Gas transport mechanisms through gas-permeable membranes in microfluidics: A perspective

Microfluidics has traditionally favored liquid-based devices, primarily due to the significant advantages offered by viscosity-dominant physics. The introduction of gases has often been avoided owing to the complexities arising from gas compression and inertia. Additionally, gas diffusion through channel walls presents unexpected challenges because of gas permeation. In contrast, the use of thin, gas-permeable membranes (GPMs) enables the controlled permeation of gases, facilitating rapid changes in the physiochemical conditions of microfluidic systems and guiding mass transport between gases and liquids. By incorporating gases, microfluidics moves beyond a mere liquid-based system and encompasses the complete essence of “fluid.” This Perspective outlines the diverse range of applications and advantages associated with the utilization of gases in microfluidics.

The integration of gases in microfluidics serves various purposes across different fields of study. In the realm of biology, for instance, the presence of oxygen is crucial in terms of nutrient supply for the culture of cells, tissues, and organs.^1,2 Consequently, on-chip methods for manipulating oxygen delivery and creating oxygen concentration gradients have been developed.^3–5 Another critical demand is the monitoring of oxygen concentration.^5,6 In the field of chemistry, gas–liquid reactions enable precise control over gas–liquid volumes, gas–liquid interface areas, flow rates, and bubble sizes.^7,8 Physics and engineering have harnessed the unique characteristics of bubbles, which exhibit robust responses to diverse stimuli.⁹ Furthermore, the study of two-phase flows yielded insights into the complexities of interfaces¹⁰ and nonlinear behavior.¹¹ Additionally, the innovative concept of “aerofluidics” emerged, enabling the underwater manipulation of gases.¹² Evaporation and condensation processes also garnered significant attention in the fields of nanofluidics,¹³ botany,¹⁴ and energy.¹⁵ Whether employed for fundamental research or practical applications, the key values of controllability and rapid response are inherent to microfluidics. GPMs effectively partition interfaces, enabling diffusion-based transport. Moreover, the emergence of GPMs with high transport rates and selective permeability offers ample opportunities for expanding the boundaries of microfluidic applications.

Gas transport mechanisms through GPMs can be categorized based on the direction and nature of the transported molecules, as illustrated in Fig. 1. A typical microfluidic device comprises a liquid channel, a gas channel, and a GPM that acts as a separator. In some cases, an environmental system can replace the gas channel, and a device wall made of gas-permeable material can be used in lieu of the GPM. In such setups, gas molecules either “dissolve” into the liquid channel or are “extracted” in the reverse direction, while those that dissolve slowly or do not dissolve lead to the formation and removal of “foam” bubbles at the liquid–GPM interface. Vapor molecules of the liquid exhibit behavior similar to that of gas molecules; they can evaporate or pervaporate through the GPM facilitating diffusive transport of vapor molecules. The introduction of vapor into the gas channel allows for “condensation” in a dry liquid channel by promoting the diffusion of vapor molecules. The specific types of transport are determined by various factors, including pressure differences, gas solubility, gas–GPM interactions, membrane roughness, humidity, and temperature. In this Perspective, we focus on the efficient utilization of gases in microfluidics, with a primary emphasis on microfluidic devices integrated with GPMs. After outlining various transport categories and their representative applications, we delve into the fundamental theories that explain these transport phenomena. We also conduct comparative analysis of GPMs based on their inherent properties and intended applications, highlighting the importance of emerging gas-selective GPM technologies in advancing microfluidics.

