Fabrication of graphene oxide/montmorillonite nanocomposite flexible thin films with improved gas-barrier properties

Nanocomposites are potential substitutes for inorganic materials in fabricating flexible gas-barrier thin films. In this study, two nanocomposites are used to form a flexible gas-barrier film that shows improved flexibility and a decreased water vapor transmission rate (WVTR), thereby extending the diffusion path length for gas molecules. The nanoclay materials used for the flexible gas-barrier thin film are Na+-montmorillonite (MMT) and graphene oxide (GO). A flexible gas-barrier thin film was fabricated using a layer-by-layer (LBL) deposition method, exploiting electronic bonding under non-vacuum conditions. The WVTR of the film, in which each layer was laminated by LBL assembly, was analyzed by Ca-test and the oxygen transmission rate (OTR) was analyzed by MOCON. When GO and MMT are used together, they fill each other's vacancies and form a gas-barrier film with high optical transmittance and the improved WVTR of 3.1 × 10−3 g per m2 per day without a large increase in thickness compared to barrier films produced with GO or MMT alone. Thus, this film has potential applicability as a barrier film in flexible electronic devices.


Introduction
When exibility is required in new electronic devices, exible gas-barrier lms that prevent water vapor transmission within such devices are of signicant research interest. 1 Previous developed technologies have used both organic and inorganic lms with SiO x or Al x O y multilayered structures. 2 The inorganic layers were typically fabricated by vacuum processes, such as chemical vapor deposition, atomic layer deposition, and sputtering. [3][4][5] However, vacuum processing has the disadvantages of low production efficiency and high production costs; inorganic layers also frequently crack under bending stresses, which permits H 2 O and O 2 to ow through exible electronic devices. H 2 O and O 2 react electrochemically with the cathode as active metals, forming additional H 2 gas inside the device. This H 2 gas forms bubbles at the cathode, which destroys the device. 6 For this reason, a mixture of GO and other materials such as polymer or nano clay which can be laminated by a non-vacuum process has seen increased interest as a good alternative to standard vacuum-processed inorganic layers. [7][8][9] In this study, a exible gas-barrier thin lm is fabricated using the layer-bylayer (LBL) deposition method, based on graphene oxide, nano clay materials of Na + -montmorillonite (MMT) and the two polymers of polydiallydimethylammonium chloride (PDDA) and poly(vinyl alcohol) (PVA). [10][11][12] Nanoclay layers can be fabricated through LBL assembly processing.
LBL refers to the general process of laminating using hydrogen bonding, covalent bonding, or electrostatic attraction under non-vacuum conditions. Therefore, gas-barrier lms fabricated through LBL processes possess steady structures formed in a cost-effective and short tact time. [13][14][15] We fabricated a gas-barrier thin lm with a decreased WVTR and OTR compared to a lm having the same number of layers by combining the nanoclays with the polymers to form adhesion layers. GO has a large aspect ratio and MMT is a plate-shaped material and both materials have good dispersibility in water, so they are suitable materials for lengthening the moisture permeation pathway in a gas-barrier thin lm. [16][17][18][19] The reaction between MMT and PVA has a negative charge, 20 while that of GO and PDDA has a positive charge. The fabricated exible gas-barrier lms showed good transparency and improved WVTR characteristics because the two materials were alternately laminated using electrostatic attraction in the LBL process. The water vapor transmission properties of the exible gas-barrier thin lms were analyzed by Ca-test. A bending test conrmed that this exible gas-barrier thin lm could be applied to exible devices.

