.2 Interlayer

Different from controlling the substrates, the introduction of interlayer in membrane design and synthesis [83] is considered to be an effective method to manipulate IP reaction and improve the structure of selective layers, thereby enhancing the CO₂ separation capacity. Substrates with an interlayer usually have smoother surface, smaller pore size and higher porosity, which are beneficial to IP reactions [84]. The porous and smooth structure is conducive to the distribution of the homogeneous monomers on the interlayer, thus forming uniform, defect-free nanofilms [85,86]. Also, the strategy of constructing interlayers has the added benefit of preventing pore penetration, which further prevents the intrusion of nanofilms into the substrate, minimizing the transport resistance significantly [87,88]. The key is to select appropriate interlayer materials.

Currently, PDMS is widely used as an interlayer and has been applied at the industrial scale [89]. Salih et al. [32] showed that the PDMS interlayer effectively reduced the defects of the nanofilm and inhibited the hydrolysis of TMC, thereby improving the membrane performance for CO₂ separation. However, the influence of high diffusion resistance and aging of PDMS on the performance of the membrane cannot be ignored. In recent years, nanomaterials as interlayers have gained more attention due to the effective regulation of the selective layer structure and membrane performance. Niu et al. [49] successfully prepared IP membranes using graphene quantum dots and N-GQDs as interlayers, respectively (Fig.10(a)). The addition of N-GQDs interlayer reduced the diffusion of β-CDs, improved the cross-linking degree of polyester membrane and improved the membrane separation performance. Meanwhile, the N-GQDs concentrates CO₂-philic groups, such as –OH and –NH₂, which can better adsorb CO₂ to achieve the purpose of improving the CO₂ selectivity. Ma et al. [90] chose ultra-thin two-dimensional (2D) MOF nanosheets as the interlayer of IP membrane (Fig.10(b)). The interlayer of 2D MOF nanosheets significantly reduced the gas transport resistance and improved the gas permeance. The instinct chemical and mechanical stability of MOFs also improved the gas separation stability of the IP membrane.

Fig.10 Process of preparing interlayer in IP process with (a) GQDs/N-GQDs. Reprinted with permission from Ref. [49], copyright 2022, Elsevier. (b) 2D MOF (Zn2(bim)4) nanosheets. Reprinted with permission from Ref. [90], copyright 2021, Elsevier.

Full size|PPT slide

4.2 Solvents

4.2.1 Aqueous phase solvents

Currently, deionized water is widely used as an aqueous phase solvent for CO₂ separation membrane fabrication. The application of other aqueous phase solvents like alkanes or ionic liquids (ILs), overcome the limitations of water-soluble monomers while avoiding side reactions caused by water, providing a new way and a broad platform for designing advanced IP membranes. The development of more varieties of immiscible liquid/liquid interfaces through the use of more readily available organic solvents is highly desirable for designing IP membranes with improved chemical stability. For example, when dimethylacetamide is used as aqueous solvent, it expanded the water-insoluble aromatic amine monomer as the monomer of the IP process. And contrary to the diffusion direction during the traditional IP process, TMC diffuses from hexane to dimethylsulfoxide to trigger the IP reaction. Finally, a smooth and dense selective layer was formed, which further improved the performance of the membrane [91]. In addition, ILs can be used as a universal solvent for various water-insoluble amine monomers, thereby greatly broadening the scope of aqueous phase monomers [92]. The use of ILs can also avoid the side reaction caused by water and produce a defect-free PA layer with a high degree of cross-linking, effectively improving the membrane separation performance [93]. ILs are considered to be a promising candidate for green solvents [94]. In addition to directly replace the aqueous solution with ILs as a solvent, it is also feasible to add a certain amount of ILs into aqueous solution. For example, Ma et al. [95] improved the solubility of water-insoluble hydroxyl monomers by adding a certain amount of ILs, and the ILs also limited the diffusion rate of the monomers, thus forming a thin selective layer, and finally enhancing the membrane separation performance.

