Improving oxygen barrier and reducing moisture sensitivity of weak polyelectrolyte multilayer thin films with crosslinking

Abstract

Thin films of polyethylenimine (PEI) and poly(acrylic acid) (PAA), deposited using layer-by-layer assembly, were studied to understand the influence of various crosslinking methods on their oxygen and water vapor barrier. Glutaraldehyde (GA), 1-[3-(dimethylamino)propyl]-3-ethylcarbodiimide methiodide (EDC) and thermal crosslinking were evaluated with respect to film thickness and barrier properties. The thickness of an 8 bilayer PEI/PAA assembly decreased from 640 to 551 nm after crosslinking with a 0.1 M aqueous GA solution and obtained an oxygen permeability one order of magnitude better than an SiO_x thin film (<5.9 × 10⁻²¹ cm³ cm cm^¬2 s^¬1 Pa^¬1) This same crosslinking treatment reduced the oxygen transmission rate (OTR), measured at 23 °C and 100% RH, from 0.61 to 0.09 (cm³ m⁻² day^¬1). Increasing the number of bilayers and heating time with thermal crosslinking also reduces the water vapor transmission rate (WVTR). These nanocoatings are a promising alternative to currently used barrier layers for flexible electronics and food packaging.

---

Introduction

The past two decades have shown layer-by-layer (LbL) assembly of polyelectrolyte multilayers, to be one of the most promising thin film technologies because of its simplicity, robustness, and versatility.^1–3 LbL assembly makes use of various interactions, such as electrostatics and hydrogen bonding between polymers or particles,^4,5 to build thin films with tunable properties. Superhydrophobicity,^6,7 flame resistance,^8,9 antimicrobial,^10,11 electrical conductivity,^12,13 selective patterning^14,15 and drug delivery^16–18 have all been achieved with LbL assembly of appropriate ingredients. Gas barrier layers are very important for a variety of packaging applications, from food to flexible electronics.^19–21 Several thin and flexible gas barriers have been developed for these applications, such as metalized films,²² SiO_x coatings,²³ polymer-inorganic oxide multilayers,²⁴ and chitosan-based films.²⁵ The lowest reported oxygen permeability was recently achieved with LbL assembly of polymer and clay.²⁶ Even without clay, highly interpenetrated PEI/PAA assemblies show an unprecedented oxygen barrier (P_O2 < 3.2 × 10⁻²¹ cm³ (STP) cm/(cm² s Pa)) with only 8 bilayers (16 total polymer layers) on PET.²⁷ After crosslinking with glutaraldehyde, film thickness decreased from 451 to 305 nm and the oxygen transmission rate decreased by almost an order of magnitude at 100% RH. This result showed the ability to impart better moisture resistance by chemically crosslinking these films.

Electrostatic assemblies, with ionic bonding and physical crosslinks, are prone to lose some of their integrity by absorbing water or changing charge density with pH,^28–30 which leads to diminished gas barrier behavior.³¹ Covalent crosslinking is an effective way to prevent this degradation of LbL thin films.^32,33 It is known that more densely-packed molecular organization often exhibits improved mechanical behavior,³⁴ chemical stability,³⁵ conductivity,³⁶ and reduced oxygen or ion permeability.^37,38 Polyelectrolyte mutilayers have already been successfully crosslinked with bifunctional aldehydes,^39,40 carbodiimides,^41–43 anhydrides,⁴⁴ UV irradiation^32,45,46 and/or heat.^35,40,47,48 It has been shown that heat-crosslinked poly(allylamine hydrochloride) (PAH)/poly(acrylic acid) (PAA) assemblies result in a passivating layer.⁴⁷ This same crosslinked system exhibits improved corrosion resistance,⁴⁹ modulus,⁴⁶ and ion transport selectivity.³⁷ As already mentioned, creating covalent bonds between amine and carboxylic acid groups also showed improved oxygen barriers in polyethylenimine (PEI) assembled with poly (acrylic acid) (PAA),²⁷ but there has never been a systematic study of the influence of crosslinking on gas permeability of these types of thin film assemblies.

In an effort to further improve the gas barrier of PEI/PAA assemblies and better understand the influence of crosslinking, glutaraldehyde (GA), 1-[3-(dimethylamino)propyl]-3-ethylcarbodiimide methiodide (EDC) and heat were examined with respect to concentration, temperature and/or time. The extent of thickness reduction can be tailored from 2 to 50% using different crosslinking methods and parameters. FTIR spectra suggest that both GA and EDC produce a higher extent of crosslinking with increasing concentration and time. Oxygen transmission rates at 0% and 100% RH show a strong connection between gas barrier and crosslinker concentration. 0.1 M GA and 0.01 M EDC exhibit the lowest OTR at 100% RH. The lowest oxygen permeability was achieved by an 8 BL film crosslinked with 0.1 M GA for 30 min. This PEI/PAA film has a 551 nm thickness and P_O2 < 5.9 × 10⁻²¹ cm³ cm cm^¬2 s^¬1 Pa^¬1, which is one order of magnitude lower than typical SiO_x nanocoatings.⁵⁰ Chemically-crosslinked films did little to reduce water vapor transmission rate, but a 50 BL film, thermally-crosslinked at 180 °C for 5 h, reduces WVTR by 46% relative to the uncoated PET substrate. By optimizing the crosslinking conditions, one can create the best OTR and WVTR all-polymer coating. These simple post-treatments of PEI/PAA multilayers could potentially be used to improve food and flexible electronics packaging, gas separation membranes, and self-healing coatings.