We classified the transport mechanisms into three distinct groups, taking into consideration the types of penetrant, namely, dissolving gases, bubbles, and vapors. Furthermore, each of these categories was further subdivided into two subcategories depending on the directionality between the liquid channel and the gas channel. In this section, we provided a concise overview of the associated phenomena, discussed methods for enhancing transport within microfluidic devices, and highlighted pertinent applications. Additionally, we included a summary of relevant publications since 2013 in Table I. Our classification is as follows:

When gas molecules diffuse from the gas channel, they can be absorbed by liquids and incorporated into the solution. The absorbed gases may either remain stable by retaining their molecular structure or dissociate into smaller parts. Unreactive gases, such as nitrogen (N₂), oxygen (O₂) in water, and noble gases, tend to remain stable once dissolved. However, when reactive gases, such as carbon dioxide (CO₂) or ammonia (NH₃), dissolve in water, they react with water molecules to produce ions. Generally, gases that dissociate have a greater capacity to dissolve compared to unreactive gases.⁴⁸ The diffusion rate and direction between the gas channel and the liquid channel can be modulated by controlling the concentration gradient along GPMs. Once gases are dissolved, they are contained in the liquid and carried by flow. Therefore, liquids can act as carriers, containers, and reactors for gas molecules. For example, blood oxygenators were developed by increasing the surface area for gas exchange with the aim of eventually meeting certain clinical criteria [Fig. 2(a)].²⁰ For practical utilization, it is necessary to simultaneously facilitate the dissolution of O₂ and the extraction of CO₂, surpassing respiratory consumption and production rates of these gases. Recently, an 8-layer oxygenator demonstrated in vivo animal testing for 24 h, maintaining a high oxygen transfer rate (27 ml of O₂/min) while meeting clinical criteria for pressure drops.⁵³ For a chemical reaction, tube-in-tube reactors represent a stepwise gas–liquid reaction in a microfluidic system [Fig. 2(b)].^7,49,54 A tube made of a gas-permeable polymer was inserted into another tube to enable controlled chemical reactions via effectively induced dissolution of gas. Additionally, gas mixtures were separated on a microfluidic device by utilizing GPMs with different gas permeabilities.¹⁸ Gas dissolution is frequently used to maintain and modulate chemical conditions in liquids. For example, the levels of O₂ and CO₂ were maintained in microfluidic devices during cell, tissue, and organ culture.^5,55 Furthermore, CO₂ and NH₃ gas channels were designed to modulate the pH of the liquid channel [Fig. 2(c) ].⁵⁰ These multilayer microfluidic channels enabled on-site sourcing of molecules while still confining the liquid within viscous-dominant microchannels. Concentration gradients of gases are more useful than constant concentration conditions to investigate the diverse responses at different conditions on a single device.⁵⁶ 

Conversely, dissolved gases sometimes need to be extracted from the liquid. This requires an opposite concentration gradient. Flushing gases can be delivered into the gas channel to achieve this; however, maintaining a steep concentration gradient can be challenging as gas extraction increases the concentration of the gas channel. Therefore, gas-absorbing materials can be utilized to achieve a high extraction rate. For instance, O₂ gas was absorbed by O₂ scavenger solution—pyrogallol/sodium hydroxide (NaOH) solution—via a polydimethylsiloxane (PDMS) wall and continuously consumed by a reaction to maintain a steep concentration gradient.⁵⁷ Similarly, toxic gases produced in a reaction can be simultaneously extracted and dissolved into a parallel channel filled with a reactant.⁵⁸ This simultaneous process prevents the leakage of toxic gases via distillation. Furthermore, sequential chemical reactions can be designed on a centrifugal microfluidic platform by incorporating gas diffusion into the process.²⁶ Gas extraction is also used to analyze gaseous components in liquids. As shown in Fig. 2(d), gases can be extracted before being delivered to a sensing unit, such as in the cases of sensing CO₂ in seawater,^19,51 sulfur dioxide (SO₂) gas in wine,^31,59 or O₂ and NH₃ in samples.^28,30 Each unit incorporated electrical or optical sensing mechanisms to facilitate on-chip determination, thereby reducing time and labor costs. In certain instances, gas-reactive biological cells can serve as biosensors that exhibit fluorescence signals on the dissolution of target odorant molecules [Fig. 2(e)].¹⁶ Unlike other sensors, cell-based biosensors can switch their target gas molecules using different types of cells while retaining sensing selectivity. They require only simple calibration to establish the relationship between gas concentration and signal intensity.