4.2.2 Organic phase solvents

The organic phase solvents used in IP process typically select high aqueous monomer solution diffusion coefficient and low aqueous monomer solution coefficient solvents such as nonpolar solvents n-hexane and n-heptane. When different nonpolar solvents are used as organic phase solvents, Yuan et al. [46] found that n-hexane possesses higher amine aqueous monomer solubility and diffusivity than other organic solvents and can also endow the IP membrane a unique resistance to CO₂-induced plasticization. Therefore, n-hexane is currently often used as an organic phase solvent to prepare IP membranes for CO₂ separation.

Adding some polar co-solvents to the organic phase can also further regulate the monomer diffusion. Wang et al. [96] used weakly polar ethyl formate as a co-solvent to adjust IP process. MD simulation showed that ethyl formate significantly reduced the interfacial tension and promoted the rapid diffusion of amine monomers to the organic phase, which was helpful to design PA membranes with a higher cross-linking degree, thus effectively improving membrane performance. However, the study on the enhancement of membrane performance by co-solvent in CO₂ separation IP membranes is limited and needs to be further developed.

4.3 Additives

When preparing the IP membranes, the use of additives to the aqueous/organic phase is a common approach and has been widely utilized in industrialization [97]. These reaction-controlling additives, such as Na₂CO₃ and NaHCO₃, help to remove accumulated HCl and promote cross-linking of the selective layer. In addition, they can produce CO₂ nanobubbles that lead to interfacial instability, resulting in a rougher skin layer that favors improved CO₂ permeance [98]. In addition to participating in polycondensation reactions, some additives (usually surfactants) can also indirectly affect IP reactions through the control of diffusion processes [51]. These surfactants improve the storage of aqueous monomers and accelerate the IP reaction, which also reduces the surface tension at the phase interface and the mass transfer resistance of aqueous monomers diffusing into the organic phase, favoring the formation of thinner and more uniform selective layers and thus effectively improving membrane performance. Currently, Na₂CO₃ and NaHCO₃ are the most common additives to produce IP membranes for CO₂ separation. It should be noted that there is still a lack of clear and unified analysis and explanation of the mechanism of action of additives on IP membranes used for CO₂ capture. In addition, there are relatively few adjustment additives in the IP process compared to other polymer material manufacturing processes.

5 The potential of MD simulation and ML in IP membrane for CO₂ separation

Up to now, the design of new IP membranes is still dominated by the trial-and-error process that is based on intuition and experience [99]. Therefore, some excellent materials may be overlooked since the synthesis of new materials and subsequent performance testing is a laborious, time-consuming, and incomplete process [100]. Recently, MD simulation and ML have been applied in the membrane design and separation mechanism study, as they are powerful and disruptive for designing a high-performance membrane.

MD simulation studies the changes of behavior and properties of the system at different time scales according to the interaction between particles and Newton’s law of motion. It usually estimates the reaction conditions by establishing a model to guide the actual experimental process. It enables an intuitive revelation of mechanisms and laws underlying experimental phenomena that are difficult to obtain directly. This significantly reduces exploration time, and enhances research efficiency, cost-effectiveness, and predictability, while providing valuable guidance for membrane design [101]. For example, through MD simulation, Song et al. [102] developed a model of PA membranes to study the influence of different MPD/TMC ratios on the membrane in IP process. LAMMPS was used to cross-link MPD and TMC in a simulation box and optimized the descriptors of OPLS-AA force field. Then the initial system was homogenized in the NVT ensemble and finally the polymer was relaxed and balanced using simulations. Through the analysis, it can be found that different monomer ratios (0.25–5) significantly affected the synthesized selective layer, in which the cross-linking degree and unreacted groups were positively related to the monomer ratios. And the membrane simulated by the intermediate monomer ratio showed a similarity built by the experimental synthesis. In another study, through MD simulations, the influence of diamines with different alkyl chain lengths (PIP, DAPP, 1-(2-aminoethyl) piperazine and m-xylylene-diamine) on the performance of IP membranes was investigated [103]. The force field and charge were distributed by COMPASS II force field, and the model was optimized by Forcite module. Then, the membrane-ionized water system was constructed with amorphous cell module. The simulation results showed that the membrane synthesized by PIP (short alkyl chain) had a tight pore size of 1.1 nm and polar hydroxyl groups, which made it highly hydrophilic, so the membrane showed a high permeation flux and it also displayed a rejection of 91% for magnesium sulfates. And the simulation results were in excellent agreement with the experimental data.

