I. Abstract of the Paper’s Core Research

This study has developed a collision‑atomization‑based aerosol‑assisted printing system. Using an HPAN ultrafiltration support membrane as the substrate, cationic PEI and anionic PSS polyelectrolyte aqueous solutions were prepared, and PEM nanofiltration membranes were fabricated via layer‑by‑layer alternating deposition of controllable ultrafine aerosols. The study systematically elucidated the film‑formation mechanism throughout the entire process and validated the water‑separation performance. The key findings are as follows:

Ultrafine Uniform Aerosol Deposition System

Impact atomization generates a narrow‑distribution population of 3–5 μm droplets, with adjustable air pressure enabling precise control over droplet size and impact kinetic energy. A mobile scanning nozzle eliminates the coffee‑ring and surface‑roughness defects associated with fixed‑nozzle spraying, resulting in a significant reduction in film‑surface roughness and markedly superior deposition uniformity compared to conventional spray guns.

Elucidation of the Three-Step Complete Film-Forming Mechanism

For the first time, the aerosol‑mediated film‑formation process has been systematically divided into three stages: (1) inertial impaction followed by uniform spreading of droplets; (2) rapid solvent evaporation driving the migration and aggregation of polyelectrolytes to form a monolayer film; and (3) electrostatic adsorption of cations and anions leading to molecular interpenetration and the construction of a dense sieving network. XPS depth profiling confirms that, within the multilayer film, PEI and PSS are highly intertwined, with no well‑defined layered structure.

Precise tuning of printing parameters for enhanced separation performance

The membrane thickness and pore size can be linearly tuned by adjusting the number of printing passes—1 to 5 for single-layer deposition and 1 to 2 layers for double-layer deposition: increasing the number of passes results in a denser membrane, with the average pore size decreasing from 0.56 nm to 0.40 nm; the permeate flux is simultaneously controllable within the range of 5 to 60 L·m⁻²·h⁻¹·bar⁻¹. The double-layer (PEI₃PSS₃)₂ membrane achieves a bisphenol A rejection rate exceeding 80%, outperforming the commercial NF270 nanofiltration membrane.

The advantages of efficient, low-cost preparation are becoming increasingly evident.

64.8 cm² functional membrane requires only 6 mL of polyelectrolyte solution, and the entire preparation process takes just 18 minutes; compared with the conventional immersion method that lasts several hours, reagent consumption is reduced by more than 90%, and processing efficiency is improved by a factor of ten.

Long-term stable operating performance

In a 96-hour continuous cross-flow filtration test, neither the water permeation flux nor the rejection rates for salts and trace contaminants showed any significant decline; the interpenetrating membrane structure confers excellent resistance to compaction and fouling.

II. Complete Process Flow for PEM Nanofiltration Fabricated via Aerosol-Assisted Printing

1) Substrate pretreatment: Commercial PAN ultrafiltration membranes were subjected to alkaline hydrolysis with 2 M NaOH to obtain a hydrophilic HPAN substrate, followed by repeated rinsing with deionized water to remove residual alkali.

2) Preparation of polyelectrolyte ink: a pure water system containing 8 g/L PEI and 16 g PSS, with no additional high-viscosity additives, suitable for collision atomization.

3) AAP aerosol system commissioning: collision nebulizer pressure 0.5–1.5 bar, nozzle-to-substrate distance 15 mm, scan speed 1500 mm/min, raster reciprocating printing;

4) Alternating-layer printing: Print 1–5 layers of single-layer PEI, allow them to dry, then print the corresponding number of PSS layers; repeat this process to form a double-layer structure in one or two cycles.

5) Forms by natural drying at room temperature, requiring no high-temperature heat treatment;

6) Comprehensive characterization: AF/SEM surface morphology, XPS depth‑profile elemental analysis, zeta potential, and cross‑flow filtration performance testing (for neutral sugars, bisphenol A, antibiotics, and inorganic salt rejection).

III. Five Core Process Advantages of Aerosol-Assisted Printing

A leapfrog improvement in preparation efficiency

Traditional dip-coating of a single film takes several hours, whereas aerosol printing can fabricate an entire set of multilayer functional layers in just over ten minutes, making it well-suited for continuous roll-to-roll mass production.