Experimental

Materials

Branched polyethylenimine (PEI) (Aldrich, St. Louis, MO) (M_W ∼ 25 000 g mol⁻¹) was dissolved into 18.2 MΩ deionized (DI) water to create a 0.1 wt% cationic solution. The pH of the PEI solution was reduced to 10 by adding 1.0 M hydrochloric acid (HCl). Poly(acrylic acid) (PAA) (Aldrich) (M_W ∼ 100 000 g mol⁻¹) was dissolved in DI water to create a 0.2 wt% anionic solution. The pH of the PAA solution was raised to 4 by adding 1.0 M sodium hydroxide (NaOH). Glutaraldyhyde (GA) (Aldrich) and 1-[3-(dimethylamino)propyl]-3-ethylcarbodiimide methiodide (EDC) (Aldrich) were used as crosslinking agents, in the form of 0.01, 0.05 and 0.10 M aqueous solutions.

Substrates

Single-side-polished (100) silicon wafers (University Wafer, South Boston, MA) were used as substrates for ellipsometry, FTIR and atomic force microscopy (AFM). Silicon wafers were piranha treated with 3 : 7 ratio of 30% hydrogen peroxide to 99% sulfuric acid and then stored in deionized water before being used. Caution: Piranha solution reacts violently with organic material and needs to be handled properly. Prior to use, the silicon wafers were rinsed with acetone and deionized water. Poly(ethylene terephthalate) (PET) film, with a thickness of 179 μm (trade name ST505, Dupont–Teijin), was purchased from Tekra (New Berlin, WI) and used as the substrate for barrier measurements. PET films were rinsed with deionized water and methanol and then corona-treated with a BD-20C Corona Treater (Electro-Technic Products Inc., Chicago, IL) before use. Corona treatment improves adhesion of the first polyelectrolyte layer to polymeric substrates by oxidizing the surface.⁵¹

Layer-by-layer deposition

Thin film deposition is shown schematically in Fig. 1. A given substrate was first dipped into a cationic PEI solution for 5 min, followed by 30 s of deionized water rinsing and blow drying with filtered-air. The substrate was then dipped into the anionic PAA solution for 5 min, with another rinsing and drying step. Each PEI/PAA deposition cycle creates one “bilayer (BL)”. Starting from the second deposition cycle, the remaining number of BLs were created using one minute dip times for both solutions. All deposition was carried out using a home-built robotic system.⁵² One BL of PEI/PAA is abbreviated as (PEI/PAA)₁. Films were crosslinked afterwards by dipping into a 0.01 M, 0.05 M or 0.1 M GA (or EDC) solution for 3, 30 or 300 min. Thermal crosslinking was performed by heating the film in an oven at 120, 150 or 180 °C for 1, 2, or 5 h. In order to examine the influence thickness growth following crosslinking, films were grown to 10 BL and then crosslinked with GA, EDC or heat. After crosslinking, film growth was continued to 14 BL. All other crosslinked films throughout this study are made using crosslinking as the final step.

Fig. 1 Schematic of the layer-by-layer deposition process. Steps 1–4 are repeated until the desired number of PEI/PAA bilayers are deposited.

Film characterization

Thickness of LbL films on silicon wafers was measured with an alpha-SE Ellipsometer (J. A. Woollam Co., Inc., Lincoln, NE). Films thicker than 2 μm were measured with a P-6 profilometer (KLA-Tencor, Milpitas, CA). Fourier transform infrared spectroscopy (FTIR) was performed in the dry state with an ALPHA-P10098-4 spectrometer (Bruker Optics Inc., Billerica, MA) in ATR mode. The IR spectra analyses were done by normalizing the intensity of the band at 1548 cm⁻¹ (–COO⁻). Surface structure of the coated silicon wafers were imaged with a Multimode Scanning Probe Microscope (AFM) (Veeco Digital Instruments, Santa Barbara, CA) in tapping mode. Oxygen transmission rate (OTR) tests were performed by MOCON (Minneapolis, MN) in accordance with ASTM D-3985, using an Oxtran 2/21 ML instrument at 23 °C and 0 or 100% RH. Water vapor transmission rate (WVTR) tests were performed by MOCON with ASTM F-1249, using a Permatran 3/33 G instrument (MOCON) at 23 °C and 100% RH.