As the most advanced branch of artificial intelligence, ML has been successfully applied to membrane separation domain. The key to applying ML to membrane separation is its inherent ability to handle massive and high-dimensional data, which is beyond human capability [104]. Through a suitable algorithm, ML can extract the potential trends and laws of a given data set, and establish a mathematical model based on this, so as to judge and predict the new data set. This means that advanced membrane development can be accelerated through appropriate ML models. Wang et al. [105] developed an ML model to design novel IP membranes for solvent recovery. First, the monomer structure, membrane, solvent and solute properties and operating conditions in the experimental data of 1347 organic solvent nanofiltration membranes were featured to train the ML model with gradient-enhanced regression algorithm. Then the solvent permanence and solute rejection of 167 new IP membranes designed from 40 monomers were predicted, and the most promising membrane was synthesized to weigh the accuracy of the model. The results showed that the ML model based on monomer structure had good accuracy and can be easily extended to the design of new membranes for gas separation. Also, Deng et al. [106] developed a novel multitask ML model (based on an artificial neural network) to guide the synthesis of IP membranes. Through the limited data sets collected and online learning involving expert experience, a good ML model is trained (Fig.11). Four kinds of membranes were synthesized under the excellent preparation conditions recommended by the model, which all showed excellent mono/divalent ion separation performance.

Fig.11 Schematic diagram for the model design, development, and application. (a) Network structure of the multitask multilayer perception, (b) online learning process with the introduction of expert experience, (c) illustration of the model training and evaluation, and (d) interpretation of model and guided design of membranes. Reprinted with permission from Ref. [106], copyright 2022, American Chemical Society.

Full size|PPT slide

It is not hard to find that MD simulation and ML have great potential in membrane fabrication, which can guide the design of membranes at the molecular level. Their advantages in monomer preparation, membrane fabrication conditions and performance prediction of IP membrane can be extended to the field of CO₂ separation, showing great attraction. However, in MD simulation, molecules are often simplified, which is far from an accurate analysis of the actual membrane separation process. And the selection of descriptors and establishment of molecular models are also difficult to overcome. In ML, the lack of data, suitable ML algorithms, and interpretability of AI models may hinder the application of ML in IP membranes for CO₂ separation. There is still a long way to expand the relevant data set to train more accurate ML models. Additionally, combining MD simulation or ML with experimental techniques allows for the full utilization of their strengths, accelerating the development of advanced membranes. This integration also enables researchers to focus on more profound questions and conduct better scientific research [100,107].

6 Conclusions and outlook

In a nutshell, IP process with its innate assets contributes it a top-notch candidate for CO₂ separation membrane composition, while still reaping the financial merits and perks of petrochemical resources. Throughout this review, we introduce systematically the most applied exploited monomers and nanoparticles and finely crafted strategies for IP parameters, targeting for an impeccable decarbonizing membrane. Furthermore, we innovatively explicated the idea of MD simulation and ML for screening promising membrane materials and predicting synergic membrane performance. By reviewing the literature, it has been found that the IP is a prevailing technology, permitting accelerated CO₂ separation membrane upscaling with lower reagents purity requirements and high-speed film-forming dynamics. The monomers varieties, improvement in nanofillers incorporation and simplified IP process expand IP prospective applications, expressly CO₂ capture and sequestration from ambient air or power plants in decarbonizing industries. IP process for CO₂ membrane preparation also enables technical economies of scale to tiny and movable emission sources, compensating for the inherent obstacles of carbon capture under weak energy infrastructures [108].

While the IP process suggests sky-high suitability, bargain-basement costs and accessibility of multifarious nanofillers for ultra-high perm-selective membrane fashioning, IP membranes have encouraging commercialization prospect for CO₂ capture. Nonetheless, in an effort to accelerate IP membranes for CO₂ separation practical applications in industry, there are still a few prominent impediments that need to be pressingly unraveled to ameliorate CO₂ separation membrane operation efficiency and durability.

