Continuous polymer films deposited on top of porous substrates using plasma-enhanced atomic layer deposition and molecular layer deposition

Molecular layer deposition (MLD) is a thin film deposition technique that allows for nanometer-scale control over the thickness of the deposited polymeric layer.^1–3 MLD uses sequential, self-limiting surface reactions to deposit organic and hybrid organic-inorganic films.⁴ MLD is analogous to atomic layer deposition (ALD) and can deposit polymeric films with conformality and atomic layer control.⁵ MLD processing has the potential to produce polymer films with a thickness of only a few nanometers while maintaining a high level of control over chemical composition and properties.^6,7

MLD films can be useful for the fabrication of polymer membranes.^8,9 An extremely thin polymer MLD layer can be deposited such that the thin film-porous substrate combination provides a functional high performance membrane.^10–14 An MLD-fabricated membrane can have high uniformity with minimal defects. Furthermore, transport resistance can be reduced by using a minimal MLD layer thickness and precise functionalization. Similar approaches using ALD have also been employed to fabricate composite polymeric membranes with enhanced properties.^15,16 These modified membranes may be useful for water purification.^17,18

Current membrane technology for applications such as water treatment utilizes composite membranes. These membranes consist of a thin dense polymer layer on top of a much thicker porous support. These thicker porous supports are typically polysulfone or polyethersulfone (PES) polymers. The thin dense polymer layer is often a cross-linked polyamide fabricated by interfacial polymerization.^19,20 Control of polymer layer thickness, interstitial voids, and nanopores is difficult using interfacial polymerization.²¹ 

Performing MLD on porous substrates presents a number of challenges. Porous substrates have high aspect ratios and previous ALD studies have shown that conformal deposition requires careful control of the reaction conditions.^22–24 In addition, conformal deposition may not be desired because conformal growth will eventually fill and block the pores of the porous support as shown in Fig. 1(a).^25,26 The conductance of the membrane will be progressively lowered as the pore diameters are reduced by the MLD cycles.^27–29 

This paper describes a method to deposit continuous polymer MLD films only on top of porous substrates using a combination of plasma-enhanced atomic layer deposition (PE-ALD) and MLD. PE-ALD is a subcategory of ALD that utilizes plasma-generated chemical species as a precursor in the ALD process.³⁰ PE-ALD can cap the pores of porous supports because the PE-ALD is limited to the top of the pores as shown in Fig. 1(b).^31–33 PE-ALD is confined to the top because the transmission of active plasma species into high aspect ratio pores is restricted by surface recombination loss within the pores.³⁴ Diffusion-limited kinetics can also localize the PE-ALD to the top when the reactive sticking coefficient of the active plasma species is high.^24,34 An MLD film can then be deposited on the continuous PE-ALD capping layer.

The strategy for depositing a continuous polymer film only on top of a porous substrate using PE-ALD and MLD is shown in Fig. 2. This strategy is implemented using porous anodic aluminum oxide (AAO) or PES support substrates with nanometer-scale pore diameters. The pores of the substrate are capped using Al₂O₃ PE-ALD. A polyamide MLD film is then deposited on the Al₂O₃ pore capping layer. The pore caps can then be removed using an etching solution applied from the backside of the membrane.³⁵ The result is a thin continuous MLD film attached to the top of the porous substrate. Alternative methods based on film transfer techniques can also achieve continuous polymer films on top of porous supports.³⁶ However, these polymers films are not bonded as securely and are difficult to scale.³⁷ 

One difficulty with the strategy shown in Fig. 2 is that the backside etching can also remove the Al₂O₃ layer underneath the polyamide MLD film depending on the etching time. Removal of the Al₂O₃ layer allows the polyamide MLD film to detach from the underlying support. The polyamide MLD film can be more firmly anchored to the porous support if there are cracks in the Al₂O₃ layer prior to the MLD. In this case, the MLD can coat both the Al₂O₃ layer and the porous support through the cracks in the Al₂O₃ layer. The MLD film deposited on the porous support acted to anchor the MLD film to the porous support.