A concentration gradient of gas molecules can be generated within a liquid-flowing channel, with the dissolution channel acting as the source and the extraction channel functioning as the destination. For instance, a pair of a source and a sink can be used to generate a CO₂ gradient in the flow of cell culture media.⁶⁰ Additionally, a CO₂ concentration gradient can be harnessed for the removal of particles in a low-energy water filtration process [Fig. 2(f)].⁵² The dissolved CO₂ molecules dispersed along the liquid stream create a pronounced ion concentration gradient across the channel, thereby prompting the diffusiophoretic migration of particles based on their surface charges and sizes. Furthermore, the distinct diffusiophoretic velocities of these particles facilitate their separation, which is contingent upon surface properties.⁶¹ 

The presence of bubbles in a liquid has a significant impact on the entire fluid network, giving rise to bubble logic⁶² and fluidic actuators.⁶³ Additionally, bubbles can serve as carriers of encapsulated gases and shell materials, thereby facilitating their applications across various fields.^9,64,65 Within a microfluidic system, bubbles can be generated through several mechanisms, including multiphase flow,⁶⁶ electrolysis,⁶⁷ chemical reactions,⁶⁸ cavitation-induced nucleation,⁶⁹ and physical stimuli using acoustic, optical, or thermal energy.⁶⁹ An alternative approach involves applying high gas pressure across GPMs to induce bubble formation within the liquid. These bubbles can maintain their size if the bubble is stabilized by a shell made of surfactants, lipids, or hydrophobic materials.^70,71 Positive or negative pressure can be applied to a PDMS membrane to produce or remove bubbles, respectively [Fig. 3(a)].⁷² Bubble inclusions could be applied for a bubble-based gating mechanism. However, the utilization of a nonporous membrane is restricted to basic operations due to challenges in controlling bubble size. Instead, by employing a microchannel membrane (i.e., porous membrane), high-speed imaging for the growth of uniformly sized bubbles can be achieved. This enables analytic studies in well-confined geometries [Fig. 3(b)].⁷³ The dynamic interfacial behavior was investigated to determine the effect of protein adsorption during bubble formation. Furthermore, efforts are being made to realize high-throughput monodisperse bubble formation using microchannel membranes.⁷⁵ Furthermore, membrane-based bubble formation might be highly valuable in situations requiring on-the-spot bubble formation. For instance, bubble-based perfusion in culture chambers⁷⁶ would be controlled in a localized and individual manner using membrane-based bubble formation.

Generally, GPMs are more commonly used for bubble removal rather than bubble formation. Both unintentionally and intentionally formed bubbles need to be eliminated once they have fulfilled their purpose.^69,77,78 Failure to remove these bubbles can result in their adhesion to walls, leading to increased hydrodynamic resistance and nonlinearity in the fluidic system.¹¹ Bubble removal devices have been comprehensively reviewed in recent papers.^69,77–79 Therefore, we specifically focus on membrane-based bubble removal in this Perspective article. Bubble removal in continuous flow has been developed for use as a pre- or post-treatment component. Bubbles can be evacuated through dense membranes or hydrophobic porous membranes by creating a pressure difference across the membranes. Dense PDMS membranes are employed when a vacuum can be applied to the device [Fig. 3(c)].³⁹ The PDMS-based bubble removal channel can be integrated as either a pre- or post-component in a multi-step microfluidic device. Furthermore, the feasibility of using a vacuum eliminates the necessity for applying high pressure to the liquid. Structural optimization, such as incorporating a bubble trap,⁷² a sloped structure,⁸⁰ or a longer channel,³⁹ can enhance bubble removal via dense membranes. In contrast, when aiming for higher removal rates, hydrophobic porous membranes are integrated into microfluidic devices. Hydrophobic porous membranes can successfully remove bubbles without requiring a vacuum.^33,34 As demonstrated in Fig. 3(d), a device solely dedicated to bubble removal achieved a high removal rate (1.5 ml/min) while maintaining a compact size (10 × 18.5 × 6 mm³ for width × length× height).³⁵ Although the flow rate becomes complex upon connection with another microfluidic device, the independently operating module is convenient and cost-effective. An exceptional example involves the use of liquid-infused membranes, which facilitate antifouling bubble removal.³⁶ Furthermore, the combination of membranes with different bubble wettability has demonstrated the proof of concept for microfluidic bubble removal.⁸¹ Bubble removal also aided in filling dead-end pores [Fig. 3(e)].⁷⁴ The presence of air in these dead-end pores was reduced by bubble removal, allowing liquids to occupy the volume quickly. Additionally, pre-vacuumed polymers can absorb air without the need for vacuum application after liquid loading, thereby eliminating the need for additional gas channels. Guided bubble removal can be leveraged to fabricate controlled interfaces of biopolymer membranes [Fig. 3(f)].³⁸ Applying a vacuum to stably trapped bubbles in the absence of liquid flow eliminated the need for intricate pressure balancing to prevent pressure fluctuations during membrane growth. Notably, gas diode membranes exhibited directionality in bubble transport, which can be employed in microfluidic devices for stepwise chemical reactions.⁶⁸ 