Experimental
Gas-barrier lm fabrication GO (500 mg L À1 ) was purchased from the Graphene Supermarket. A mixture of 0.01 wt% GO in 200 mL of deionized (DI) water was magnetically stirred (450-550 rpm) for 24 h to disperse the GO uniformly. A solution of 0.02 wt% PDDA (M w ¼ 200 000-350 000, 20 wt% H 2 O) in DI water was prepared by magnetic stirring for 24 h. A mixture of the GO and PDDA solutions was magnetically stirred for 24 h to combine by the electrostatic attractions between the functional groups of the GO surface and the PDDA. In solution, yielding sheets of GO with positively charged surfaces. 21,22 The PDDA(GO) solution was adjusted to the pH of 10 using 1 M NaOH. 23 MMT was dispersed as a 0.05 wt% suspension in DI water by magnetic stirring for 24 h and then centrifuged at 4500 rpm for 1 h. Aer the rst centrifugation, large MMT particles were dispersed at the bottom of the solution. These large submerged MMT particles were extracted and centrifuged at 1700 rpm for 15 min to yield MMT particles of uniform size between 2 and 4 mm. The resulting solution was magnetically stirred for 24 h and mixed in a 3 : 1 volumetric ratio with 0.5 wt% PVA (M w ¼ 30 000-70 000, 87-90% hydrolyzed, purchased from Sigma-Aldrich). The mixture was magnetically stirred for 24 h at 85 C and stirred for another 24 h at room temperature in order to absorb the PVA between the MMT layers in solution. The PVA(MMT) solution was adjusted to a pH of 3.5 using 1 M HCl and comprised negatively charged complexes.
Scheme 1(a) shows the gas-barrier thin lms fabricated by the LBL process. A cleaned polyethylene naphthalate (PEN) lm with a thickness of 125 mm (purchased from DuPont Teijin) was treated by ultraviolet (UV) ozone for 20 min in order to form OH-radicals on its surface. The PEN substrate was dipped in the PDDA(GO) solution for 10 min at room temperature by an automated dipping system. Next, the substrate was rinsed with DI water for 3 min and dried with ltered air. Aerward, the substrate was dipped in the PVA(MMT) solution for 10 min. Aer the nal rinsing and drying, a single PDDA(GO)/PVA(MMT) bilayer was formed on the PEN substrate. For the second bilayer, the process was repeated with the dipping times in the PDDA(GO) and PVA(MMT) solutions reduced to 1 min. The lm thickness, light transmittance, and WVTR of the gas-barrier thin lms could be varied by changing the number of bilayers, as dened by the number of times the LBL dipping process was repeated.

WVTR and OTR
The performances of the gas-barrier lms fabricated by LBL processing on a PEN substrate were evaluated by measuring the WVTR using the Ca-test and the OTR using the MOCON OX-TRAN 2/21 MH (as specied in ASTM D-3985) at 23 C and 50% RH. The Ca-test is a conventional method for measuring WVTR, with a minimum rate of 10 À6 g per m 2 per day. Ca is highly sensitive to water and water vapor; when it reacts with water vapor passing through a gas-barrier lm, calcium oxide is formed. This oxidizes the Ca; therefore, an insulating lm experiences an increase in resistance value under an applied constant voltage, which can be observed by monitoring the decrease in current through the lm.
Analysis using zeta-potential, TEM, SEM, XRD, AFM and UVvis spectrophotometry To characterize the electrostatic attraction between the PDDA(GO) and PVA(MMT) solutions, a zeta-potential analyzer (ELSZ-1000, Otsuka Electronics) was used. The focused ionbeam (FIB) technique was used to prepare cross-sections of the gas-barrier lms for analysis by transmission electron

Results and discussion
The PEN substrate, surface-treated by UV ozone, is alternately dipped into the PDDA(GO) and PVA(MMT) solutions. The surface charge of each solution was analyzed by the zetapotential analyzer. We adjusted the pH of each solution and compared the zeta-potential values to obtain the appropriate adhesion. The zeta-potential value of the PDDA(GO) solution increases as the pH is decreased. At low pH, the carboxyl groups on the surface of GO are protonated and increase the number of COO À groups; thus, the repulsive forces between the GO sheets are reduced. 24 The experiment was performed under acidic conditions at a pH of 3.5, appropriate for the fabrication of gasbarrier lms, because very low pH values induce insufficient charge densities and consequently disturb the assembly of each lm layer. 24 For the PVA(MMT) solution, the zeta-potential decreases as the pH increases. Thus, the pH of the PVA(MMT) solution is adjusted to 10. 25 The PDDA(GO) solution has a positive surface charge of 39.73 mV, while PVA(MMT) has the negative surface charge of À10.98 mV.
Through XRD analysis conrmed that the incorporation between the GO and MMT. Fig. 1 shows the XRD patterns of GO, MMT, and GO-MMT. For GO, an obvious diffraction peak appears at 2q ¼ 11.56 , which corresponds to the d-spacing of 0.76 nm. MMT yields two intense diffraction peaks at 2q ¼ 7.

