Engineered Microdefects in Nano‐Membranes for Enhanced Ion Selectivity and Membrane Durability in Vanadium Redox Flow Batteries

Abstract

In vanadium redox flow batteries (VRFBs), ultrathin perfluorinated sulfonic acid (PFSA) membranes with a highly ordered morphology are proposed to enhance proton/vanadium ion selectivity, overcoming the limitations of conventional membranes. However, dense structures hinder proton transport and cause progressive deformation during cell operation, reducing overall performance. In this study, sub‐25 nm ultrathin PFSA membranes are demonstrated by harnessing engineered microdefects as size‐exclusive pores, promoting proton transport while maintaining structural integrity. To achieve this, only 14 molecularly thin PFSA Langmuir monolayers are stacked with adjusted packing density, and proton‐conducting pathways are sufficiently established via pre‐swelling. Unlike conventional strategies for robust dense membranes, this approach suppresses deformation of the channel morphology during cell operation, with higher ion selectivity than Nafion 211. Finally, the optimized ultrathin membrane exhibits superior cyclic and rate performance compared to the commercial membrane, delivering ≈76% energy efficiency and long‐term stability at 200 mA cm⁻² without capacity decay.

Sub‐25 nm ultrathin perfluorinated sulfonic acid (PFSA) membranes with engineered microdefects outperform commercial membranes that are 1000 times thicker, achieving ≈76% energy efficiency with superior capacity retention during long‐term operation in vanadium redox flow batteries.

1. Introduction

Perfluorinated sulfonic acid (PFSA) membranes are widely used as perm‐selective proton exchange membranes (PEMs) in vanadium redox flow batteries (VRFBs) due to their exceptional chemical stability and proton conductivity.^{[ 1} , ² , ³ , ⁴ , ^{5 ]} However, the distinct and interconnected hydrophilic channel morphology of PFSA‐based PEMs causes poor proton selectivity over active vanadium ion species in electrochemical reactions, limiting the efficient and reliable operation of the VRFBs.^{[ 6} , ⁷ , ⁸ , ⁹ , ^{10 ]} To address this limitation, several studies have attempted to either directly tune channel morphology^{[ 9} , ¹¹ , ^{12 ]} or incorporate hydrophilic inorganic fillers into PFSA membranes,^{[ 13} , ¹⁴ , ¹⁵ , ^{16 ]} aiming to regulate ion transport pathways and enhance ion selectivity in VRFBs. Meanwhile, as an alternative to PFSA‐based PEMs for VRFBs, tailored hydrocarbon‐based PEMs,^{[ 17} , ¹⁸ , ¹⁹ , ²⁰ , ^{21 ]} and their composites^{[ 22} , ²³ , ²⁴ , ²⁵ , ²⁶ , ^{27 ]} have been investigated to address the limitations of PFSA‐based PEMs. In addition to ion‐conducting polymers, non‐ion conducting size‐exclusive membranes^{[ 28} , ²⁹ , ³⁰ , ^{31 ]} have also shown promise as the alternatives in VRFBs by exploiting the difference in the hydrodynamic radius of hydronium ion (H₃O⁺, 0.15 nm)^{[ 32 ]} and vanadium ions, where the stokes radii of V²⁺, V³⁺, VO²⁺, and VO₂ ⁺ are 0.32, 0.32, 0.21, and 0.28 nm,^{[ 33 ]} respectively. However, for all membrane types, it remains a challenge to simultaneously achieve both high proton conductivity and high proton selectivity to vanadium ions, overcoming the intrinsic trade‐off between conductivity and selectivity inherent in the material properties.^{[ 34} , ³⁵ , ^{36 ]}