Raw material consumption has been significantly reduced.

Localized, precise deposition eliminates the need to immerse large reservoirs; for the same membrane area, electrolyte consumption is reduced to less than one-tenth of that in dip‑coating, thereby lowering the production cost of water‑treatment membranes.

Ultra-fine mist droplets eliminate coffee-ring defects.

3–5 μm microdroplets evaporate rapidly, and when combined with mobile grating scanning, the membrane surface is smooth and uniform, free of edge buildup or surface irregularities, resulting in highly consistent separation performance.

Performance is digitally precise and adjustable.

By simply adjusting the number of printing passes and the number of bilayer cycles, the membrane thickness, pore size, and water permeation flux can be linearly tuned, enabling the on-demand fabrication of either loosely structured macromolecular filtration membranes or dense membranes for retaining trace contaminants.

Compatible with irregular-shaped / curved substrates

Non-contact droplet deposition enables the uniform fabrication of separation layers on surfaces ranging from planar membranes to hollow fibers and non‑standard filtration supports, thereby expanding the design space for membrane modules.

IV. Five Core Technological Innovations of This Study

Innovation 1: A Comprehensive Mechanistic Model for Aerosol-Based Three-Step Film Formation

The system elucidates the entire process of droplet impact, single‑polyelectrolyte layer formation, and interpenetration of cationic and anionic molecules, thereby establishing a dual‑mechanism film‑formation theory that integrates inertial deposition with electrostatic entanglement.

Innovation 2: A standardized mobile scanning process for eliminating coffee-ring artifacts

Addressing the challenge of localized buildup with fixed spray guns, raster‑based uniform scanning achieves a globally consistent, ultra‑thin separation layer, significantly enhancing batch-to-batch film stability.

Innovation: Three-Step / Double-Layer, Dual-Dimension Pore Size Control Method

A single device requires no raw-material changes and, by continuously adjusting printing parameters, achieves a nanopore size of 0.40–0.56 nm, meeting water‑treatment needs across all applications.

Innovative New Structure of a Four-Molecule Interpenetrating PEM Membrane

Aerosol‑driven dynamic deposition promotes deep interpenetration of PEI and PSS, yielding a membrane that differs from conventional well‑defined layered dip‑coated films, with markedly improved resistance to compaction and long‑term filtration stability.

Innovative Low-Cost, Scalable Additive Manufacturing Paradigm for PEM Membranes

It overcomes the bottleneck of mass production in dip-coating, with low equipment investment and a fast cycle time, offering a completely new pathway for the industrialization of polyelectrolyte nanofiltration membranes.

V. Current Industry Challenges and Corresponding Solutions

Challenge 1: The conventional dip-coating process is time-consuming, results in significant chemical waste, and cannot support continuous mass production.

✅ Aerosol direct writing achieves rapid formation in just a few minutes, significantly reducing reagent consumption and enabling seamless scaling to roll-to-roll production lines.

Challenge 2: Conventional large‑diameter spray guns produce coarse droplets, and coffee‑ring defects lead to variability in membrane performance.

✅3–5 μm ultra-fine atomization + dynamic moving scanning ensures uniform surface coverage with no localized buildup, significantly reducing batch-to-batch variability.

Challenge 3: The pore size of nanofiltration membranes is difficult to fine-tune continuously, and a single formulation is typically optimized for only one specific water quality.

✅ Print‑path linear control adjusts membrane density, enabling one‑click switching between loose and dense membranes to suit various treatment scenarios involving humic acids and trace organic compounds.

Challenge 4: Conventional layered membranes are prone to delamination and rapid flux decline under long-term high-pressure filtration.

✅ Aerosol deposition induces deep molecular interpenetration, resulting in an integrated membrane layer and stable long-term performance over 96 hours.

Challenge 5: Dip coating is limited to flat substrates, and the fabrication of hollow fiber and other shaped membranes is highly challenging.

✅ Non-contact aerosol mist flow enables uniform coating of the separation layer on curved surfaces and the outer walls of hollow fibers.