Results and discussion

Influence of crosslinking on film thickness

Growth of PEI/PAA assemblies, from 3 to 18 BL, is shown in Fig. 2. This system exhibits exponential growth typical of weak polyelectrolytes, similar to the well-studied PAH/PAA system.^53–57 Exponential growth is attributed to an “in and out” diffusion mechanism that involves an endothermic polycation/polyanion complexation process.^41,58,59 Three different growth zones can be observed here: (1) island (2) exponential and (3) linear. In the first few bilayers of deposition, polymers only cover part of the substrate surface, which results in relatively uneven surface coverage (known as islands) and slow growth (1–4 BL). After several layers are deposited, the surface of the substrate becomes fully-covered, as islands coalesce with one another. In this stage, the “in and out” diffusion mechanism dominates, resulting in the fastest growth (4–10 BL). Eventually film growth slows and ultimately grows linearly due to the amount of charged groups for overcompensation reaching a saturation point (i.e., maximum value) beyond 10 BL.⁵³

Fig. 2 Ellipsometric thickness of PEI/PAA assemblies as a function of bilayers deposited

Crosslinking creates covalent linkages among polymer chains inside of an LbL film, which results in a “blocking layer” effect that decreases the extent of polymer interdiffusion and slows subsequent growth.^27,60 After crosslinking, a densified layer is formed at the surface that prevents polymer interdiffusion as LbL deposition proceeds. Table 1 summarizes thickness reduction in PEI/PAA assemblies, calculated with 14 BL films with and without crosslinking after 10 BL, using various concentrations and times. For GA, all crosslinked films have reduced thickness ranging from 15 to 36%, indicating that crosslinking did reduce film growth. Increasing crosslinker concentration also increases the extent of thickness reduction, especially at shorter crosslinking time. The trends become unclear at longer times due to the swelling effect that counteracts the influence of crosslinking. For EDC, the extent of thickness reduction is not as much as GA (∼0 to 18%), suggesting the blocking layer effect is weaker. Similar to GA, increasing the concentration of EDC results in more thickness reduction at shorter crosslinking time. The film crosslinked with 0.01 M EDC for 300 min is even thicker than its uncrosslinked counterpart. It appears that thickness reduction comes with effective creation of blocking layers and higher extent of crosslinking. From Table 1, higher concentration of GA and EDC both lead to higher extent of crosslinking, while GA stops the PEI/PAA interdiffusion more effectively.

Table 1 Thickness reduction of (PEI/PAA)₁₄ with GA or EDC cross-linking after 10 BL

	GA			EDC
a This negative value indicates film swelling.
Time (min)	0.01 M	0.05 M	0.1 M	0.01 M	0.05 M	0.1 M
3	15%	32%	36%	2%	4%	15%
30	34%	29%	35%	11%	7%	18%
300	31%	24%	31%	−1%^a	7%	∼0%

---

Thickness reductions of 14 BL films with thermal crosslinking at 10 BL, at various temperatures and times, are summarized in Table 2. Compared to GA and EDC, heating results in thinner films because more water molecules were driven out of the film. Increasing crosslinking temperature and time both lead to greater thickness reduction. At 180 °C, a 5 h crosslinked film is 50% thinner than an uncrosslinked film, which is a difference of more than 1 μm. Higher temperature and longer heating time contribute to a more effective blocking layer, which stops the interdiffusion. By applying different crosslinking methods to LbL films, one can easily manipulate the nanoscale film thickness.

Table 2 Thickness reduction of (PEI/PAA)₁₄ with and without thermal crosslinking after 10 BL

Time (h)	120 °C	150 °C	180 °C
1	25%	29%	40%
2	21%	36%	44%
5	36%	40%	50%

---

FTIR analysis of crosslinked films

Fig. 3 shows the chemical reaction that occurs with each type of crosslinking. Glutaraldehyde reacts with the primary amine groups of PEI, forming a “Schiff base” (Fig. 3(a)), while EDC and heat create amide bonds between PEI and PAA. EDC acts as an activator in the process, generating a urea derivative as a side product (Fig. 3(b)).⁴² Thermal crosslinking is a simple condensation reaction between PEI and PAA that generates water as a byproduct (Fig. 3(c)). Each of these crosslinking mechanisms were verified by FTIR.

Fig. 3 Chemistry for PEI/PAA assemblies exposed to GA (a), EDC (b) or heat (c).

Fig. 4 shows the ATR-FTIR spectra of 10 BL PEI/PAA films crosslinked with varying GA concentrations and times. Assemblies crosslinked with 0.01 M GA exhibit similar spectra to the control, without generating new peaks (or shifted peaks), regardless of time (Fig. 4(a)). Increasing the GA concentration to 0.05 M caused the peak at 1630 cm⁻¹ (–NH₂, stretching) to diminish (Fig. 4(b)). The C=N stretching band is located from 1645–1665 cm⁻¹,⁶¹ so it is assumed that generating the Schiff base influenced the appearance of the adjacent peak. It should be noted that the intensity for Schiff bases is relatively low, so the peak is not clear in these spectra. This “smoothing out” effect is much clearer in films crosslinked with 0.10 M GA, which indicates that higher concentration results in more crosslinking (Fig. 4(c)). These results confirm the proposed crosslinking mechanism (Fig. 3(a)) and suggest that the higher concentration solutions provide greater film crosslinking after a given time, which has also been observed by others.^62,63

Fig. 4 ATR-FTIR spectra of (PEI/PAA)₁₀ films crosslinked for varying times and with varying concentration of glutaraldehyde (a–c). These spectra are intentionally overlaid with arbitrary offset for clarity.