The development of novel monomers is particularly important. Some monomers on a laboratory scale may be expensive. The low cost and high efficiency of monomers as well as their ability to achieve high performance in gas separation are prerequisites for large-scale production [23]. Due to the aggregation-prone nature of nanofillers, the existing methods employed to functionalize nanofillers cannot adequately overcome the embarrassment in the aggregation of nanofillers. Moreover, nanofiller’s intrinsic characteristics, such as high aspect ratio and surface area to volume ratio, temporal instability of nanofiller suspension, and lower threshold in nanofiller embedment, may block the high nanofiller loading, thereby restraining the potentials of large-scale TFN manufacturing. In the future, the vigorous expansion of in situ nanofiller synthesis technique to flexibly tune filler size and organization can offer a flexible tool to controllably improve compatibility between polymer bulk and nanofiller [56,109], expecting to breakthrough technological bottlenecks of TFN membrane for CO₂ separation.

To realize the industrialization of IP process in high-performance CO₂ separation membranes, more attention should be paid to the stability of long-term operation. TFC membrane with an ultra-thin separation layer is terribly vulnerable to CO₂-induced plasticization [110] and carrier saturation [111]. Albeit some polymers with stiff chains and compatible inorganic materials were invented and blended with polymer bulk to cater the negative impacts of plasticization, high CO₂-philic materials still need to be introduced side-by-side to compensate for the dropped gas diffusion capacity originating from the rigidification of membrane matrix. Thus, researchers are constantly pursuing much more CO₂-philic materials such as ILs as a fairly enticing candidate for incredibly high CO₂-sorption membrane construction.

Large-area membrane fabrication is requisite for accomplishing the real potentiality of such a membrane for realistic demands. But the fabrication of large-scale membrane modules and defect-free membrane structures are exactly challenging [112]. Healing techniques based on coating with elastomers is possible to be an interesting solution for such a thorny problem. Further, the existing CO₂ separation membranes derived from IP procedure are flat-sheet structures. Hollow fiber membranes with less fouling propensity and larger effective infiltration area could contribute higher productivity [113]; nonetheless, the designation of large-area hollow fiber modules is another plague issue and deserves further exploration. During the hollow fiber membrane commercialization and industrial consolidation, some new theories have been proposed and likely receive thorough investigation in the future. For example, the counter cations on defects of TFC selective layer plausible impede CO₂ transmission, showcasing tremendous prospects for CO₂ removal process; secondary IP reaction between functionalized nanofillers and reactive monomers may greatly increase the nanofiller loading [114], which provides a tried-and-true routine for CO₂ separation membrane formation with high achievements.

In practical application, flue gas usually contains impurity gases, such as SO₂, NO_x, O₂, H₂S, and H₂O. The existence of these impurities usually leads to the performance degradation of IP membranes [115]. In particular, SO₂ is more acidic and condensable than CO₂, and the competitive adsorption between SO₂ and CO₂ can compromise the transport of CO₂ through the membranes. Meanwhile, these impurities can also affect the service life and long-term stability of the membranes. But it should be noted that laboratory tests usually involve single gas or CO₂/N₂ simulated mixed gas, and the CO₂ separation performance of the obtained membrane may be still far from the standard of practical application. In addition, technical and economic feasibility analysis should be carried out through process simulation and cost estimation to determine the best operating conditions for future work [116]. To truly realize the industrialization of IP membrane for CO₂ separation, it is necessary to develop together with other fields.

Finally, it cannot be ignored that IP membranes used for water treatment and gas separation have different mechanisms. Compared with water treatment membranes, IP membranes used for CO₂ separation require smaller pore size and sharper pore size distribution to achieve superior separation performance. In addition, the Donnan effect caused by membrane charge is rarely considered in CO₂ separation, and the efficient separation of CO₂ is mainly achieved by introducing some promoting groups. In large-scale applications, the acid and base stability and the fouling resistance of water treatment membrane should be considered [26], and CO₂-induced plasticization and competitive adsorption of some impurity gases are the main concerns of CO₂ separation. When designing a new IP membrane for CO₂ separation, these differences should be considered and identified.