Recently, we demonstrated ultrathin PFSA membranes (sub‐50 nm) with highly ordered channel morphology, which exhibit distinct material properties compared to bulk, micron‐thick PFSA commercial membranes, despite being made from the same PFSA ionomers.^{[ 37 ]} These ultrathin PFSA membranes, formed by layer‐by‐layer deposition of dense PFSA Langmuir monolayers at the air/water interface, exhibit an in‐plane aligned channel morphology, which leads to tortuous ion transport pathways in the through‐plane direction, in contrast to commercial PFSA membranes with randomly interconnected hydrophilic channels. Consequently, they exhibited proton selectivity ≈500 times higher than that of the 25 µm‐thick commercial PFSA membrane for VRFBs. However, the densely packed monolayers with the highly aligned channel morphology, through the repeated compression of the monolayer and thermal annealing, significantly impede proton transport along the through‐plane direction, resulting in very high membrane resistance at the beginning of the operation.^{[ 37} , ³⁸ , ^{39 ]} VRFBs with densely packed ultrathin PFSA membranes were not operated immediately due to the high ohmic overpotential leading the initial cell voltage over the open circuit voltage of 1.6 V, which is the upper cut‐off voltage for charge–discharge cycles.^{[ 38 ]} Although cell operation was achieved through repeated trials of charge–discharge cycles as the membrane resistance decreased, a decline in cell performance and capacity was observed, implying electric field‐induced structural distortion in the dense layers, which might form macropores (Scheme 1a) to facilitate proton transport for the redox reactions during operation. Despite this distortion, however, the ultrathin PFSA PEMs were still able to operate stably for up to 800 cycles at a high current density of 200 mA cm⁻², showing comparable performance to that of a 10‐µm‐thick PFSA PEM.

Scheme 1.

Comparison of a) macropores and b) microdefects in the PFSA layers.

Inspired by these results, we devised a strategy to intentionally induce microdefects, though too small to be defined as pores, as size‐exclusive proton paths in the ultrathin PFSA PEMs. By employing microdefects that transport protons but not vanadium ions, the ultrathin PFSA PEMs can function as both ion‐conducting PEMs and size‐exclusive membranes (Scheme 1b).^{[ 40} , ⁴¹ , ^{42 ]} Moreover, the benefits of the highly ordered hydrophilic channel morphology can be utilized without electric field‐induced distortion, due to the inherently placed microdefects.

In this study, we develop high‐performing ultrathin PFSA membranes with engineered microdefects. Sub‐25 nm ultrathin PFSA membranes are prepared by stacking only 14 PFSA Langmuir monolayers from the air/water interface with adjusted packing density. To establish proton selective transport pathways sufficiently, the ultrathin membranes are pre‐swollen with electrolytes (3 m H₂SO₄ aqueous solution), while keeping highly‐ordered hydrophilic nanostructures. The PFSA membrane, with swollen and aligned channel structures, still shows proton selectivity 80 times higher than that of the 25 µm‐thick Nafion 211, and due to pre‐induced microdefects originated from loosely packed structure, it operates in a VRFB cell at a reduced ohmic overpotential without requiring the proton path development process through nanostructure distortion. This newly designed feature enables the ultrathin membrane to exhibit superior cell performance compared to Nafion 211 across the entire current density range of 40 to 200 mA cm⁻². Specifically, the cell with the ultrathin membrane maintains ≈95% of its original capacity after 100 charge–discharge cycling tests at 200 mA cm⁻², while the cell with Nafion 211 loses ≈50% of its capacity under the same conditions. Moreover, after more than 500 cycles, it exhibits higher efficiency and lower capacity loss compared to Nafion 211. These results show that the ultrathin PFSA membrane possesses facilitated proton conduction through the introduction of engineered microdefects into the highly aligned channel morphology, demonstrating its potential to serve as an alternative to commercial PFSA‐based membranes.

2. Results and Discussion

Ultrathin PFSA membranes are manufactured from pre‐aligned PFSA Langmuir monolayers at the air/water interface. Notably, PFSA ionomers are semi‐crystalline polymers composed of flexible linear polymer chains^{[ 1 ]} and exhibit interfacial activity due to their amphiphilic nature, which originates from their molecular structure (Figure S1, Supporting Information). Consequently, through direct spreading of the dispersion, PFSA ionomers can be readily adsorbed onto the air/water interface, where they freely float and primarily interact with neighboring ionomers through electrostatic interactions.^{[ 43 ]} Owing to these molecular characteristics and their free mobility at the interface, PFSA ionomers can be easily packed through physical compression. Thus, the packing density of PFSA ionomers in the monolayer at the interface can be controlled by physical compression, thereby forming voids within the monolayer that could potentially act as microdefects.