A porous membrane enables continuous evaporation while maintaining the gas–liquid interface within its pores. Various methods can be employed to enhance the rate of evaporation. Typically, a refreshing gas flow was used to maintain a steep vapor concentration gradient.⁴⁴ Thermal energy can also be utilized to facilitate evaporation [Fig. 4(a)].⁴⁵ The porous membrane served as a pathway for vapor and offered a stable gas–liquid interface for crystal growth. Similarly, an effective evaporation technique concentrated the solution, leading to a rapid concentration of samples containing short-lived radiolabeled tracers.⁸³ Another option for an evaporative membrane entails the adoption of a dense material, such as PDMS. Microfluidic devices composed of dense material have gas-permeable walls, which permit evaporation through them (i.e., pervaporation). However, the millimeter-scale thickness of the channel wall lowers the concentration gradient and restricts severe pervaporation. Over prolonged periods, such as during cell incubation, pervaporation can significantly increase osmolarity.⁸⁴ To address water loss, common solutions include humidifying the entire incubation system or placing an “iso-osmotic bath.”⁸⁵ In contrast, when there is a high surface-to-volume ratio and a thin dense membrane, pervaporation becomes effective enough to be a dominant phenomenon. Solvent mixture separation via pervaporation is achieved in channels that are hundreds of millimeters in length [Fig. 4(b)].⁴² The removal efficiency was studied using 2D numerical simulation, taking into account the advection in the nominal direction due to permeation flux, thereby elucidating the reason for the limitation in removal efficiency. Pervaporation also enables solute concentration in confined structures, allowing for low concentrations to reach their saturation concentration.⁸⁶ A similar mechanism in a longitudinal microchannel provides a valuable tool for analyzing the mutual diffusion of binary solution.⁸⁷ 

Furthermore, the continuity of liquid generates pervaporation-induced flow.⁸⁸ Leveraging this mechanism, a simple microfluidic pump was initially designed by mimicking leaves.⁸⁹ Subsequently, flow generation becomes locally and independently controllable by designing individual control channels for liquid channels, enabling various dynamic functions, such as molecular concentrators, pumps, and filters [Fig. 4(c)].⁴⁶ Numerical and experimental studies showed that the advection caused by the pervaporation-induced flow is sufficient to compare with diffusion of small molecules. Pervaporation-induced flow was also utilized to assemble particles [Fig. 4(d)].⁴⁰ The particle assemblies exhibit a porous structure that enables solvent penetration, thereby ensuring that the evaporation area remains constant. As a result, the assembly rate remains rapid even as the particle assembly grows and occupies a larger volume. Then, the particle assembly served as a homogeneous/heterogeneous nanoporous membrane whose pore properties are determined by the particle properties, including pore size, functional group, and wettability. This methodology of particle assembly has been employed for several applications, including transport control of small molecules,⁴⁰ solidification of colloidal dispersion,⁹⁰ and plasmonic supercrystals.⁹¹ 