Two porous substrates were examined in this study: ceramic AAO membranes and PES membranes. The AAO membranes (InRedox) were used as a model porous ceramic substrate.²⁴ The anisotropic AAO membrane is defined by a 1.5-μm-thick active layer with 10 nm pores and 60-μm-thick support layer with 100 nm diameter pores. The pore concentration and porosity in the active layer were 1.6 × 10¹¹ cm⁻² and 12%, respectively. Sample diameters were 13 mm.

PES ultrafiltration membranes (MilliporeSigma) were utilized as a model flexible porous polymer substrate. Like the AAO membranes, the PES membranes were asymmetric with an active layer containing pores. The range of pore sizes in the active layer was between 5 and 50 nm. The glass transition temperature of PES is 220 °C. The PES membranes had no fiber support.

The deposition of Al₂O₃ pore caps by PE-ALD was performed at room temperature in a custom-built remote plasma ALD reactor.³⁸ The remote plasma system was able to minimize sample etching and heating.³⁹ Samples were exposed to sequential doses of trimethylaluminum (TMA) (97%, Aldrich), nitrogen (N₂) (4.8 grade, AirGas), and oxygen plasma (4.4 grade O₂ gas, AirGas). The nitrogen dose served to purge the TMA from the reactor. No purge step was utilized after the oxygen plasma step because of the short lifetime of the reactive radical species in the oxygen plasma.⁴⁰ Exposure times of TMA/N₂/oxygen plasma were 3, 1, and 2 s, respectively, followed by pumping times of 45, 25, and 18 s, respectively. The static exposure pressures of TMA and N₂ were 0.75 and 1.0 Torr, respectively. Short exposures to the RF plasma are sufficient for generating active species and to minimize heat shrinkage of the polymer resulting from heating.⁴¹ All the reactions were performed at the ambient temperature of 23 °C.

The 200 W oxygen plasma was produced by a 2 kW RF Plasma Generator (Paramount HALO 3156300-000A, Advanced Energy). This remote plasma reactor had a reaction volume that was separated from the plasma source by a gate valve. The measured flow rate of O₂ through the plasma source was 3.0 SCCM. The pressure of oxygen during oxygen plasma exposures was 45 mTorr above base pressure.

The AAO samples were mounted normal to the plasma flux at a distance of 23 cm from the edge of the RF coils. The PES samples were placed parallel to the plasma flux on top of a removable metal tray that bisected the tube of the reactor. Silicon coupons were placed in the reactor next to the porous samples to serve as a witness of the PE-ALD growth.

A spatial MLD reactor was utilized to deposit the polymer MLD films on the porous substrates. The design and operation of this rotating cylinder reactor for ALD or MLD have been detailed elsewhere.^{25,26,42–44} Samples were mounted to the inner drum of the spatial MLD system using Kapton adhesive. The adhesive was applied to all sides of the porous samples to prevent backside deposition. The system was maintained at near-isothermal conditions within a custom convection oven that contained the dosing lines and valves.

The polymer MLD was conducted using m-phenylenediamine (MPD) (99%, Sigma Aldrich) and trimesoyl chloride (TMC) (98%, Sigma Aldrich) as the reactants.⁴² The two precursor zones in the spatial MLD reactor were individually dosed with MPD and TMC.⁴² All chemicals were stored in stainless steel containers within the convection oven. Each precursor was connected to the reactor by two bellows globe valves on either side of a bellows needle valve for metering flow.

The MLD reactions were performed at a drum rotation rate of 5 rpm. This rotation rate corresponds to a 1.2 s residence time in the reactive dose zones and a 4.8 s residence time in the purging zones. The reaction temperature was 130 °C. Nitrogen purge gas (4.8 grade, Airgas) was maintained at a flow rate of 200 SCCM and established a base pressure of 1.0 Torr. Precursor flow was manually adjusted to add 200 mTorr above the base pressure. This pressure was measured for each dosing line. A silicon coupon was also placed in the reactor to serve as a witness substrate for thickness measurements.