As depicted in Figure 1a,b, two different compression protocols were used to achieve different packing densities: 1) For a loosely packed monolayer, the PFSA monolayer was compressed once; and 2) for a densely packed monolayer, the PFSA monolayer was prepared by a multiple compression‐expansion procedure. In general, when the PFSA Langmuir monolayer is simply compressed once, it exhibits a relatively uniform structure; however, small defects or cracks are still present.^{[ 44 ]} In contrast, the multiple compression‐expansion process results in a highly dense monolayer, as repeated compression allows the ionomers to rearrange, leading to a more compact structure that effectively minimizes microdefects.^{[ 37} , ^{45 ]} Here, the PFSA monolayer was compressed at least five cycles of a multiple compression‐expansion procedure based on our previous study to achieve a more densely packed monolayer with aligned side chains over a large interfacial area (A ≈ 200 cm²) (See Discussions S1 and S2, Supporting Information for the details of the compression procedures). As each monolayer was compressed, the surface pressure (Π) increased for both procedures. However, the monolayer compressed once displayed a gradual slope in the isotherm profile compared to the PFSA monolayer prepared by multiple compression‐expansion, indicating a softer monolayer due to its loosely packed structure (Figure 1c). Indeed, the PFSA monolayer prepared by simple compression exhibited a lower maximum static compression modulus (E _{max. comp, S_com} = 107.86 mN m⁻¹) derived from the compression modulus, E _comp = −A(∂Π/∂A) _T ,^{[ 46} , ⁴⁷ , ^{48 ]} compared to the repeatedly compressed one (E _{max. comp, RC_com} = 121.54 mN m⁻¹) (Figure S2a, Supporting Information). Furthermore, the monolayer compressed once showed higher compressibility (β)^{[ 43} , ^{49 ]} throughout the entire compression process, indicating that it was more easily compressed due to its loose structure (Figure S2b, Supporting Information). Finally, to manufacture the ultrathin PFSA membranes, the compressed PFSA monolayer on the interface was transferred onto the substrate by Langmuir–Blodgett (LB) deposition. As shown in Figure 1d, during LB deposition, monolayers are sequentially transferred in a layer‐by‐layer (LBL) manner, where hydrophilic side chains interact during the upstroke, while hydrophobic backbones come into contact during the downstroke. Regardless of the packing density, the monolayer exhibited a transfer ratio close to 1, confirming stable and uniform deposition throughout the process (Figure S3, Supporting Information). Additionally, since the deposition is performed at the center of the LB instrument, a single stroke enables the simultaneous transfer of the PFSA monolayer onto both sides of the substrate. This process ultimately results in the formation of an ultrathin PFSA membrane on both surfaces of the supporting substrate.

Figure 1.

Schematic illustrations of preparation steps for PFSA ionomer monolayer by (a) simple compression method and (b) repeating compression‐expansion method. c) Surface pressure‐area (Π‐A) isotherms of PFSA ionomer monolayer on the air/HCl aqueous interface, compressed by each method. The black dashed line represents the Π‐A isotherm of the PFSA Langmuir monolayer simply compressed by physical barriers at a rate of 5 mm min⁻¹. Additionally, the red line represents the Π‐A isotherm of the simply compressed monolayer before the deposition process, while the blue line represents the Π‐A isotherm of the repeatedly compressed monolayer during the final compression before deposition. The initial trough area is 243 cm², and each monolayer was compressed at a rate of 5 mm min⁻¹ for the following deposition steps. d) Schematic diagram of the manufacturing process of ultrathin PFSA ionomer membranes via Langmuir–Blodgett deposition of PFSA monolayer on the air/water interface. During deposition, the substrate is immersed in the center of the trough, which results in the monolayer being transferred onto both sides of the substrate. Schematic illustrations were reprinted with permission from ref. [37]. Copyright 2021 American Chemical Society.