In addition, condensation offers reverse working principles of pervaporation, and water supply through a GPM can be leveraged to collect water from the environment. Microfluidic channels can simplify the phenomena,⁹² and GPMs can serve as both walls and an all-around vapor source for fluidic channels. However, controlling on-chip vapor condensation is often challenging owing to its unpredictable nature. Only in a few cases is the condensation considered for manipulating liquids in microfluidics. For example, liquid marbles covered with hydrophobic powder can be refilled through condensation via cooling [Fig. 4(e)].⁴⁷ Although the condensation process is much slower than the evaporation process, condensation-based refilling offers a potential method for noninvasive volume recovery of a liquid marble. Similarly, a humid environment facilitates the reconnection of disconnected channels filled with hydrophilic fluorescent molecules.⁴⁶ Furthermore, we believe that the controllability of liquid connection via evaporation and condensation holds potential for gating mechanisms.⁸² Membrane-based condensation may enable humidity-based gating, in contrast to the pH control proposed in Fig. 4(f). For example, dissolutions of acidic/basic gases can provide similar pH control in nanopores, while humidification of the gases can accelerate capillary condensation. Moreover, water harvesting structures,⁹³ guided transportation,⁹⁴ and self-humidifying membranes⁹⁵ can be employed to collect water from the environment, increasing responsiveness to environmental conditions for humidity-dependent actuations.

In summary, transport of gases through GPMs in microfluidic devices offers a range of possibilities depending on types of penetrants and the specific system designed to guide them. These transport phenomena have been utilized in numerous ways, serving as the foundation for sophisticated microfluidic applications. The use of gases enables dynamic changes within liquid systems, affecting driving forces, actuation, and chemical stimuli. This dynamism includes rapid condition changes, repeatability, and the reversibility of these changes. The integration of a membrane enhances the device's practicality, either with or without interfacial pores. Thin membranes are fabricated and integrated to induce a higher gradient of chemical potential across them, thereby increasing the transport rate. Additionally, the thin structure reduces the device's volume, enhancing its responsiveness to changing conditions, albeit sometimes requiring a minimum volume for adequate gas absorption. Furthermore, the gas selectivity of GPMs opens up possibilities for unique applications that go beyond the functions based solely on the nature of the penetrants. However, certain applications, such as bubble formation and condensation, have been explored only to the extent of demonstrating their feasibility. Even for other gas transport mechanisms, recent developments do not appear to significantly diversify away from the previous research achievements. To overcome this challenge, it is necessary to focus on the type of GPM employed. By carefully selecting GPMs, high permeability or selective gas transports can be achieved, which enables devices to possess greater robustness and reduced size while maintaining the same performance. In light of these considerations, we conducted a concise yet comprehensive investigation into GPMs, aiming to provide guidance for their utilization in the field of microfluidics.

GPMs can be designed and fabricated in either dense or porous forms. In this section, we provided a summary of fundamental theories from a microfluidics-focused perspective, complemented by schematics presented in Fig. 5. It is worth noting that we do not discuss the integration of GPMs into microfluidic devices, which has already been extensively covered in the previous literature.^96–98 

There exists a reciprocal correlation between the permeation and selectivity. Therefore, the trade-offs between the permeability and selectivity of gases in homogeneous polymer membranes have been extensively studied.^103,104 Although originally formulated for homogeneous polymer membranes, the trade-off relation offers insight for evaluating various types of GPMs, including porous membranes, and serves as an initial guideline for selecting a suitable GPM for specific microfluidic applications. However, as indicated in Table I, the range of available GPMs for microfluidic devices is indeed limited. For dense membranes, PDMS is commonly employed because of its excellent performance in gas separation and high permeability for specific gases, particularly water. Nonetheless, this limits the application to the inherent selectivity of the PDMS membrane itself. Hence, there are new functionalities to be unlocked by incorporating the gas selectivity of other membranes, which can enhance microfluidic systems. For instance, improved gas selectivity and permeability can enhance the performance of sensing devices in terms of selectivity and robustness. Furthermore, selective collection of reactants in chemical and biological processes becomes feasible, enabling stepwise and segregated processes.