The Al₂O₃ PE-ALD on AAO samples was performed to study the ability of Al₂O₃ PE-ALD to cap the pores and then the ability of the NaOH etching to remove the pore caps. After the Al₂O₃ PE-ALD films were deposited on the AAO porous supports, the backside of each pore-capped AAO sample was etched with a 0.03M NaOH solution. This solution was prepared with ultrapure water and NaOH pellets (97%, Fisher Chemical). The etching process was designed to etch only from the backside of the samples and ensure repeatable etching conditions. AAO samples with 20 PE-ALD Al₂O₃ cycles were placed in an AAO holder cell (InRedox). The front side faced a sintered porous metal support disk.

A crossflow of ∼0.1 ml/min of the NaOH solution was maintained on the backside of the AAO sample. This crossflow ensured constant solution concentration. The crossflow was pressurized by an N₂ headspace above the reservoir of the NaOH solution. The N₂ pressure was 5 psi. After the designated times, the NaOH solution was cleared by a flow of N₂ gas. Etching was performed at the ambient temperature of 23 °C. Following each etch, the samples were rinsed in ultrapure water and then dried in an oven at 70 °C.

After the PE-ALD and MLD films were deposited on the PES porous supports, samples were prepared using a punch to define a 50 mm diameter. Samples were immersed in a stirred bath of 100 ml of 0.03M NaOH solution. The PES samples were fully immersed in the etching solution with the MLD layer on top of the Al₂O₃ pore caps. The MLD layer protected the Al₂O₃ capping layer from etching from the top side. Samples were etched for 20 or 40 min and then rinsed with ultrapure water. The samples were air dried at room temperature before measuring the gas permeance.

Spectroscopic ellipsometry measured film thicknesses on the silicon witness coupons using a spectroscopic ellipsometer (M-2000, J.A. Woollam Co., Inc.) and modeling software (completeease, J.A. Woollam Co., Inc.). Measurements of the Al₂O₃ film thicknesses were performed at a 70° incident angle. The spectra were fit using a Cauchy model. Measurements of the polyamide MLD film were performed at three angles: 50°, 60°, and 70°. The spectra were fit with two Gaussian oscillator curves and an anisotropic index difference in the z axis with parameters developed from multisample analysis.

Gas permeance measurements were performed using a custom-built system comprised of a compressed nitrogen tank sourcing pressure to a dead-end membrane flow cell. Commercial membrane cells were used to hold the AAO (InRedox, AAO Membrane Holder) and PES samples (Sterlitech, HP4750). Pressure was controlled with a regulator and measured with a pressure transducer (OMEGA, PX309-300G5 V). Flow rate was measured with a mass flow controller in the read operation (Alicat, MC-100SCCM-D).

To improve the handling and integrity of AAO samples, each AAO disk was adhered to a ring of Kapton adhesive. These rings had a 10 mm inner diameter that overlapped and sealed to the 13 mm AAO disks. The outer diameter of these Kapton rings was 25 mm and provided a sealing surface for the O-ring of the membrane holder to prevent the fracture of the brittle AAO material. For PES, gas permeance measurements were conducted on 50 mm diameter samples. During these gas permanence experiments, the samples were backed by a sintered porous metal disk with the active layer facing upstream. Mass flow measurements were performed using a pressure drop of 100 psi. Gas permeance was calculated as volumetric flow rate per pressure drop per unit area.

Images of fractured samples were obtained with scanning electron microscopy (SEM) using a Hitachi SU3500 at 30 kV accelerating voltage and a working distance of 11–28 mm. A 1 nm layer of Au was sputtered onto the samples to minimize the effects of sample charging. All other SEM imaging was performed using a Zeiss GeminiSEM 500 at 3 kV accelerating voltage and a working distance of 3–5 mm. A 1–3 nm layer of Au–Pd was sputtered onto these samples to minimize the effects of sample charging. Additional imaging was performed using an optical microscope (Micromanipulator, 6000) with standard optics (Mitutoyo, Plan Apo Objective).