Regardless of packing density of PFSA layers, the ultrathin PFSA membrane exhibited an anisotropic scattering pattern as shown in the grazing‐incidence small‐angle X‐ray scattering (GISAXS) experiments (Figures 2a; S4a, Supporting Information). The GISAXS scattering pattern was located at q _z = 0.20 to 0.25 Å⁻¹, with a localized pattern along the azimuthal angles (ω) between 0 and 20° (Figure 2b,c; Figures S4b,c,S5, Supporting Information). Interestingly, as shown in a previous study,^{[ 39 ]} the maximum scattering intensity, q ^* _z = 0.22 Å⁻¹, with their second‐order reflection peaks (q ^** _z = 2 × q ^* _z) at q ^** _z = 0.43 Å⁻¹ was also observed, implying the highly ordered structure in out‐of‐plane direction. Moreover, quantitatively, the critical azimuthal angle (ω ^*) of ultrathin PFSA membrane formed by LB deposition was ω ^* = 3°, which is close to ω ^* = 0° suggesting nearly perfect lateral order of the PFSA monolayers (Figure 2d).^{[ 37} , ^{50 ]}

Figure 2.

a) 2D GISAXS scattering patterns of the ultrathin PFSA membrane, PFSA thin films prepared via spin‐coating and dip‐coating, and the ultrathin PFSA membrane fully swollen in an electrolyte (3 m H₂SO₄ aqueous solution). b) 1D intensity profiles of PFSA membranes in both dried and swollen states. The black solid and dashed lines represent the 1D intensity profiles of the dried membrane along the z‐ and y‐axes, respectively. Similarly, the red solid and dashed lines correspond to the pre‐swollen membrane along the z‐ and y‐axes, respectively. c) Azimuthal angular profiles of PFSA membranes and thin films, showing intensity variations as a function of the azimuthal angle; each profile is extracted at q ^* _z . d) Critical azimuthal angle (ω ^*) of ultrathin PFSA ionomer membranes in both dried and swollen states, as well as for dried PFSA thin films prepared via spin‐coating and dip‐coating. e) Schematic illustrations of the well‐ordered channel morphologies of the ultrathin PFSA membrane in both dry and wet states.

In addition to the packing density, thermal annealing can affect the nanostructures of the pre‐aligned PFSA ionomers on the aqueous substrate, as discussed in our recent study.^{[ 39 ]} The thermal energy, under the transition temperature (T_α ) of PFSA ionomers, imparted ionomer chain mobility, leading to a more packed structure, as indicated by a narrower ion‐conducting channel and reduced proton and vanadium ion transportation. Therefore, to maintain the pre‐formed nanostructure including microdefects, the PFSA layers were kept under the swollen state in the electrolyte without the annealing process. Even in a swollen state within the electrolyte, the PFSA layers displayed an anisotropic scattering pattern in the GISAXS similar to that of the ultrathin membrane in its dried state; however, the peak was broadened along q _z = 0.11 to 0.25 Å⁻¹ with the maximum scattering intensity shifted to, q ^* _z = 0.18 Å⁻¹ (Figure 2a,b). Notably, as swelling altered the spacing between the ordered lamellar structures, the second‐order reflection peak disappeared. However, the membrane still exhibited an intensity profile localized along the z‐axis with a lower critical azimuthal angle (ω ^* ≈ 7°) compared to thin films produced by spin‐coating or dip‐coating, indicating that the well‐aligned channel morphology was well preserved (Figure 2c,d). The average d‐spacing of the aligned structure was estimated from d = 2𝜋/q ^*. The ultrathin PFSA membrane in the swollen state exhibited a larger inter‐domain spacing (d = 3.46 nm) compared to its dry state (d = 2.81 nm) (Figure 2e). Considering that the PFSA monolayers were deposited in an LBL manner on the substrate, the average thickness of one PFSA layer is estimated to be half of d (1.41 nm in the dry state), which agrees well with the reported thickness of a PFSA monolayer (≈1.3 nm)^{[ 44 ]} and the thickness estimated from the AFM analysis (Figure S6, Supporting Information).