Notably, when dealing with mixed-gas environments, relying solely on permeability data from individual gases may not always yield accurate results. Gas selectivity in a mixture can differ from the ideal gas selectivity calculated based on individual gas permeabilities.¹⁰⁵ Like temperature, humidity plays a significant role in the permeability of gases, including vapor itself.^106,107 While the pressure effect on permeability is generally negligible in the typical pressure range in microfluidics. In Fig. 6, the gas permeabilities of PDMS^108–118 and Teflon AF2400^{109,114,118–125} were investigated from literature studies. Both materials are commercially available and well known for their high gas permeability.^114,126 The high permeability is a prerequisite in microfluidic applications because moderate working pressures are acceptable; otherwise, the microfluidic device undergoes deformation or rupture at high pressure. PDMS is a silicon-based rubbery polymer and is one of the most frequently used GPMs in microfluidics due to its mechanical properties, biocompatibility for microfluidics, and accumulated knowledge in fabrication and modifications.¹²⁷ As shown in Fig. 6(a), PDMS has high gas permeabilities and even higher water permeability.^114,126 Furthermore, PDMS has a hydrophobic nature with extremely low water solubility. However, PDMS exhibits anomalously high permeability, while Bian et al. recently suggested an explanation by hydrogen-bonded chains of nanoscopically confined water.¹²⁶ 

Teflon AF2400 is a fluorocarbon-based glassy polymer. As a fluorocarbon-based polymer, Teflon AF2400 also demonstrates high permeability, as shown in Fig. 6(b). The O₂ and CO₂ gas permeabilities of Teflon AF2400 are comparable to those of PDMS, implying that Teflon AF2400 membranes have the potential for use in cell culture systems.¹²⁸ Therefore, Teflon AF2400 can be a PDMS alternative when similar O₂ and CO₂ gas permeability is required, but water permeation is prohibited. Furthermore, the chemical and mechanical durability of Teflon AF2400 enables the use of organic solvents without significant swelling and permeation.^123,129 Teflon AF2400 is commercially available in a variety of forms. In particular, tubular Teflon AF2400 membranes have inspired a new concept of fluidic platform, i.e., the aforementioned “tube-in-tube reactor.”^7,54 In addition to the above two materials, there are several popular alternatives for dense membranes. For example, poly[1-trimethylsilyl-1-propyne] (PTMSP) is popular for its ultrahigh permeability,¹³⁰ and polyimide is likewise popular for its excellent gas selectivity.¹⁰⁴ 

Essentially, no gas selectivity is realized by the porous membrane when its pore size (d) surpasses the mean free path of penetrants (λ); for air, λ = 64–68 nm at room temperature and atmospheric pressure.¹³¹ This implies that, if d > λ, sophisticated gas–membrane interactions in the ambient condition can be neglected.

On the other hand, hydrophilic membranes should be utilized to prevent the absorbance of hydrophobic molecules¹³⁸ or adopt surface modification for biocompatibility.¹³⁹ In the case of a hydrophilic membrane, an additional strategy is required to prevent pressure-driven permeation [Fig. 5(c)]. The gas pressure should be higher than the liquid pressure to maintain the gas–liquid interface, facilitating bubble generation.⁷³ Smaller pores within a hydrophilic membrane can stabilize the gas–liquid interface in both directions of a pressure gradient.²⁴ However, there are several critical issues that require careful consideration. The lack of a thin structure poses challenges for the membrane to effectively mitigate transport of gases, resulting in a significant reduction in transport rates. As illustrated in Fig. 5(c), this reduction is primarily attributed to the additional diffusion resistance presented by the liquid barrier within the hydrophilic membrane.¹⁴⁰ Moreover, hydrophilic membrane pores are susceptible to small fluctuations of surroundings or condensation, thereby causing complete wetting of the pores and subsequent pressure-driven permeation. Therefore, the utilization of hydrophilic membranes for GPMs is only valid when a stable gas–liquid interface is guaranteed during the operation of a microfluidic device.