The ATR-FTIR spectra of EDC crosslinked films are shown in Fig. 5. All films show weaker intensity for the –COOH vibration band (located at 1710 cm⁻¹) because the carboxylic acid groups were activated by EDC to form amide bonds with PEI (Fig. 5(a)). The peak near 1630 cm⁻¹ represents the C=O stretching band for the amide bond (amide I). From 0.01 to 0.1 M, this peak becomes more intense with increasing crosslinking time, which can be explained by the generation of amide bonds. The –COOH vibration band gradually smooths out with increasing crosslinking time, from 30 min to 300 min, at all EDC concentrations. Additionally, the amide I peak intensity increased significantly, regardless of crosslinking time. A new peak generated at 1490 cm⁻¹ can be observed after 300 min of crosslinking, for all EDC concentrations, which represents the –O–CH₂ band from the ester intermediate produced by PAA and EDC (Fig. 3(b)).

Fig. 5 ATR-FTIR spectra of (PEI/PAA)₁₀ films crosslinked for varying times and with varying concentration of EDC (a–c). These spectra are intentionally overlaid with arbitrary offset for clarity.

Thermally crosslinking PEI/PAA assemblies requires a minimum temperature of 150 °C for 5 h. At 120 °C there are no spectral changes relative to the control, regardless of heating time (Fig. 6). Raising the temperature to 150 °C produces similar results at 1 and 2 h, but the amide I peak intensifies, and –COOH peak becomes weaker than the control, at 5 h (Fig. 6(b)). These spectral changes indicate that some crosslinking has occurred. At 180 °C, the intensity of the amide I peak becomes even more pronounced. All crosslinking times show intensified amide I peaks, while the intensity of peaks around 1543 cm⁻¹ (the asymmetric stretching band for –COO⁻) decreased. There is likely some overlap of this peak with the amide II band (40–60% N–H bending and 18–40% C–N stretching).⁶⁴ Among all the 180 °C films, a decreasing trend is revealed between the apparent crosslink density, as evidenced by the intensity of the amide I band (∼1630 cm⁻¹), and the heating time in these PEI/PAA thin films (Fig. 6(c)).

Fig. 6 FTIR spectra of (PEI/PAA)₁₀ films crosslinked for varying times and with varying temperatures (a–c). These spectra are intentionally overlaid with arbitrary offset for clarity

Topography of crosslinked films

AFM surface images of (PEI/PAA)₁₀ films, with and without 0.1 M GA, 0.1 M EDC or 150 °C exposure, are shown in Fig. 7. All films were crosslinked for 30 min for direct comparison. The GA-crosslinked film has a rougher surface than the control (Fig. 7(b) and (f)), while both EDC and thermally crosslinked films have smoother surfaces than the control (Fig. 7(c), (d), (g) and (h)). GA crosslinking creates covalent bonds between PEI layers by generating a Schiff base (Fig. 7(a)), which tightens the initial molecular arrangement. EDC and thermal crosslinking, on the other hand, create covalent linkages by connecting amine groups from PEI and carboxylic acid groups from PAA, which fill the uneven “free sites” and make the overall surface smoother. Root-mean-square (RMS) surface roughnesses were measured with 20 × 20 μm scans. The order of RMS roughness is: GA (118.1 nm) > control (90.5 nm) > thermal (58.1 nm) > EDC (55.4 nm). After crosslinking, GA increased roughness by 30%, while thermal and EDC reduced roughness by 35% and 38%, respectively. A similar smoothing effect by EDC, on PAH/PAA assemblies, was previously demonstrated by Caruso.⁴² Crosslinking provides a direct way to manipulate surface morphology of LbL films, which could be useful for surface patterning or tailoring hydrophilicity.

Fig. 7 AFM surface images of (PEI/PAA)₁₀ films prepared with varying crosslinking conditions: control (a)(e), 0.1 M GA (b)(f), 0.1 M EDC (c)(g), 150 °C (d)(h). (a)–(d) are height images and (e)–(h) are phase images. All crosslinking was performed for 30 min.

Oxygen and moisture barrier of crosslinked films

Oxygen transmission rates (OTR) of PEI/PAA assemblies on PET were measured at 23 °C and 0% or 100% RH, as shown in Fig. 8. (PEI/PAA)₈ films, crosslinked with varying concentrations (0.01, 0.05 and 0.10 M) of GA for 30 min, show a three orders of magnitude reduction in OTR relative to the uncoated substrate (OTR = 8.48 cm³ cm cm^¬2 s^¬1 Pa^¬1 for 179 μm thick PET film) under dry conditions (Fig. 8(a)). An 8 BL PEI/PAA thin film with no crosslinking, and crosslinked with 0.10 M GA, reaches the instrumental undetectable limit (OTR < 0.005 cm³ cm cm^¬2 s^¬1 Pa^¬1). It is interesting to see that intermediate GA concentrations resulted in increased OTR. Higher GA concentration correlates to a higher extent of crosslinking in these assemblies (Fig. 4), so it is assumed that the observed reduction in barrier came from reduced film thickness (OTR is a thickness dependent property). Table 3 summarizes oxygen barrier and film thickness data for these chemically-crosslinked assemblies. With 0.01 M GA, an 8 BL assembly's thickness is reduced by 30%, while it is only reduced by 14% with 0.10 M GA. At 100% RH, 0.01 M and 0.05 M GA both fail to reduce OTR relative to the uncrosslinked (PEI/PAA)₈ thin film, which correlates with the 0% RH results. The highest crosslink density, from 0.10 M GA, reduces the OTR from 0.61 to 0.09 cm³ m⁻² atm^¬1 day^¬1 (Fig. 8(b)), which is two orders of magnitude below uncoated PET. Although similar to GA, crosslinking with EDC results in some interesting differences in thin film OTR.