In PFSA membranes, ion transport occurs through ion channels, making the transport properties closely related to the channel morphology. Considering that ions are ultimately transported in the through‐plane direction, microdefects induced by adjustments in packing density can create additional ion transport pathways between the planarly‐oriented hydrophilic domains, significantly impacting the ion transport properties, proton conductivity (σ) and vanadium permeability (P). For the measurement, the free‐standing composites were prepared by transferring the ultrathin PFSA membranes, with controlled compression densities, onto a track‐etched polycarbonate membrane with cylindrical pores of 50 nm diameter (PC50). To exclude the possible mechanical deformation and defects of the PFSA layers at around the pores of PC50, the 14 PFSA monolayers at least, the minimum number of layers sufficient to fully cover the pores during the LB deposition regardless of packing density (Figure S7, Supporting Information), were deposited onto PC50. Based on the packing density, the composites were named PN14L with the loosely packed monolayer through simple compression and PN14D with the densely packed monolayer through repeated compression. Last, to indicate whether thermal annealing and pre‐swelling in a 3 M H₂SO₄ aqueous solution were conducted, the thermal annealing temperature and the dry or wet state were annotated at the end of the sample name. For example, PN14L samples subjected to thermal annealing at 80 °C in the dry and wet states were labeled as PN14L_80dry and PN14L_80wet, respectively.

The highly ordered channel morphology of ultrathin PFSA membranes was well retained even after deposition onto the PC supporting membranes (Figure S8, Supporting Information), in agreement with the observations in GISAXS analysis. Notably, SEM surface analysis showed that ultrathin PFSA/PC composite membranes with loosely packed layers commonly exhibited black spots on the surface. Furthermore, these spots disappeared in membranes prepared with densely packed PFSA layers or after thermal annealing (Figure S9, Supporting Information). Considering that regions with lower electron density appear darker in SEM images,^{[ 51} , ^{52 ]} these spots are presumed to correspond to microdefects, representing small voids or relatively loosely packed regions that may exist within the compressed PFSA monolayer.

To investigate the ion transport properties, the vanadium ion permeation and proton conduction through the membranes were evaluated. As shown in Figure S10a,b (Supporting Information), even though the PC supporter was not effective at blocking vanadium ions, most ultrathin PFSA/PC composite membranes exhibited P values several orders of magnitude lower than that of the commercial PFSA membrane (N211) because their highly ordered channel morphologies with ≈3 nm in size, which effectively inhibit vanadium ion permeation. However, this morphology also restricts proton conduction, resulting in lower σ values compared to N211 (Figure S10c, Supporting Information). To examine the packing density and post‐treatment effects on the ion transport properties, the P and σ of a PFSA membrane on the PC supporter were estimated using a series model with ex situ measurements of proton conductance (κ) and vanadium permeance (p) for each composite membrane as depicted in Experimental details in Supporting Information. Notably, the loosely packed and pre‐swollen layer (Loose layer_wet) exhibited one order of magnitude higher σ (σ = 56.4 mS m⁻¹) compared to the densely packed and thermally annealed PFSA layer (Dense layer_80dry) at 80 °C (σ = 3.7 mS m⁻¹) (Figure 3a). Additionally, the loosely packed PFSA layer thermally annealed at 80 °C (Loose layer_80wet) exhibited similar σ (σ = 47.8 mS m⁻¹) to the untreated layer (Loose layer_wet), suggesting that the packing density of the layer has a more significant impact on proton conduction than post‐treatment. Likewise, in vanadium ion permeability, both Loose layer_wet and Loose layer_80wet (P = 3.4 × 10⁻¹⁶ m² s⁻¹ and 6.0 × 10⁻¹⁶ m² s⁻¹ for Loose layer_wet and Loose layer_80wet, respectively) exhibited two orders of magnitude higher P values than Dense layer_80dry (P = 7.1 × 10⁻¹⁸ m² s⁻¹) (Figure 3a). Nonetheless, due to the excellent vanadium ion barrier properties provided by the highly ordered PFSA Langmuir monolayers, both the loosely and densely packed ultrathin PFSA membranes (Loose layer_wet and Loose layer_80wet, and Dense layer_80dry) showed three and five orders of magnitude lower P compared to N211 (P = 5.2 × 10⁻¹³ m² s⁻¹ for N211), respectively, resulting in the 80 and 200 times higher ion‐selectivity of protons over vanadium ions compared to commercial membranes (Figure 3b).