References

[1] Huang Y L , Ye Z L , Wan X K , Yao G , Duan J Y , Liu J J , Yao M D , Sun X , Deng Z X , Shen K . . Systematic mining and evaluation of the sesquiterpene skeletons as high energy aviation fuel molecules. Advanced Science, 2023, 10(23): 2300889

[2] Nie J R , Jia T H , Pan L , Zhang X W , Zou J J . Development of high-energy-density liquid aerospace fuel: a perspective. Transactions of Tianjin University, 2022, 28(1): 1–5

[3] Tang P F , Yang B , Li R , Wang Y C , Li X D , Yang G C . Ti₃C₂ MXene: a reactive combustion catalyst for efficient burning rate control of ammonium perchlorate based solid propellant. Carbon, 2022, 186: 678–687

[4] Cao J W , Zhang Y C , Pan L , Shi C X , Zhang X W , Zou J J . Synthesis and characterization of gelled high-density fuels with low-molecular mass gellant. Propellants Explosives Pyrotechnics, 2020, 45(7): 1018–1026

[5] Lysien K , Stolarczyk A , Jarosz T . Solid propellant formulations: a review of recent progress and utilized components. Materials, 2021, 14(21): 6657

[6] Shorunov S V , Zarezin D P , Samoilov V O , Rudakova M A , Borisov R S , Maximov A L , Bermeshev M V . Synthesis and properties of high-energy-density hydrocarbons based on 5-vinyl-2-norbornene. Fuel, 2021, 283: 118935

[7] Moghaddam A S , Rezaei M R , Tavangar S . Experimental investigation of characteristic length influence on a combustion chamber performance with liquid and gelled UDMH/IRFNA bi-propellants. Propellants Explosives Pyrotechnics, 2019, 44(9): 1154–1159

[8] ZouJ JZhangX WPanL. High-Energy-Density Fuels for Advanced Propulsion: Design and Synthesis. Hoboken, New Jersey: Wiley, 2021

[9] Jyoti B V S , Baek S W . Rheological characterization of ethanolamine gel propellants. Journal of Energetic Materials, 2016, 34(3): 260–278

[10] Liu Y , Zhang H Z , Pan L , Xue K , Zhang X W , Zou J J . High-energy-density gelled fuels with high stability and shear thinning performance. Chinese Journal of Chemical Engineering, 2022, 43: 99–109

[11] Natan B , Hasan D . Advances in gel propulsion. Journal of Energetic Materials, 2019, 18(4): 303–323

[12] CaoJ WPanLZhangX WZouJ J. Physicochemical and rheological properties of Al/JP-10 gelled fuel. Chinese Journal of Energetic Materials, 2020, 28(5): 382–390 (in Chinese)

[13] Xue K , Cao J W , Pan L , Zhang X W , Zou J J . Review on design, preparation and performance characterization of gelled fuels for advanced propulsion. Frontiers of Chemical Science and Engineering, 2022, 16(6): 819–837

[14] Han L K , Wang R D , Chen W Y , Wang Z , Zhu X Y , Huang T Z . Preparation and combustion mechanism of boron-based high-energy fuels. Catalysts, 2023, 13(2): 378

[15] Glushkov D O , Paushkina K K , Pleshko A O , Yanovsky A O . Ignition and combustion behavior of gel fuel particles with metal and non-metal additives. Acta Astronautica, 2023, 202: 637–652

[16] Yang D L , Xia Z X , Huang L Y , Ma L K , Chen B B , Feng Y C . Synthesis of metallized kerosene gel and its characterization for propulsion applications. Fuel, 2020, 262: 116684

[17] Saberimoghaddam A , Emamifard Z , Mahdi Bahri Rasht Abadi M , Meyghani N . Investigation of the effective parameters on the preparation of the gelled IRFNA. Propellants Explosives Pyrotechnics, 2019, 44(12): 1621–1627

[18] Rahimi S , Hasan D , Peretz A . Development of laboratory-scale gel-propulsion technology. Journal of Propulsion and Power, 2004, 20(1): 93–100

[19] Wang F S , Chen J , Zhang T , Guan H S , Li H M . Experimental study on spray characteristics of ADN/water-based gel propellant with impinging jet injectors. Propellants, Explosives, Pyrotechnics, 2020, 45(9): 1357–1365