VI. Future Diversified Industrial Application Scenarios

1 Advanced Treatment of Trace Contaminants in Municipal Drinking Water

Removal of endocrine disruptors such as bisphenol A and antibiotics, with mass production of low-flux, dense nanofiltration membranes.

2 Separation of dyes and humic acids from industrial wastewater

Large-pore, loosely structured PEM membrane with high water permeation flux for pretreating industrial organic wastewater.

3 Seawater / Brackish Water Softening and Desalination

A negatively charged selective nanofiltration membrane efficiently retains divalent sulfate anions, thereby reducing softening energy consumption.

4 Mass production of hollow-fiber nanofiltration membrane elements

Uniform functional-layer printing on the outer surface of aerosols, along with the fabrication of compact, integrated filtration modules.

5 High-Throughput Membrane Formulation Screening Platform for Laboratories

Rapidly fabricate gradient membranes with varying layer numbers, conduct batch testing of pore size and rejection performance, and shorten the R&D cycle for new materials.

Image and Text Guide

• (a)   AAP system schematic: Two Collison atomizers nebulize aqueous solutions of PEI (8.0 g/L) and PSS (16.0 g/L), generating micron‑scale droplets; the resulting aerosol flows through a 14 mm stainless steel exit nozzle and is directed vertically toward the substrate. A 3D printer controls the nozzle to raster‑scan the entire substrate at 1500 mm/min with a 7 mm interline spacing, covering 64.8 cm² per layer in approximately 1.5 minutes. The nozzle is positioned 15 mm above the substrate, operating at a pressure of 0.5–1.5 bar.
• (b)   Dimensions of membrane structure regulation: Number of scans n (= deposition quality) — Each complete scan deposits one layer; 1–5 scans yield a PEIₙ or PSSₙ layer. Double-layer number — Cyclic deposition of PEI/PSS yields films consisting of 1–2 bilayers, such as (PEIₙPSSₙ)₁/(PEIₙPSSₙ)₂.
• Base : A polyacrylonitrile (PAN) ultrafiltration membrane with a 400 kDa molecular weight cut-off was hydrolyzed with 2.0 M NaOH to yield a negatively charged HPAN support, which facilitates the electrostatic adsorption of the first layer of cationic PEI.

• (a,b)   Volume size distribution of PEI/PSS droplets under different atomization pressures (0.50–1.50 bar): the particles are predominantly concentrated within a narrow range of 3–5 μm, which facilitates uniform deposition.
• (c)   Median particle diameter Dv50: When the pressure increases from 0.5 to 1.5 bar, PEI droplets decrease from 3.6 to 2.4 μm, while PSS droplets shrink from 3.1 to 2.3 μm—higher pressure yields finer droplets. At the same pressure, PEI droplets are larger (the two solutions have viscosities of 5 and 20 mPa·s, respectively; the more viscous solution is easier to atomize into finer droplets).
• (d)   Nozzle exit droplet velocity: from 0.7 to 1.3 m/s as pressure increases from 0.5 to 1.5 bar (at 1 bar, the exit velocity is approximately 1.1 m/s, decreasing to about 0.6 m/s at a distance of 5 mm from the substrate).
• (e)   COMSOL (k‑ε/RANS) simulation of the velocity field at 1 bar: The gas flow exits the nozzle vertically downward, then abruptly turns and curls upward upon encountering the substrate—this curved flow results in higher velocities and greater deposition on the right side (the curved side of the nozzle), whereas in the moving‑scan mode, the nozzle sweeps across the entire substrate, yielding uniform deposition.