Figure 3.

a) Material properties for proton and vanadium ion transport of ultrathin PFSA membranes according to compression method, thermal annealing, and pre‐swelling. The PFSA layers are labeled as Loose and Dense layers based on their preparation methods. The Loose layer refers to a PFSA layer formed by a simple compression method, while the Dense layer refers to one prepared through a repeated compression process. The VO²⁺ ion permeability and proton conductivity were calculated from the VO²⁺ ion permeance and membrane resistance using the series model of composite membranes. b) proton/VO²⁺ ion selectivity of each PFSA layer. c) The IR drop at the initial VRFBs cycle and (d) the IR drops of each ultrathin PFSA/PC composite membranes during VRFBs operation at the 40 mA cm⁻². Ultrathin PFSA/PC composite membranes are labeled as PN#D and PN#L, where # represents the total number of PFSA layers deposited on the PC support, and D and L indicate that the PFSA membrane consists of densely packed and loosely packed layers, respectively. Since PFSA monolayers were simultaneously transferred onto both sides of the substrate during the LB deposition process, each side of the PC support received half of the total number of layers. For example, when # = 14, seven layers were deposited on each side of the PC support. Last, e) schematic illustrations compare the defect evolution between the dense and loose layers under cell operating conditions.

In addition to ex situ measurements, ultrathin PFSA/PC composite membranes were applied as ion‐selective membranes in a VRFB single cell to evaluate in situ membrane performance influenced by packing density and post‐treatments. Fundamentally, the low operational ohmic resistance of the VRFB cell is required to ensure sufficient charge capacity and stable cell operation across the discharge and charge cut‐off voltage range from 1 to 1.6 V.^{[ 53} , ⁵⁴ , ^{55 ]} At a low current density of 40 mA cm⁻², starting with a 3.5‐valence vanadium electrolyte, the VRFB cell with the dried PN22D_80dry did not operate initially due to a high overpotential, showing a substantial IR drop (≈7.95 V), in contrast to the cell with N211. With the repeated operation, the IR drop (ΔV) during the initial cycles gradually decreased, indicating a reduction in internal resistance, and the VRFB cell began operating after the 9th trial (Figure 3c). However, with a loosely packed ultrathin PFSA membrane in a swollen state (PN14L_wet), the VRFBs were stably operated, showing a low IR drop (≈0.08 V) comparable to that of the single cell with N211 (≈0.07 V). For example, during the initial cyclic test at a low current density (40 mA cm⁻²), the single cell with PN14L_wet operated stably without the gradual IR drop reduction for the operation, while the cell with PN22D_80dry showed a progressive reduction in IR drop as the cycles proceeded (Figure 3d).

When the overpotential is primarily caused by the high ohmic resistance of the membrane, the electrochemical reaction in the VRFB cannot proceed effectively. As a result, the cell voltage drop was mainly dominated by the ohmic overpotential, and the high resistance prevented the redox reactions of the electrolytes from occurring under the cut‐off voltage.^{[ 56 ]} In particular, in in situ proton conductivity (σ _{in situ}) of PFSA membranes extracted from the in situ area resistances of VRFB single cell with composite membrane (R _{in situ SC, PN}) (Experimental details, and Tables S1 and S2, Supporting Information), loosely packed PFSA membrane in swollen state (1.29 mS m⁻¹) showed three orders of magnitude higher proton conductivity than that of densely compressed PFSA membranes with thermal annealing (1.57 × 10⁻³ mS m⁻¹). Considering the similarity in the planarly‐oriented channel morphology between loosely and densely packed membranes, the ≈1000‐fold difference in proton conductivity is presumably attributed to packing density from the distinct compression methods for PFSA monolayer at the air/liquid interface. Simple compression likely induces microdefects in loosely packed membranes, leading to enhance proton transport across the transverse direction of the membrane, while maintaining size exclusion for vanadium ions, as evidenced by 80‐fold higher ion‐selectivity compared to commercial membranes (Figure 3b).^{[ 41} , ⁵⁷ , ^{58 ]} This suggests that the microdefects induced by controlled compression of the PFSA monolayer are sufficiently small to maintain size exclusion for vanadium ions.