The intermediate transport behavior of dense and porous membranes can be achieved by reducing the pore size of the porous membrane. When the pore size becomes smaller than the mean free path (i.e., d < λ), it results in selective gas transport, driven by several phenomena, namely, Knudsen diffusion, surface diffusion, capillary condensation, and molecular sieving. These transport processes occurring at the pore level are inherently mixed, even within a single pore, and this situation varies simultaneously from pore to pore.¹⁴¹ Detailed mechanisms of transport and the origin of selectivity are extensively covered in the existing literature.^142–144 Building upon the intermediate performance achieved by balancing gas selectivity and permeability through these phenomena, numerous GPMs have emerged and continue to be developed. Section IV delves into these GPMs and explores their potential for utilization in microfluidics.

We offer an overview of newly emerging gas-selective GPMs, with a specific focus on membrane technology and microfluidic applications of gas separation. This on-chip gas separation is poised to make a significant impact in the field of microfluidics, promising precise and efficient processing of gas and liquid within microfluidic platforms.

Recent advancements in membrane technology aim to address the perennial challenge of striking the right balance between permeability and selectivity. In response to this challenge, novel membrane materials, such as polymers of intrinsic microporosity (PIMs) and thermally rearranged (TR) polymers, have been developed. These super-glassy polymers exhibited exceptional performance, albeit with certain durability limitations. Researchers are actively exploring innovative strategies to overcome these obstacles.^145,146 Notably, the ionization of polymers has been employed to further enhance membrane properties.¹⁴⁷ Additionally, heterogeneous membranes have been introduced, including thin-film composite (TFC) membranes¹⁴⁸ and mixed-matrix membranes (MMMs).¹⁴⁹ TFC membranes involve the formation or deposition of an ultrathin dense polymer layer onto a rigid membrane support, enabling the simultaneous integration of the selectivity of dense membranes and the mechanical strength of the support layer. MMMs, on the other hand, consist of organic polymers blended with inorganic additives to overcome the limitation of each. Generally, inorganic material improves gas selectivity, while polymers provide mechanical stability without losing the membrane role. The inorganic additives include zeolites, metal-organic frameworks (MOFs), nanocarbons, and other nanoparticles. Furthermore, the heterogeneous membranes utilize the other emerging polymers, such as PIMs and TRs.^145,150,151

To date, various types of membranes have been developed for diverse gas separation applications, including gas–gas separation, dehumidification, and pervaporation separation.^145–152 However, their suitability for microfluidics may be limited. This is mainly due to challenges in achieving improved permeability of gases with highly selective membranes, which are crucial for miniaturization and robust processing, the core aims of microfluidics. Emerging gas-selective membranes offer the potential to specifically enhance precision and efficiency in microfluidic applications. Furthermore, there are still several advantages in microfluidics when utilized for gas separation. Serial placement of GPMs in a microfluidic device realized miniaturized serial gas processing flow.⁵⁴ A high surface-to-volume ratio reduces the amount of samples required and allows for supporting liquid solutions, such as enzyme, to work effectively.¹⁸ The application of selective membranes allows for the pre-separation of sample gases, effectively eliminating interfering gases that may produce similar signals to the target gas or disrupt the sensing mechanism.¹⁵³ Additionally, these membranes enable the retention or removal of gaseous cellular by-products, such as nitric oxide (NO) and methane (CH₄), while facilitating the transport of O₂ and CO₂. Such selective control of gaseous components within a liquid environment can be beneficial for studying the impact of by-products on cell culturing within a gas-controlled microfluidic device. This approach enables a shift in focus from idealized scenarios where only single gases are considered, to more realistic situations involving mixtures of multiple gases. By incorporating emerging gas-selective membranes into microfluidic platforms, researchers gain the capability to explore and manipulate complex gas mixtures, thereby advancing the understanding and application of gas-related processes in diverse fields.