• (a)   Schematic of PE layer formation: continuous spraying → additional droplet impact and spreading → increased scanning cycles → formation of a complete polyelectrolyte thin film.
• (b)   Deposition particle size (inset in the AFM phase map): The peak diameter of PEI particles is approximately  14 nm , PSS approximately  12 nm According to the mass‑conservation calculation based on one droplet–one particle (ODOP), the expected value is 574/651 nm; however, the measured value is much smaller than the calculated one, indicating that the droplet does not fully evaporate upon reaching the substrate but rather during the impact. Unfold And distribute the PE across the surface, forming smaller particles.
• (c)   Layer thickness increases linearly with the number of scans: for the PEI layer, y = 42.8x; for the PSS layer, y = 45.5x (approximately 44 nm per scan, R² = 0.96–0.99). SEM measurements yield a PEI₃ layer thickness of 109 ± 26 nm and a PSS₃ layer thickness of 96 ± 17 nm, falling between the “solid‑film model” (98/56 nm) and the “close‑packed spheres model” (188/107 nm), indicating that the actual layers are closely packed.
• Deposition mechanism : Droplet impact is dominated by inertia (Stokes number Stk = ρₚdₚ²V/(18μL)); increasing the droplet velocity from 0.7 to 1.3 m/s raises the PEI deposition efficiency from 6.6% to 12.0% (whereas PSS achieves only 1.2%, owing to smaller droplets and lower momentum).

• (a)   ATR-FTIR spectrum: The peak at 1690 cm⁻¹ (C=O of COO⁻ generated by PAN hydrolysis) diminishes with PEI/PSS deposition; the peaks at 1178, 1124, 1035, and 1007 cm⁻¹ (asymmetric/symmetric stretching of PSS’s S=O) confirm successful PSS deposition.
• (b)   Surface and Cross-Section FESEM: As deposition proceeds, pore size decreases and the surface becomes denser and free of defects; cross-sectional thicknesses are 109 ± 26 nm for PEI₃, 116 ± 19 nm for (PEI₃PSS₃)₁, 217 ± 5 nm for PEI₃PSS₃PEI₃, and 263 ± 4 nm for (PEI₃PSS₃)₂. Adding a PSS layer does not double the thickness (the increase is ≤20%). 。
• (c)   XPS depth profiling of the N/S ratio: (PEI₂PSS₂)₂ remains approximately constant at 1.7 over 0–15 s → indicating uniform interpenetration of PEI and PSS, forming a homogeneous PEC; (PEI₄PSS₄) increases from 2.0 to 2.2 → reflecting weakened interlayer PSS penetration; after 20 s, it abruptly rises to 2.7/5.0 → reaching the HPAN substrate.
• (d)   Schematic of PEM assembly: Adjacent PE layers with opposite charges undergo strong interpenetration, forming polyelectrolyte complex (PEC) layers (1 PEC ≈ 1 bilayer thickness), rather than the ideal, orderly layer-by-layer stacking.

• (a)   Pure water permeance (PWP): For the single-layer PEI, there was no significant reduction in flux due to water swelling; however, after alternating PSS/PEI layers, the permeance decreased with increasing number of bilayers. During three scan cycles, the permeance of the bilayer membrane dropped from 12.0 to 5.4 L·m⁻²·h⁻¹·bar⁻¹, while for the bilayer membrane, it decreased from 13.3 to 3.3 LMH·bar⁻¹ between the first and fifth scan cycles.
• (b,c)   For single- or double-layer membranes, the intrinsic rejection coefficient Ri for neutral organic solutes (erythritol, xylose, glucose, and sucrose) increases with increasing molecular weight, while the molecular weight cut-off (MWCO) decreases as the scan rate increases.
• (d)   Average pore radius: for the monolayer membrane, 0.56 → 0.46 nm; for the bilayer membrane, 0.56 → 0.40 nm (calculated via scanning from 1 to 5, based on a steric‑hindrance pore model).
• (e)   BPA rejection: 12.7% → 67.9% for single-layer membranes; highest for double-layer membranes.   80.8% (1→5 scanning: 38.7%→80.8%), significantly outperforming the commercial NF270 (approximately 40%).
• (f)   (PEI₃PSS₃)₂ exhibits selective retention of micropollutants: AP removal is only 45.1% (dominated by size exclusion, with a MWCO ≈ 294 Da), whereas CIP and OFX both exceed 90% → indicating a defect-free membrane; salt rejection rates are 22% for NaCl, 34% for MgSO₄, 58% for Na₂SO₄, and 13% for MgCl₂ (governed by electrostatic interactions).