For ultrathin PFSA membrane, ion permeation rates will vary with pathway tortuosity owing to identical material and channel morphology. According to the tortuosity model,^{[ 59 ]} compared to loosely packed membranes, densely compressed membranes theoretically exhibit ≈600 times higher tortuosity due to their three orders of magnitude lower proton conductivity and smaller channel size (Discussion S3, Supporting Information).^{[ 37 ]} However, given that the channels are aligned horizontally regardless of compression density, the reduction in tortuosity appears to result from microdefects induced by loose compression that interconnect the Langmuir monolayers. The ultrathin PFSA membranes, composed of loosely packed layers, allowed for unimpeded proton flux through microdefects and pre‐swollen ion channels with still preventing the vanadium ions permeation (Figure 3b), enabling stable cell operation. In contrast, in densely packed ultrathin membranes, the lack of proton transport pathways connecting the Langmuir monolayers led to high overpotential under initial operation conditions, as ions encountered greater resistance in passing through the membrane. The significant potential drop within the highly resistive densely packed PFSA layers could lead to structural distortion in the aligned morphology, potentially causing unintended larger defects, resulting in a gradual decrease in IR drop (Figure 3e). When the VRFB cell with densely packed PFSA membrane operated at the ninth trials, the σ _{in situ} of Dense PFSA_80dry (1.08 mS m⁻¹) increased to a value comparable that of the loosely packed membrane (Loose PFSA_wet), indicating significant structural deformation within the cell (Table S2, Supporting Information).

However, as shown in Figure 3d, the composite (PN22L_wet), which has thicker PFSA layers than PN14L_wet, showed a continuous reduction in ΔV, even though it consisted of loosely packed layers. This result suggests that microdefects can effectively function when the layered structure is less complex. Even with more loosely packed PFSA layers, proton movement may be restricted by the increased complexity of the layered structure. Due to the higher ohmic resistance caused by increased hydrophobic layers, microdefects in many PFSA layers may become larger to facilitate proton conduction during cell operation. Based on this result, a minimum number of PFSA layers in the composite is preferred. In this study, a 14‐layer PFSA membrane composed of loosely packed monolayers provides the most favorable conditions for stable operation (Figure 3d,e).

The VRFB single‐cell tests at various current densities (from 40 to 200 mA cm⁻²) were conducted to evaluate the performance of the ultrathin PFSA membrane with microdefects induced by loosely packed layer and their impact on the cell efficiencies. The rate performance of VRFBs with each membrane is shown in Figure 4 , including the Coulombic efficiency (CE), voltage efficiency (VE), and energy efficiency (EE). PN14L_wet and PN22L_wet are composite membranes with loosely packed PFSA layers but differ in the number of layers, while PN22L_100wet is an annealed composite membrane at 100 °C with loosely packed PFSA layers. While PC50 exhibited significant fluctuation in efficiency and discharge capacity decay due to severe vanadium ion crossover (Figure S11, Supporting Information), all ultrathin PFSA/PC composite membranes showed stable cell operation and even displayed superior capacity retention compared to commercial Nafion membranes (Figure S12, Supporting Information). Moreover, except the first three cycles at 40 mA cm⁻², PN14L_wet exhibited a higher CE as 97% across the entire current density range compared to the other membranes (Figure 4a), even though its vanadium ion permeability is 1.5 times higher than that of PN22L_100wet.^{[ 39 ]} Notably, similar to the commercial N211, PN14L_wet did not show a gradual decrease in CE during the initial seven cycles. This indicates that there was no electric field‐induced structural distortion in the microdefects embedded in PN14L_wet, as previously discussed. As shown in Figure 4b, PN14L_wet demonstrated better VE efficiency as well due to the additional proton pathways provided by the microdefects and thinner PFSA layers. At high current densities (150 and 200 mA cm⁻²), PN14L_wet achieved VE values ≈5% higher than those of the commercial N211. This trend aligns with the polarization curve analysis, where PN14L_wet (−9.96 × 10⁻⁴ V cm² mA⁻¹) exhibited lower voltage drop per unit current density compared to N211 (−10.30 × 10⁻⁴ V cm² mA⁻¹), indicating the lower ohmic resistance (Figure S13, Supporting Information). Consequently, despite being only 20 nm thick, the single cell with PN14L_wet outperformed the commercial one, which is 1000 times thicker, in EE across all current densities (Figure 4c).