View Article Online
Scheme 2 shows that GO mixed PDDA nanocomposites increased the diffusion path compared to untreated nanocomposites. It also conrmed that the diffusion path is extends without the increase of thickness when GO was mixed with the PDDA through the SEM image comparing the thickness of the gas-barrier lms. Furthermore, when GO and MMT are alternately laminated, they materials are more stably bonded by hydrogen-bonding or crosslinking effects between GO and MMT, thus lling the vacancies formed when the layers are laminated individually and blocking pathways for H 2 O permeation. [28][29][30][31][32]    decreasing current under a constant applied voltage. The WVTR can be expressed as: where d is the Ca density, M is the molar mass of the indicated reagent, R is the resistance of the Ca connected to Ag electrodes, and h denotes the Ca height. R i and h i are the initial values of R and h, respectively. WVTR is proportional to the conductance, as indicated by a decrease in the Ca height Dh versus the elapsed time Dt. 35,36 Fig. 6(a) shows the WVTR of the uncoated PEN lm of 2.2 Â 10 À1 g per m 2 per day, that of the 10-layer PDDA(GO)-coated PEN lm of 3.6 Â 10 À2 g per m 2 per day, that of the 10-layer PDDA/PVA(MMT)-coated PEN lm of 4.8 Â 10 À2 g per m 2 per day, and that of the 10-layer PDDA(GO)/PVA(MMT)-coated PEN lm being 3.1 Â 10 À3 g per m 2 per day. When GO and MMT are alternately laminated, compared to when each is used alone, as shown in Scheme 2, extending the diffusion path length for gas molecules by inserting GO into the PDDA. Hence, WVTR value of the PDDA(GO)/PVA(MMT) lms was greatly reduced. The OTR values of lms were analyzed by MOCON. As shown in Fig. 6(b), the lowest OTR is measured for the lm containing GO and MMT as alternating layers. The OTR of the PEN substrate is measured as 39.31 cc per m 2 per day, which decreases to 6.91 cc per m 2 per day, and 3.69 cc per m 2 per day for the 10-layer PDDA/PVA(MMT)-, and 10-layer PDDA(GO)/PVA(MMT)-coated PEN substrates, respectively. This result shows that the oxygen permeability is decreased for the same reason causing the decreased WVTR, which is the lengthening of gas-molecule diffusion pathways.
Since transmittance is an important factor in gas-barrier lm, the light-transmittance properties of the gas-barrier lms are measured using a UV-vis spectrophotometer. Fig. 7(a)     This journal is © The Royal Society of Chemistry 2018 the lm comprising stacked GO and MMT shows the highest optical transmittance compared to the others. 34,37 As shown in Fig. 7(b), the PDDA(GO)/PVA(MMT) exible gas-barrier lm is sufficiently transparent. This result shows that the characteristics of WVTR are further improved without signicant deterioration in light transmittance by the alternating lamination of GO and MMT in lms of the same number of layers. Fig. 8 graphs a comparison of the WVTR values measured before and aer the bending testing of the PDDA(GO)/ PVA(MMT) gas-barrier lm. The bending test was repeated 10 000 times to the minimum radius of 3 cm. The WVTR value is changed from 5.0 Â 10 À3 g per m 2 per day before testing to 3.1 Â 10 À3 g per m 2 per day aerward. This result shows that no signicant change occurs in the WVTR value, even aer repeated bending; therefore, this gas-barrier lm is applicable to exible electronic devices. 38

Conclusions
In this study, we used GO and MMT as alternative inorganic materials for a exible gas-barrier thin lm. The mixtures of PDDA(GO) and PVA(MMT) formed cationic and anionic complexes. Therefore, PDDA(GO) and PVA(MMT) were successfully deposited using the LBL method as a non-vacuum process to form a gas-barrier thin lm with good interlayer bonding through electrostatic attraction. This method is costeffective, short in fabrication time, and applicable to largescale production. Our experimental results conrmed that a lm comprising PDDA(GO)/PVA(MMT) multilayers showed decreased WVTR, better exibility, and higher optical transmittance compared to single-material gas-barrier thin lms without a large increase in thickness. The PDDA(GO)/ PVA(MMT) multilayer-coated PEN gas-barrier thin lm shows great potential for use in exible applications.

Conflicts of interest
There are no conicts to declare.