[20] Padwal M B , Natan B , Mishra D P . Gel propellants. Progress in Energy and Combustion Science, 2021, 83: 100885

[21] Guo X D , Li F S , Bian G Z , Liu G P . Coating treatment of Mg powders and their water reaction characteristics. Advanced Materials Research, 2013, 781–784: 2463–2470

[22] Bergthorson J M , Yavor Y , Palecka J , Georges W , Soo M , Vickery J , Goroshin S , Frost D L , Higgins A J . Metal-water combustion for clean propulsion and power generation. Applied Energy, 2017, 186: 13–27

[23] GhedjattiIYuanS WWangH X. Energy generation from metal-water reaction for power systems, underwater and aerospace propulsion applications. In: 2019 16th International Bhurban Conference on Applied Sciences and Technology. New York: IEEE, 2019, 49–54

[24] NishiiKMannamiYAkiyamaMMuroharaMHiroyukiKKomurasakiK. Experimental study on bulk metal-water combustion for small spacecraft propulsion. In: AIAA Propulsion and Energy 2020 Forum. Reston, Virginia: AIAA, 2020, 3744

[25] Boryaev A A . Calculation and experimental estimation of the efficiency of using lithium, sodium, magnesium, and aluminum as fuels in hydro-reactive propellants. Thermal Science and Engineering Progress, 2021, 23: 100881

[26] Zou M , Yang R , Guo X , Cao C , Li J . Advances in aluminum/water propellants. Chinese Journal of Energetic Materials, 2007, 15(4): 421–424 (in Chinese)

[27] Dong R K , Mei Z , Xu S Y , Zhao F Q , Ju X H , Ye C C . Molecular dynamics simulation on reaction and kinetics isotope effect of nano-aluminum and water. International Journal of Hydrogen Energy, 2019, 44(36): 19474–19483

[28] Murugesan R , Chakravarthy S R , Kandasamy J , Sarathi R . Experimental investigation on aluminum-based water ramjet for propelling high-speed underwater vehicles. Journal of Propulsion and Power, 2023, 39(6): 886–895

[29] Hahma A , Gany A , Palovuori K . Combustion of activated aluminum. Combustion and Flame, 2006, 145(3): 464–480

[30] Huang H T , Zou M S , Guo X Y , Yang R J , Li Y K . Analysis of the aluminum reaction efficiency in a hydro-reactive fuel propellant used for a water ramjet. Combustion, Explosion, and Shock Waves, 2013, 49(5): 541–547

[31] Gautham M G , Ramakrishna P A . Propulsive performance of mechanically activated aluminum-water gelled composite propellant. Journal of Propulsion and Power, 2020, 36(2): 294–301

[32] Risha G A , Connell T L Jr , Yetter R A , Sundaram D S , Yang V . Combustion of frozen nanoaluminum and water mixtures. Journal of Propulsion and Power, 2014, 30(1): 133–142

[33] Gautham M G , Ramakrishna P A . Combustion characteristics of aluminum-water gelled composite propellant. Journal of Propulsion and Power, 2018, 34(5): 1345–1354

[34] Guo C , Li T , Zhao Y , Bao S , Zhang H , Wu R . Aluminum/water reaction mechanism of aluminum-based hydrogels. Chinese Journal of Energetic Materials, 2022, 30(6): 557–563 (in Chinese)

[35] Nakayama A , Kakugo A , Gong J P , Osada Y , Takai M , Erata T , Kawano S . High mechanical strength double-network hydrogel with bacterial cellulose. Advanced Functional Materials, 2004, 14(11): 1124–1128

[36] Li L C , Zheng R L , Huang Y , Sun R Q . Self-sorting assembly in multicomponent self-assembled low molecular weight hydrogels. Progress in Chemistry, 2023, 35(2): 274–286

[37] Espinosa-Andrews H , Velasquez-Ordonez C , Cervantes-Uc J M , Rodriguez-Rodriguez R . Water behavior, thermal, structural, and viscoelastic properties of physically cross-linked chitosan hydrogels produced by NaHCO₃ as a crosslinking agent. Journal of Materials Science, 2023, 58(13): 6025–6037