Fig. 8 Oxygen transmission rate of PEI/PAA assemblies on PET at 0% (a)(c) and 100% RH (b)(d). These films were crosslinked with GA (a)(b) or EDC (c)(d) for 30 min.

Table 3 Oxygen permeability of (PEI/PAA)₈ assemblies on PET film at 23 °C

Recipe	OTR (cm^3 m^−2 atm^¬1 day^¬1)		Film Thickness (nm)	Permeability (×10^−16 cm^3 cm cm^¬2 s^¬1 Pa^¬1)
a Film permeability was decoupled from the total permeability using a previously described method.^65 b The low end detection limit for an Ox Tran 2/21 L module is 0.005 cm^3 m^−2 atm^¬1 day^¬1.
	0% RH	100% RH		Film^a	Total
Bare PET	8.48	6.60	N/A	N/A	17.3
8 BL	<0.005^b	0.61	640	<0.000068^b	<0.0096
8 BL+0.01 M GA	0.23	0.76	445	0.0024	0.47
8 BL+0.05 M GA	0.11	0.78	501	0.0013	0.23
8 BL+0.10 M GA	<0.005	0.09	551	<0.000059	<0.0096
8 BL+0.01 M EDC	0.0062	0.21	483	0.000068	0.013
8 BL+0.05 M EDC	0.12	0.59	526	0.0015	0.25
8 BL+0.10 M EDC	0.0062	0.32	603	0.000085	0.012

---

At 0% RH, the OTR of (PEI/PAA)₈ increased slightly from undetectable to 0.0062 cm³ m⁻² atm^¬1 day^¬1 with 0.01 M EDC. OTR became much higher with 0.05 M (0.12 cm³ m⁻² atm^¬1 day^¬1), while 0.10 M EDC decreased the OTR back to 0.0062 cm³ m⁻² atm^¬1 day^¬1 (Fig. 8(c)). With a constant crosslinking time, the extent of reaction increased as the EDC concentration increased. It is clear from Table 3 that film thickness is less influenced by EDC and the resulting lower density films exhibit a poorer barrier than GA-crosslinked films. At 100% RH, the 0.01 M EDC crosslinked film shows the most OTR improvement compared to the two higher concentrations (Fig. 8(d)). It is likely that the optimum crosslinking concentration for EDC is <0.01 M. With greater EDC concentration, more crosslinking generates more urea derivatives that may plasticize the film and reduce the oxygen barrier. Permeability values in Table 3 were obtained by multiplying film thickness by OTR. GA crosslinking yields reduced permeability with increasing concentration, while the most dilute EDC crosslinking exhibits the lowest oxygen permeability. Crosslinking with 0.10 M GA also produces the best barrier at 100% RH. With an oxygen permeability of <5.9 × 10⁻²¹ cm³ cm cm^¬2 s^¬1 Pa^¬1 under dry conditions, this nanocoating is better than 100 nm SiO_x (P_O2 = 1.05 × 10⁻²⁰ cm³ cm cm^¬2 s^¬1 Pa^¬1)⁵⁰ and may be more economical for commercial-scale coating of packaging film.

Thermally-crosslinked (PEI/PAA)₈ films (at 120 or 150 °C for 5 h) exhibited higher dry OTR (0.26 and 0.36 cm³ m⁻² atm^¬1 day^¬1 for 120 and 150 °C, respectively) than an uncrosslinked film. Elevated OTR was also observed at 100% RH (0.90 and 1.22 cm³ m⁻² atm^¬1 day^¬1 for 120 and 150 °C, respectively) relative to an unheated film (0.61 cm³ m⁻² atm^¬1 day^¬1). Reduced thickness and a low level of crosslinking are the two factors causing higher OTR. Simple dehydration at 120 °C can account for reduced thickness (from 640 to 382 nm) without any crosslinking. Although heating at 150 °C for 5 h results in amide crosslinks (Fig. 9), the significant decrease in thickness (from 640 to 507 nm) appears to counteract the benefit of crosslinking.

Fig. 9 Water vapor transmission rate of PEI/PAA assemblies on PET.

Gas transport through materials can be attributed to two mechanisms: (1) Fickian flow by a diffusion-solubility model, which is mostly observed in homogeneous polymeric materials or (2) flow through defects, such as pinholes or micro-channels.⁶⁶ PEI/PAA assemblies are believed to have defect-free surfaces, as shown in Fig. 7. Oxygen transport through a PEI/PAA multilayer film is expected to be better described by the diffusion-solubility model based on this observation. This model describes molecular transport through a homogeneous material in several steps. Penetrants are first adsorbed onto the barrier surface, which then dissolve into the material. Gas molecules next diffuse through the thickness, moving out of one phase and re-dissolving into another phase to continue the process. In a single-phase material, the permeability of Fickian flow depends on both diffusivity and solubility. With two components and numerous acid/base interfaces, the PEI/PAA thin film assembly is believed to have a scrambled-egg structure,⁶⁷ similar to a highly interpenetrated network.⁶⁸ The chaotic aggregation among carboxylic acid groups and amine groups makes it difficult for gas molecules to dissolve, diffuse and re-dissolve. Interactions with PEI, PAA, and the PEI/PAA interfaces all contribute to slowing molecular motion through the film. Unlike traditional macroscopic polymer blends, LbL assembly results in “nano-blends” of the two ingredients, with many more interfaces that trap gas molecules and create a high oxygen barrier.