Last, it is important to acknowledge that the integration of emerging gas-selective membranes into microfluidics is not without challenges. The development of easy and cost-effective fabrication methods for on-chip integration is a crucial aspect that needs to be addressed. Furthermore, investigations on the compatibility of these membranes with the physiochemical environment of microfluidics and their interaction with biomolecules are essential.¹⁵⁴ Also, the requirement for transparency in observations imposes limitations on the selection of membranes. Additionally, inherent issues associated with certain membranes, such as aging and plasticization, must be addressed as important considerations. Despite the barriers that need to be overcome, it is evident that the infinite combination of gases and liquids undoubtedly enriches the field of microfluidics. Continued research and development in this area holds tremendous potential for further advancements and innovations in gas-related processes within the microfluidic domain.

We categorized gas transport mechanisms in microfluidics into three distinct groups: the dissolution and extraction of gases, the formation and removal of gas bubbles, and the evaporation and condensation of solutions. Our exploration delved into the fundamental mechanisms of GPMs and provided a comprehensive review of their recent applications, drawing upon a wealth of literature from the realm of microfluidics. In particular, GPM-based gas transport offers numerous advantages, especially in scenarios where rapid switching of physiochemical conditions and on-site molecule transport into or from liquids within devices is desired. While the literature discussed here predominantly emphasizes the transport itself and the ensuing physiochemical phenomena, there exists a profound connection to potential biological applications. Beyond merely regulating O₂/CO₂ levels for cell culture, GPMs open the door to precise molecule delivery within a fluidic network—a capability of paramount importance. Gating mechanisms, such as bubble inclusion and capillary evaporation, can be harnessed for the targeted delivery of drugs, signaling molecules, and nanoparticles. Particle migration methods, such as diffusiophoresis and pervaporation-induced flow, can similarly be applied to cells, facilitating the transport of vital biological materials, such as nucleic acids, proteins, and liposomes. These dynamic mechanisms provide researchers with powerful tools for conducting intricate biological experiments within microchannels, similar to those traditionally conducted in cell culture plates. Furthermore, another promising application lies in the assembly of antibody-tagged particles for simultaneous detection. Moreover, emulating the fluidic system in plants not only enables the replication of their microfluidic capabilities, but also allows for a comprehensive understanding through the utilization of devices that offer advantages in manipulating parameters, such as geometry, contact angle, and environmental conditions.^155,156 However, despite these compelling advantages and promising applications, we have observed stagnation in the development and innovation of GPM-based microfluidic applications. Our analysis has indicated that a dearth of diversity in available membranes, specifically scarcity of polymer choices, severely restricts the range of gases and liquids that can be selectively transported. To overcome this limitation, it is imperative that we foster a deeper understanding of both conventional and emerging membrane technologies. By increasing the diversity of GPMs, we can significantly broaden the scope of microfluidic applications and fully unlock the potential of liquid and gas manipulation within microfluidic systems. We believe that by expanding our knowledge in this area, we can usher in a new era of innovation and transformative applications in microfluidics.

This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korean government (MSIT) (NRF-2020R1A2C3003344). This work was also supported by the Technology Innovation Program (Development of Underwater Breathing Device Technology without An Oxygen Tank, 20026073) funded by the Ministry of Trade, Industry & Energy (MOTIE) of Korea.

The authors have no conflicts to disclose.

Sangjin Seo: Conceptualization (lead); Writing – original draft (lead); Writing – review & editing (equal). Taesung Kim: Conceptualization (equal); Funding acquisition (lead); Supervision (lead); Writing – review & editing (equal).

Data sharing is not applicable to this article as no new data were created or analyzed in this study.