Figure 4.

a) Coulombic efficiency (CE), b) voltage efficiency (VE), and (c) energy efficiency (EE) of the PC50/PFSA composite membranes.

As observed in our recent study,^{[ 37 ]} there was a discrepancy between the trends in capacity retention during long‐term VRFB operation and the trend in vanadium permeability for bulk Nafion membranes and densely packed PFSA membranes. Although the vanadium ion permeability was 100 times lower for the composite with densely packed PFSA layers, the capacity retention decay (≈60%) was similar across both bulk Nafion membranes and densely packed PFSA membranes. This result was presumably attributed to nanostructure distortion and the formation of macrodefects in the densely packed layers. In contrast, in this study, the PN14L_wet composite, which embeds microdefects, performed better and maintained its performance during long‐term charge–discharge cycling operation at 200 mA cm⁻², as shown in Figure 5a. Furthermore, similar to the trend in CE and vanadium ion permeability, PN14L_wet retained over 90% of its initial charged capacity, while N211 lost about half of its capacity within the first 100 cycles, as depicted in Figure 5b. Notably, even as the cycle number increased, PN14L_wet exhibited significantly lower capacity fade and consistently demonstrated higher EE than commercial Nafion membranes, highlighting its long‐term stability without changes in ion transport properties or physical damages to the membrane surface (Figures S14 and S15, Supporting Information).

Figure 5.

a) The long‐term performance of the VRFBs with N211 and PN14L_wet, and (b) discharge capacity retention at 200 mA cm⁻².

Overall, our findings emphasize the critical role of engineering ion‐selective microdefects in ultrathin membranes. Ultrathin PFSA membranes with highly ordered and dense structures exhibited reduced cell performance in comparison to ion transport properties obtained from the ex situ measurements due to nanostructure distortion and macropore formation during operation. Unlike ex situ measurements, which examine vanadium ion permeation and membrane resistance separately under controlled conditions, in situ environments involve complex processes like diffusion, migration, and convection, leading to discrepancies between ex situ and in situ membrane performance. These findings underscore two key considerations: first, pre‐swelling is essential to promote the formation of sufficiently hydrated ion channels, which mitigate ohmic resistance within the membrane during cell operation; second, while dense membranes limit vanadium ion crossover, they compromise proton conductivity, highlighting the need to balance selectivity and conductivity through tailored microdefects.

3. Conclusion

In this study, we demonstrate the highly ordered ultrathin PFSA membranes with engineered microdefects to enhance proton conduction under the VRFB cell operation while maintaining structural integrity. For the microdefects, the PFSA ionomer membranes are fabricated using LB deposition of PFSA monolayers with adjusted packing density through simple compression at the air/water interface. By stacking only 14 PFSA layers, sub‐25 nm ultrathin PFSA membranes with intentional microdefects are prepared, and size‐exclusive proton transport pathways are sufficiently established through pre‐swelling with electrolytes. Despite loosely packed layers and pre‐swelling, ultrathin PFSA membranes keep highly ordered channel morphology with a channel size of 3.46 nm. Measurements of the ion transport properties and VRFB single‐cell tests reveal that the microdefects in the ultrathin PFSA membranes suppress deformation of the channel morphology during cell operation with 80 times higher proton selectivity than Nafion 211, ultimately preventing a decline in cell performance. Finally, the sub‐25 nm ultrathin PFSA/PC composite membrane, consisting of 14 PFSA layers, exhibits superior cell performance compared to the 25‐µm‐thick commercial N211 membrane in cyclic tests and achieves higher energy efficiency (≈76%) along with high structural stability and excellent capacity retention after 500 cyclic tests at 200 mA cm⁻².

Conflict of Interest

The authors declare no conflict of interest.

Supporting information

Supporting Information

Kim J. Q., Lee Y., Lee J., Rho Y., So S., Choi S. Q., Engineered Microdefects in Nano‐Membranes for Enhanced Ion Selectivity and Membrane Durability in Vanadium Redox Flow Batteries. Small 2025, 21, 2500505. 10.1002/smll.202500505

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.