Figure 3 shows the growth of Al₂O₃ PE-ALD films on silicon witness substrates using TMA and oxygen plasma at an ambient temperature of 23 °C. Linear growth was observed with a growth per cycle of 2.4 Å/cycle. This growth rate is higher than the earlier reported growth rate of 1.7 Å/cycle at 25 °C.⁴⁵ There was a short nucleation period of two PE-ALD cycles. The oxygen plasma exposures were limited to 2 s. Longer plasma exposure times led to surface corrosion effects. Short exposure times also assisted in limiting the depth of the PE-ALD deposition within the pores.^24,31,32,34

The PES sheets with dimensions of 25 × 5 cm were too large relative to the size of the cylindrical reaction chamber to be positioned normal to the plasma flux. Consequently, the PES sheets were loaded lengthwise into the reactor with their surfaces parallel to the plasma flux. To test for uniform deposition, 15 silicon coupons were placed in a grid at various points along the axis of the reactor and off-axis at the midplane of the reaction tube. After 85 PE-ALD cycles, the relative standard deviation of measured thicknesses for the various locations in the reactor was 1.5% with no apparent trends. These results indicate that the Al₂O₃ PE-ALD films should be uniform on the surface of the PES sheets.

The TMA precursor is soluble in PES and can infiltrate the bulk of the membrane material.⁴⁶ However, deposition in the polymer membrane is unlikely because oxygen radicals cannot easily diffuse into PES. Radical species experience recombination on surfaces.³⁴ Rather than absorbing into the polymer, oxygen plasma species induce bond scissions at the surface of PES substrates.^47,48 This study observed that a PES sample exposed to oxygen plasma for many minutes was severely cracked, thinner, and brittle.

Al₂O₃ PE-ALD was performed on AAO samples to evaluate the ability of Al₂O₃ PE-ALD to cap the pores of the AAO samples. Figure 4 shows the relationship between the gas permeance and the number of Al₂O₃ PE-ALD cycles. There was a progressive reduction in gas permeance with increasing number of PE-ALD cycles. These results illustrate that Al₂O₃ PE-ALD can tune the conductance of the AAO samples. After 16 PE-ALD cycles, the flow of N₂ through the AAO membrane was below the measuring capacity. The measured Al₂O₃ deposition thickness on the silicon witness wafers after 16 PE-ALD cycles was 3.5 nm.

If the AAO membranes had a pore diameter of 10 nm, then the pores should have required an Al₂O₃ PE-ALD thickness of 5 nm to seal the pores. In contrast, Fig. 4 shows that the pores closed after 3.5 nm of the measured PE-ALD growth. This difference may be attributed to inaccuracies in the measurement of the pore diameters in the AAO membrane. Furthermore, a nucleation period was observed during the first two PE-ALD cycles on the silicon substrates in Fig. 3. This nucleation period may not exist for the AAO substrates. Consequently, the PE-ALD films on AAO may have been slightly thicker than on the witness silicon wafers.

Figures 5(a) and 5(b) display SEM images of the active side of unprocessed AAO membranes and AAO membranes capped with 20 Al₂O₃ PE-ALD cycles, respectively. A visual change in the surface morphology is apparent between these two samples. The capped AAO surface had fewer visible openings than the bare AAO surface. Although the capped surface still appears to have open surface pores, the measured gas permeance shown in Fig. 4 was below the detection limit after 20 PE-ALD cycles, indicating full pore closure.

An AAO membrane coated with 20 Al₂O₃ PE-ALD cycles was then etched from the backside using a 0.03M NaOH solution to remove the Al₂O₃ PE-ALD pore caps from the top of the AAO membrane. Figure 5(c) shows an SEM image of the top side of the AAO membrane sample after 3 min of etching. The morphology of the active layer was similar to the morphology of the AAO sample after 20 Al₂O₃ PE-ALD cycles shown in Fig. 5(b). The active side was still unaffected because no breakthrough has occurred after 3 min of etching.