Another contributing factor to the high oxygen barrier in these films is reduced free volume.⁶⁹ The high glass transition temperature for this PEI/PAA assembly was previously used to show that PEI and PAA are more attracted to each other than they are to themselves.²⁷ This strong attraction reduces free volume that can be further reduced by crosslinking. Covalent crosslinking not only reduces the free volume of PEI/PAA, it also provides a better humid oxygen barrier by reducing swellability. In a high humidity environment, LbL films often swell and lose integrity,^70,71 which increases permeability to oxygen molecules passing through. Crosslinking the “free” functional groups (–NH₂ of PEI and –COOH of PAA) in the present assemblies decreases the number of hydrophilic groups within the film, thereby making it more hydrophobic. Crosslinking also reduces the ability of the film to swell, which was expected to improve the water barrier in these systems.

Water vapor transmission rates (WVTR) were tested on 8 BL films crosslinked with GA, EDC or heat at varying concentrations and temperatures. Unlike with oxygen, WVTR values show only 16–20% improvement (1.23–1.30 g m⁻² atm^¬1 day^¬1) relative to the uncoated substrate (1.53 g m⁻² atm^¬1 day^¬1). Heated assemblies were chosen for further WVTR testing because water is driven out during crosslinking. Fig. 9 shows WVTR as a function of the number of PEI/PAA bilayers deposited on PET, with and without exposure to 150 °C for 5 h. Water vapor transmission rate decreased with increasing bilayers. For 20 BL (3.3 μm) and 50 BL (7.3 μm) films, the WVTR decreased 15% and 9% compared to their uncrosslinked counterparts, respectively. A crosslinked 8 BL film is very similar to the 20 BL film without heat treatment, which demonstrates that thermal crosslinking creates a more moisture-resistant structure. The lowest WVTR came from the 50 BL heated film, which decreased the WVTR of PET from 1.53 to 0.82 g m⁻² atm^¬1 day^¬1 (∼46% reduction). In general, thickness and structure are both important for reducing WVTR. Although heating decreases the number of permeable pathways in the structure by creating amide bonds, the simultaneous reduction of thickness counteracts this to some extent. The electrostatic attraction among PEI/PAA charged groups can also be weakened by water, creating more diffusion pathways.

Conclusions

Super gas barrier thin films, composed of PEI and PAA, were successfully deposited via LbL assembly and crosslinked using glutaraldyhyde,1-[3-(dimethylamino)propyl]-3-ethylcarbodiimide methiodide and/or heating with varying concentration, temperature and time. Growth of crosslinked films were inhibited by the blocking layer effect, which results in reduced film thickness with the extent of reduction ranging from 2 to 50% for 14 BL films. Covalent bond formation, due to the various crosslinking mechanisms, was confirmed by FTIR. All IR spectra indicate higher crosslinking density with longer crosslinking time. AFM images highlight the ability to tailor the surface roughness of these PEI/PAA films with crosslinking method. GA makes the surface rougher, while EDC and thermal crosslinking both make the film smoother.

Oxygen transmission rates of chemically-crosslinked PEI/PAA assemblies are improved at 100% RH relative to uncrosslinked films. Maximum barrier improvements occur at different GA and EDC concentrations, suggesting that crosslinker chemistry influences the barrier. A 551 nm thick, 8 BL PEI/PAA assembly exhibits an oxygen permeability (<5.9 × 10⁻²¹ cm³ cm cm^¬2 s^¬1 Pa^¬1) that rivals SiO_x. Thermally-crosslinked films showed some improvement in moisture barriers, but this required a high number of layers. These films are inherently hydrophilic and more work needs to be done with hydrophobic crosslinkers and/or the addition of impermeable particles to make significant reductions in WVTR. This is the first systematic study of crosslinking and its influence on the barrier properties of LbL assemblies. The improved oxygen barrier under high humidity, and modest improvement in the water vapor barrier, is promising for food packaging, selective gas membranes and protection of flexible electronics.

Acknowledgements

The authors would like to thank the Texas Engineering Experiment Station (TEES), Baker Hughes and Kuraray America, Inc. for financial support of this work. We also thank Dr Nicole Zacharia for helpful discussions.