[38] Yang H H , Zhao C C , Wang Y , Wang Y Y , Shi B F , Chen P F , Yan W . Progress in study on gel propellants and their rheological properties. Journal of Xi’an Jiaotong University, 2022, 56(5): 166–179 (in Chinese)

[39] Horinaka J , Ogawa S . Cyclic deformation behavior of agarose hydrogels prepared at different gelation concentrations. International Journal of Biological Macromolecules, 2023, 248: 125904

[40] Zheng J , Zhao C , Zhu L , Chen Q , Wang Q . One-pot synthesis of highly mechanical and recoverable double network hydrogels using thermoreversible sol–gel polysaccharide. Advanced Materials, 2013, 25(30): 4171–4176

[41] Chen A Q , Guan X D , Li X M , Zhang B H , Zhang B , Song J . Preparation and characterization of metalized JP-10 gel propellants with excellent thixotropic performance. Propellants Explosives Pyrotechnics, 2017, 42(9): 1007–1013

[42] Lin C , Li Y , Tang W , Zhou S , Rao X . Facile construction of bio-based supramolecular hydrogels from dehydroabietic acid with a tricyclic hydrophenanthrene skeleton and stabilized gel emulsions. Molecules, 2021, 26(21): 6526

[43] Sang Y , Liu M . Nanoarchitectonics through supramolecular gelation: formation and switching of diverse nanostructures. Molecular Systems Design & Engineering, 2019, 4(1): 11–28

[44] Zhang L M , Wu C X , Huang J Y , Peng X H , Chen P , Tang S Q . Synthesis and characterization of a degradable composite agarose/HA hydrogel. Carbohydrate Polymers, 2012, 88(4): 1445–1452

[45] Liao J , Wang Y J , Hou B , Zhang J M , Huang H H . Nano-chitin reinforced agarose hydrogels: effects of nano-chitin addition and acidic gas-phase coagulation. Carbohydrate Polymers, 2023, 313: 120902

[46] Wang S , Zhang R , Yang Y , Wu S , Cao Y , Lu A , Zhang L . Strength enhanced hydrogels constructed from agarose in alkali/urea aqueous solution and their application. Chemical Engineering Journal, 2018, 331: 177–184

[47] Sarkar D , Mohapatra D , Ray S , Bhattacharyya S , Adak S , Mitra N . Nanostructured Al₂O₃–ZrO₂ composite synthesized by sol–gel technique: powder processing and microstructure. Journal of Materials Science, 2007, 42(5): 1847–1855

[48] Jo H S , Kim H , Yoon S Y . Synthesis and characterization of mesoporous aluminum silicate and its adsorption for Pb(II) ions and methylene blue in aqueous solution. Materials, 2022, 15(10): 3562

[49] CaoQFengFWuX S. Time and temperature dependent constitutive equations modeling of RP-1 jet fuel gel. Chinese Journal of Energetic Materials, 2016, 24(6): 592–598 (in Chinese)

[50] Qiu X P , Pang A M , Jin F , Wei W , Chen K H , Lu T J . Preparation and characterization of JP-10 gel propellants with tris-urea low-molecular mass gelators. Propellants Explosives Pyrotechnics, 2016, 41(2): 212–216

[51] Hu Y , Kim Y , Hong I , Kim M , Jung S . Fabrication of flexible pH-responsive agarose/succinoglycan hydrogels for controlled drug release. Polymers, 2021, 13(13): 2049

[52] Jarosz A , Kapusta O , Gugała-Fekner D , Barczak M . Synthesis and characterization of agarose hydrogels for release of diclofenac sodium. Materials, 2023, 16(17): 6042

[53] Dennis J D , Kubal T D , Campanella O , Son S F , Pourpoint T L . Rheological characterization of monomethyl hydrazine gels. Journal of Propulsion and Power, 2013, 29(2): 313–320

Competing interests

The authors declare that they have no competing interests.

Acknowledgements

This work was supported by the Haihe Laboratory of Sustainable Chemical Transformations.

RIGHTS & PERMISSIONS

2024 Higher Education Press