Figure 6 displays the effect of etch time on the gas permeance of the AAO sample coated with 20 Al₂O₃ PE-ALD cycles. The dotted line shows the permeance of the unprocessed AAO sample with no PE-ALD cycles. The gas permeance increased with the etch time. After a 300 s etch time, the measured permeance of the AAO sample coated with 20 Al₂O₃ PE-ALD cycles reached the permeance of an unprocessed AAO sample. The permeance then remained at the permeance of an unprocessed AAO for even longer etch times. The relatively steep slope of the gas permeance over the first 300 s etch times is attributed to the removal of the Al₂O₃ PE-ALD pore caps. The slight increase of the gas permeance for longer etch times may be caused by the widening of the pores in the original AAO sample.³⁵ 

These results confirm that the Al₂O₃ PE-ALD layer was limited to the top of the porous AAO substrate. In addition, there was minimal Al₂O₃ PE-ALD growth inside the pores. The Al₂O₃ PE-ALD pore capping layer could be controllably removed without compromising the original AAO sample. Based on these results, this method should be useful for the addition of Al₂O₃ PE-ALD pore caps and their removal from porous substrates using backside etching.

PES membranes were also coated with Al₂O₃ PE-ALD to cap the pores on the surface of the PES membrane. Figure 7 displays a progression of SEM images that reveal the effect of the Al₂O₃ PE-ALD cycles on the PES membrane. The unprocessed PES membrane is shown in Fig. 7(a). This unprocessed PES membrane had surface pores with diameters ranging from 5 to 50 nm. In comparison with the AAO substrates shown in Fig. 5(a), the pores in the PES membrane had a broader size distribution. Figure 7(b) shows that very few surface pores were visible after only ten cycles of Al₂O₃ PE-ALD. With subsequent PE-ALD cycles, the surface pores visibly disappeared with subsequent PE-ALD cycles as illustrated in Figs. 7(c)–7(e) after 20, 70, and 150 PE-ALD cycles, respectively.

The Al₂O₃ PE-ALD cycles progressively reduced the gas permeance of the PES membranes. Figure 8 shows the relationship between gas permeance and the number of Al₂O₃ PE-ALD cycles applied to the PES samples. The gas permeance decreased with the number of Al₂O₃ PE-ALD cycles. After 60 or more Al₂O₃ PE-ALD cycles, the measured gas permeance was <2 × 10⁻³ cm³/(cm² s Pa). These results are different from the results for the AAO membranes shown in Fig. 4. The pores in the AAO membrane became fully sealed after 16 Al₂O₃ PE-ALD cycles at an Al₂O₃ PE-ALD thickness of 3.5 nm. In contrast, the permeance of the PES samples never fell below the detection limit, even after 150 Al₂O₃ PE-ALD cycles.

This lack of complete blockage of the gas flow is attributed to the cracking of the Al₂O₃ PE-ALD film. This cracking occurred when placing the Al₂O₃-topped PES samples in the measurement cell. Due to the compressive force on the samples from the O-ring seal, fracture of the brittle, ceramic Al₂O₃ PE-ALD layer occurred against the flexible PES substrate under compression.^49,50 Based on the SEM micrographs in Fig. 7, full pore capping of the PES occurred after >20 PE-ALD cycles. However, the film fracture from the compressive O-ring seal leads to nonzero gas permeance.

The fracture of the Al₂O₃ PE-ALD layer can be observed by SEM images. Figure 9 shows the SEM image of a PES sample that was prepared with 70 Al₂O₃ PE-ALD cycles. Figure 9(a) shows the SEM image of the film fracture caused by compression against the O-ring. Figure 9(b) displays an SEM image after the sample was fractured as a result of cutting. Fracturing of the Al₂O₃ PE-ALD layer on the AAO substrates was not observed because the AAO substrate and the Al₂O₃ PE-ALD layer have similar flexural properties.