References

1. G. Decher and J. B. Schlenoff, Multilayer thin films sequential assembly of nanocomposite materials, Wiley-VCH, Weinheim, 2012.
2. P. Bertrand, A. Jonas, A. Laschewsky and R. Legras, Macromol. Rapid Commun., 2000, 21, 319–348.
3. P. T. Hammond, Adv. Mater., 2004, 16, 1271–1293.
4. M. Schonhoff, Curr. Opin. Colloid Interface Sci., 2003, 8, 86–95.
5. W. B. Stockton and M. F. Rubner, Macromolecules, 1997, 30, 2717–2725.
6. L. Zhai, F. C. Cebeci, R. E. Cohen and M. F. Rubner, Nano Lett., 2004, 4, 1349–1353.
7. L. Zhang and J. Q. Sun, Macromolecules, 2010, 43, 2413–2420.
8. Y. C. Li, S. Mannen, A. B. Morgan, S. C. Chang, Y. H. Yang, B. Condon and J. C. Grunlan, Adv. Mater., 2011, 23, 3926–3931.
9. A. Laachachi, V. Ball, K. Apaydin, V. Toniazzo and D. Ruch, Langmuir, 2011, 27, 13879–13887.
10. P. Podsiadlo, S. Paternel, J. M. Rouillard, Z. F. Zhang, J. Lee, J. W. Lee, L. Gulari and N. A. Kotov, Langmuir, 2005, 21, 11915–11921.
11. O. Etienne, C. Gasnier, C. Taddei, J. C. Voegel, D. Aunis, P. Schaaf, M. H. Metz-Boutigue, A. L. Bolcato-Bellemin and C. Egles, Biomaterials, 2005, 26, 6704–6712.
12. B. S. Shim, J. A. Zhu, E. Jan, K. Critchley and N. A. Kotov, ACS Nano, 2010, 4, 3725–3734.
13. Y. T. Park, A. Y. Ham, Y. H. Yang and J. C. Grunlan, RSC Adv., 2011, 1, 662–671.
14. J. Park, L. D. Fouche and P. T. Hammond, Adv. Mater., 2005, 17, 2575–2579.
15. C. Y. Cho, L. Valverde, G. A. Ozin and N. S. Zacharia, Langmuir, 2010, 26, 13637–13643.
16. A. J. Chung and M. F. Rubner, Langmuir, 2002, 18, 1176–1183.
17. X. Yu and M. V. Pishko, Biomacromolecules, 2011, 12, 3205–3212.
18. P. C. DeMuth, X. F. Su, R. E. Samuel, P. T. Hammond and D. J. Irvine, Adv. Mater., 2010, 22, 4851–4856.
19. C. A. Phillips, Int. J. Food Sci. Technol., 1996, 31, 463–479.
20. S. K. Bhattacharya and R. R. Tummala, Microelectron. J., 2001, 32, 11–19.
21. M. Yan, T. W. Kim, A. G. Erlat, M. Pellow, D. F. Foust, H. Liu, M. Schaepkens, C. M. Heller, P. A. McConnelee, T. P. Feist and A. R. Duggal, Proc. IEEE, 2005, 93, 1468–1477.
22. E. H. H. Jamieson and A. H. Windle, J. Mater. Sci., 1983, 18, 64–80.
23. R. J. Spontak, A. G. Erlat, R. P. Clarke, T. C. Robinson, P. D. Haaland, Y. Tropsha, N. G. Harvey and E. A. Vogler, J. Phys. Chem. B, 1999, 103, 6047–6055.
24. J. D. Affinito, M. E. Gross, C. A. Coronado, G. L. Graff, E. N. Greenwell and P. M. Martin, Thin Solid Films, 1996, 290, 63–67.
25. B. L. Butler, P. J. Vergano, R. F. Testin, J. M. Bunn and J. L. Wiles, J. Food Sci., 1996, 61, 953–956.
26. M. A. Priolo, D. Gamboa, K. M. Holder and J. C. Grunlan, Nano Lett., 2010, 10, 4970–4974.
27. Y. H. Yang, M. Haile, Y. T. Park, F. A. Malek and J. C. Grunlan, Macromolecules, 2011, 44, 1450–1459.
28. J. J. Harris and M. L. Bruening, Langmuir, 2000, 16, 2006–2013.
29. M. Losche, J. Schmitt, G. Decher, W. G. Bouwman and K. Kjaer, Macromolecules, 1998, 31, 8893–8906.
30. J. B. Schlenoff, H. Ly and M. Li, J. Am. Chem. Soc., 1998, 120, 7626–7634.
31. M. A. Priolo, D. Gamboa and J. C. Grunlan, ACS Appl. Mater. Interfaces, 2010, 2, 312–320.
32. J. Y. Chen, L. Huang, L. M. Ying, G. B. Luo, X. S. Zhao and W. X. Cao, Langmuir, 1999, 15, 7208–7212.
33. J. H. Dai, A. W. Jensen, D. K. Mohanty, J. Erndt and M. L. Bruening, Langmuir, 2001, 17, 931–937.
34. S. Moya, L. Dahne, A. Voigt, S. Leporatti, E. Donath and H. Mohwald, Colloids Surf., A, 2001, 183, 27–40.
35. S. Y. Yang and M. F. Rubner, J. Am. Chem. Soc., 2002, 124, 2100–2101.
36. J. H. Chen, M. Asano, T. Yamaki and M. Yoshida, J. Appl. Polym. Sci., 2006, 100, 4565–4574.