MLD films were also grown on top of PES membranes coated with 40 Al₂O₃ PE-ALD cycles. The polyamide MLD films had a thickness of 290 nm after 774 cycles at 5 rpm at 130 °C. The growth rate of the polyamide MLD film is 3.7 Å/cycle. This growth rate compares with the growth rate of 4.5 Å/cycle reported earlier using the same reactants at 20 rpm at 115 °C.⁴² 

Optical micrographs of these composite membranes revealed a number of fractures as displayed in Fig. 10. Most of the fractures likely resulted from differences in the thermal expansion between the Al₂O₃ layer and the underlying PES membrane.^49,51 Al₂O₃ has a relatively low thermal expansion coefficient of 4.2 × 10⁻⁶ cm/cm K.⁵⁰ PES has a much larger thermal expansion coefficient of ∼54 × 10⁻⁶ cm/cm K.^41,52 The thermal expansion of the PES substrate upon heating for MLD at 130 °C leads to tensile stress in the Al₂O₃ layer.^49,51

Earlier work studied thermal Al₂O₃ ALD films that were grown on Teflon at higher temperatures and experienced a compressive stress upon cooling.^49,51 This compressive stress resulted from thermal expansion mismatch between the Al₂O₃ ALD film and the Teflon substrate. A tensile stress for an Al₂O₃ ALD film grown at low temperature on a polymer and then heated to a higher temperature is consistent with a compressive stress for an Al₂O₃ ALD film grown at higher temperature on a polymer and then cooled to a lower temperature.^49,51

The Al₂O₃ pore caps were removed from the polyamide MLD-Al₂O₃ PE-ALD-PES composite membranes using a 20 min exposure in a 0.03M NaOH solution. The thickness of the Al₂O₃ PE-ALD layers on these PES membranes was 16 nm. The optical micrographs revealed that the MLD layer began to separate from the PES membrane after the dissolution of the Al₂O₃ pore caps. The optical micrograph displayed in Fig. 11 reveals that the polymer MLD film was buckled after detaching from the underlying PES membrane. An inspection of the optical micrograph also revealed that the polymer MLD film was only anchored to the PES substrate at the lines of fracture in the initial Al₂O₃ PE-ALD film.

The polyamide MLD film is only attached to the PES substrate through the fractures in the original Al₂O₃ PE-ALD layer. The polyamide MLD layer was able to anchor itself securely by deposition inside the pores of the PES substrate. This explanation is supported by the observation that the polyamide MLD-Al₂O₃ PE-ALD-PES composite membranes displayed no gas permeance prior to removal of the Al₂O₃ pore caps by dissolution using the 0.03M NaOH solution. Although many fractures existed in the Al₂O₃ PE-ALD layer, polyamide MLD filled the pores in the PES substrate through the cracks in the Al₂O₃ PE-ALD layer and inhibited any gas flow.

The water permeability of these polyamide MLD-Al₂O₃ PE-ALD-PES composite membranes was measured using a pressure of 225 psi. Optical micrographs after testing showed that the buckled regions of the polyamide MLD film tore and pulled away from the underlying PES membrane. In addition, the regions in the continuous polyamide MLD film with greater distance between the anchoring lines were more susceptible to failure. This behavior suggested that the polyamide MLD film would be more robust if there were more fracture lines in the initial Al₂O₃ PE-ALD layer and more anchoring to the underlying PES substrate.

The anchoring of the polyamide MLD film was dependent on the number of fractures in the original Al₂O₃ PE-ALD layer. Consequently, the Al₂O₃ PE-ALD layer was intentionally fractured prior to MLD. By bending the Al₂O₃ PE-ALD-PES samples over a fixed radius, a regular pattern of fractures was produced in the Al₂O₃ PE-ALD layer.⁴⁹ Bending over an 11.4 mm diameter mandrel produced parallel fractures at a spacing of ∼100 μm. Polyamide MLD on these fractured Al₂O₃ PE-ALD layers followed by etching produced the optical micrograph shown in Fig. 12. The anchor lines for the polyamide MLD film are evident at a spacing of ∼100 μm.

Fractures in the Al₂O₃ PE-ALD layer could also be produced in both the x and y directions of the PES substrate. Bending over a 3.2 mm diameter mandrel along both sheet axes produced orthogonal fractures at a spacing of ∼50 μm in the Al₂O₃ PE-ALD layer. Polyamide MLD on these fractured Al₂O₃ PE-ALD layers followed by etching produced the optical micrograph shown in Fig. 13. The anchor lines for the polymer MLD layer formed a distinctive crosshatched pattern.

The polyamide MLD-Al₂O₃ PE-ALD-PES composite membranes with the greater number of anchoring points were the most robust. Samples with larger spacing between anchor points experienced failure more frequently when subjected to pressure. However, the sample shown in Fig. 13 with a polyamide MLD film thickness of 290 nm was impermeable to both nitrogen and water up to 225 psi. Perhaps thinner MLD films are needed for measurable permeability. If permeable MLD films can be demonstrated, then this technique has the potential for producing robust, thin film composite membranes. Such MLD membrane systems could prove useful for a number of important pressure-driven separation applications such as reverse osmosis, gas separation, or pervaporation.

A method to deposit continuous polymer films on top of porous substrates was developed based on PE-ALD and MLD. PE-ALD was first used to cap the pores of the porous substrate. An MLD polymer film was then deposited on the PE-ALD capping layer. This method was demonstrated using AAO and PES porous substrates. This new strategy should be useful for adding polymer MLD films to the surface of porous substrates to create new composite membranes for separation applications.

The experiments on porous AAO substrates revealed that Al₂O₃ PE-ALD film growth could cap the pores on top of AAO substrates. Following Al₂O₃ PE-ALD film growth, a continuous surface was available for the deposition of a polymer MLD film. The pores could then be reopened by dissolving the Al₂O₃ pore caps with sodium hydroxide. Gas permeance measurements confirmed the closure of the pores in AAO membranes coated with Al₂O₃ PE-ALD films and the reopening of the pores versus NaOH etch time.

The pores in PES membranes could also be closed using Al₂O₃ PE-ALD film growth. However, gas permeance measurements indicated that Al₂O₃ PE-ALD film growth did not completely limit gas transmission. The nonzero gas permeance was traced back to the fracture of the Al₂O₃ PE-ALD films on the PES membranes during sample testing. Fracture occurred due to compression of the Al₂O₃ PE-ALD/PES membrane when making the O-ring seal for subsequent testing.

The fracture of the Al₂O₃ PE-ALD film also provided a mechanism to anchor the polyamide MLD film to the PES membrane. The Al₂O₃ film fracture allowed the MLD film to attach firmly to the PES porous substrate by MLD deposition in the pores of the PES substrate. After dissolving the Al₂O₃ film, the polyamide MLD film remained anchored to the PES substrate through the original fracture lines in the Al₂O₃ PE-ALD film. The Al₂O₃ PE-ALD film could be intentionally fractured to provide more fracture lines for anchoring the MLD film to the PES substrate. The stability of the polyamide MLD film improved with a smaller spacing between initial fractures in the Al₂O₃ PE-ALD film.

This research was initiated by the U.S. Department of Energy (DOE) (No. DE-EE0005771) through a subcontract from GE Global Research. The authors thank Robert Gemmer of the Department of Energy for many useful discussions. Continued support for this research was provided from the Laboratory Directed Research and Development Program at Sandia National Laboratories, a multimission laboratory managed and operated by National Technology and Engineering Solutions of Sandia LLC, a wholly owned subsidiary of Honeywell International Inc. for the U.S. Department of Energy’s National Nuclear Security Administration under Contract No. DE-NA0003525. Further support was provided by the Membrane Science, Engineering and Technology Center (NSF IUCRC Award IIP No. 1624602) at the University of Colorado Boulder. The authors thank James Redmond, Susan Rempe, Leo Small, Eliot Fang, and John Emery of Sandia National Laboratories for helpful technical discussions. In addition, the authors thank Dmitri Routkevitch of InRedox for supplying AAO samples and Christina Carbrello of MilliporeSigma for supplying PES samples. The RF plasma system was provided by Dan Carter at Advanced Energy.