37. A. M. Balachandra, J. H. Dai and M. L. Bruening, Macromolecules, 2002, 35, 3171–3178.
38. B. C. Ku, D. Froio, D. Steeves, D. W. Kim, H. Ahn, J. A. Ratto, A. Blumstein, J. Kumar and L. A. Samuelson, Journal of Macromolecular Science-Pure and Applied Chemistry, 2004, A41, 1401–1410.
39. S. D. Tuong, H. Lee and H. Kim, Macromol. Res., 2008, 16, 373–378.
40. W. J. Tong, C. Y. Gao and H. Mohwald, Chem. Mater., 2005, 17, 4610–4616.
41. L. Richert, F. Boulmedais, P. Lavalle, J. Mutterer, E. Ferreux, G. Decher, P. Schaaf, J. C. Voegel and C. Picart, Biomacromolecules, 2004, 5, 284–294.
42. P. Schuetz and F. Caruso, Adv. Funct. Mater., 2003, 13, 929–937.
43. T. Mauser, C. Dejugnat and G. B. Sukhorukov, Macromol. Rapid Commun., 2004, 25, 1781–1785.
44. R. E. Ducker, M. T. Montague and G. J. Leggett, Biointerphases, 2008, 3, 59–65.
45. M. K. Park, S. X. Deng and R. C. Advincula, J. Am. Chem. Soc., 2004, 126, 13723–13731.
46. A. M. Lehaf, M. D. Moussallem and J. B. Schlenoff, Langmuir, 2011, 27, 4756–4763.
47. J. J. Harris, P. M. DeRose and M. L. Bruening, J. Am. Chem. Soc., 1999, 121, 1978–1979.
48. B. S. Kim, S. W. Park and P. T. Hammond, ACS Nano, 2008, 2, 386–392.
49. J. H. Dai, D. M. Sullivan and M. L. Bruening, Ind. Eng. Chem. Res., 2000, 39, 3528–3535.
50. N. Inagaki, S. Tasaka and H. Hiramatsu, J. Appl. Polym. Sci., 1999, 71, 2091–2100.
51. E. M. Liston, L. Martinu and M. R. Wertheimer, J. Adhes. Sci. Technol., 1993, 7, 1091–1127.
52. D. Gamboa, M. A. Priolo, A. Ham and J. C. Grunlan, Rev. Sci. Instrum., 2010, 81, 036103.
53. P. Bieker and M. Schonhoff, Macromolecules, 2010, 43, 5052–5059.
54. B. Sun, C. M. Jewell, N. J. Fredin and D. M. Lynn, Langmuir, 2007, 23, 8452–8459.
55. P. Podsiadlo, M. Michel, J. Lee, E. Verploegen, N. W. S. Kam, V. Ball, J. Lee, Y. Qi, A. J. Hart, P. T. Hammond and N. A. Kotov, Nano Lett., 2008, 8, 1762–1770.
56. L. Y. Shen, J. H. Fu, K. Fu, C. Picart and J. A. Ji, Langmuir, 2010, 26, 16634–16637.
57. J. H. Fu, J. Ji, L. Y. Shen, A. Kuller, A. Rosenhahn, J. C. Shen and M. Grunze, Langmuir, 2009, 25, 672–675.
58. N. Laugel, C. Betscha, M. Winterhalter, J. C. Voegel, P. Schaaf and V. Ball, J. Phys. Chem. B, 2006, 110, 19443–19449.
59. C. Picart, P. Lavalle, P. Hubert, F. J. G. Cuisinier, G. Decher, P. Schaaf and J. C. Voegel, Langmuir, 2001, 17, 7414–7424.
60. N. S. Zacharia, D. M. DeLongchamp, M. Modestino and P. T. Hammond, Macromolecules, 2007, 40, 1598–1603.
61. L. J. Bellamy, The infra-red spectra of complex molecules. Volume 1, Chapman and Hall, London; New York, 1975.
62. M. Morcellet, M. Ghoul and M. Bacquet, Water Res., 2003, 37, 729–734.
63. C. Y. Gao, W. J. Tong and H. Mohwald, Macromolecules, 2006, 39, 335–340.
64. S. Farris, J. H. Song and Q. R. Huang, J. Agric. Food Chem., 2010, 58, 998–1003.
65. A. P. Roberts, B. M. Henry, A. P. Sutton, C. R. M. Grovenor, G. A. D. Briggs, T. Miyamoto, A. Kano, Y. Tsukahara and M. Yanaka, J. Membr. Sci., 2002, 208, 75–88.
66. H. Chatham, Surf. Coat. Technol., 1996, 78, 1–9.
67. B. Philipp, H. Dautzenberg, K. J. Linow, J. Kotz and W. Dawydoff, Prog. Polym. Sci., 1989, 14, 91–172.
68. T. Aoki, M. Kawashima, H. Katono, K. Sanui, N. Ogata, T. Okano and Y. Sakurai, Macromolecules, 1994, 27, 947–952.
69. I. Pinnau and L. G. Toy, J. Membr. Sci., 1996, 116, 199–209.
70. A. J. Nolte, N. D. Treat, R. E. Cohen and M. F. Rubner, Macromolecules, 2008, 41, 5793–5798.
71. J. E. Wong, F. Rehfeldt, P. Hanni, M. Tanaka and R. V. Klitzing, Macromolecules, 2004, 37, 7285–7289.

---