Flexible photovoltaic (PV) technologies have gained considerable attention for new energy-generating applications, such as buildingintegrated PVs [1–6], vehicle-integrated PVs [7–9], and product-integrated PVs [2,10,11]. Flexible thin-film solar cells [12] are lightweight, allowing high specific power (kW/kg) and mechanical robustness, which are of particular interest, and a lower production cost by employing large-scale production, enabling high throughput (e.g., roll-to-roll manufacturing). Furthermore, they can be integrated with elements of various shapes and sizes (such as fabrics, tiles, tents, and sails) [2,10].

For practical applications, it is crucial for a flexible PV module to achieve long-term reliability compared with a glass-based rigid module. Thus, encapsulation barrier technology is important for achieving a long device lifetime. The design objectives of the barrier depend on the required water vapor transmission rate (WVTR) and the device durability. For flexible solar cells in actual industrial products, the desired WVTR should be within 10⁻⁵ g/m²/day [13,14]. Recently, halide perovskite solar cells fabricated with thin-film encapsulation moisture barriers have been reported to successfully overcome instability in humid environment [15–18].

Recently, multilayer (dyad) barriers of organic−inorganic materials have been acknowledged as extremely effective in preventing moisture and oxygen penetration [19,20]. Their effectiveness can be attributed to the barrier property of the inorganic layer being coupled with the properties of the organic layer to planarize the surface topography, decouple defects/pinholes, and prevent cracks in the incorporated inorganic layer [21]. Additionally, an extended path through the inorganic and organic layers in the dyad is known to limit diffusion and greatly increase the lag times of gases and water vapor [22].

Atomic layer deposition (ALD)-grown inorganic layers have emerged as promising encapsulation solutions for organic light-emitting diode display [23–26] and flexible solar technologies [12,27–31]. Among these, Al₂O₃ is one of the most studied candidates because of its low WVTR. On the other hand, polymeric parylene is an ultimate conformal coating solution for protection of devices, components, and surfaces in electronics, instrumentation, aerospace, medical and engineering industries [32–36].

In this study, we prepared barrier layers of a single Al₂O₃ film and dyad organic-inorganic barriers coupled with a parylene C polymer and characterized them in terms of WVTR, surface roughness, and optical transmission. The 2-dyad barrier layer on polyethylene naphthalate (PEN) was applied to encapsulate two GaAs solar cells die-bonded onto a circuited printed-circuit-board (PCB) platform. For the flexible PV module, we conducted a damp/heat (DH) test to investigate the effectiveness of the moisture barrier by measuring the current−voltage (I−V) curves.

2.1. Film growth

A commercial 6-inch ALD chamber (iSAC Co. Ltd., iOV m100) was used to deposit the inorganic films. Figure 1 illustrates the schematic design of the moisture barrier used in this study. The Al₂O₃ films were grown at 100 °C on different substrates, Si (100) wafer (10 × 10 mm²) and soda-lime glass (SLG) (10 × 10 mm²), for monitoring and characterization. The Si and SLG substrates were ultrasonically cleaned with acetone, isopropanol, and deionized water for 30 min and then dried with N₂ gas. Trimethylamine [N(CH₃)3, TMA] (iChems Co.) was used a precursor at the bubbler temperature of 50 °C for obtaining an appropriate vapor pressure for the ALD process. Ar (99.99 %) was used as the carrier gas with a fixed flow rate of 50 standard cubic centimeters per minute, and the working pressure was maintained at 300 mTorr. The ozone reactant was generated by an ozone generator using O₂ gas (99.99 %). Each deposition cycle consisted of the following sequence: TMA pulse (1.5 s) – Ar purge (3 s) – O₃ (3 s) pulse – Ar purge (5 s). The growth rate of Al₂O₃ at the reactor temperature of 100 °C was 0.091 nm/cycle, and the film thickness was controlled by adjusting the cycles of deposition. For the substrate (100 mm × 100 mm) of the moisture barrier, we utilized three commercial transparent polymeric substrates: PEN of 125 μm thickness, polyethylene terephthalate (PET) of 100 μm thickness, and ethylene tetrafluoroethylene (ETFE) of 75 μm thickness. Polymer substrates were utilized as asreceived layers without cleaning to deposit the dyads.

Figure 1. Design of moisture barriers: (a) Al₂O₃(70 nm)/polymer substrate, (b) 1-dyad of Al₂O₃(70 nm)/parylene (700 nm), and (c) 2-dyad of 2-[Al₂O₃(70 nm)/parylene (700 nm)].

Parylene C was chosen as the organic layer to relax the stress of the inorganic Al₂O₃ layers to fabricate the dyad barrier layers, as shown in Figs. 1(b) and 1(c). The parylene C film is an excellent candidate for the dyad layers because of its characteristics, such as hydrophobicity (WVTR = 0.08 g/m²/day) for prohibiting moisture ingression in the film and low Young’s modulus (4 GPa) for the layer flexibility of an encapsulated device module [37]. It also prevents water from condensing on the inorganic Al₂O₃ film, causing its corrosion [38]. The parylene layer was deposited by chemical vapor deposition using a specialized instrument (NRPC-500, NURITECH Co.). The process consists of three stages: sublimation of the solid dimer (di-para-xylene) into vapor, dissociation of the dimers into monomers (para-xylene), and condensation of the monomers to yield a polymeric poly(para-xylene) film. The deposition started at the chamber pressure below 0.01 Torr and continued until its completion at the furnace temperature of 670 °C and a vapor temperature of 125 °C. Parylene thin films can be deposited regardless of the polymer substrate used (e.g., PEN, ETFE, or PET).

2.2. Dyad moisture barrier and encapsulation

Figures 1(b) and 1(c) illustrate the design of the dyad barrier layer used in this study. All samples were fabricated with a parylene layer deposited first on the substrate, and then dyad structures of Al₂O₃−parylene were prepared. Several dyads were formed with a single layer of Al₂O₃ (70 nm) coupled with parylene-C layers of three thicknesses: 500, 700, and 1,000 nm. Water permeability was investigated by measuring the WVTR for each barrier layer.

Flexible PV modules were fabricated using dyad barrier layers. Figure 2 shows the schematic of the encapsulation of the PV module. Table I summarizes the water permeabilities of all materials used for encapsulation. Two flexible GaAs III-V solar cells (5 mm × 5 mm) were placed on a patterned PCB made of a sheet of Cu (25 μm)-onpolyimide (50 μm)-Cu (25 μm). Then, ethylene-vinyl acetate (EVA) was placed on the PCB whose edges were surrounded by an edge sealant (butyl rubber), and a sheet of barrier layer of 2-dyad was laid out on top of the entire module. Finally, layer lamination was accomplished by heating at a temperature of 136 °C for a period of 710 s.

Figure 2. Structure of flexible device encapsulation.

Table I. WVTR of moisture barrier materials used for PV module..

	Flexible PCB* (Cu / PI / Cu)	EVA [39]	PET [39]	PEN [40]	Edge sealant (butyl rubber)	Barrier (2-dyad)
WVTR (g/m2/day)	7.61 × 10−4	6.7–17.1	5.8–23.0	0.30	2.5–3.5 × 10−1	2.01 × 10−4

*tri-layer sheet: Cu (25 μm) / Polyimide (50 μm) / Cu (25 μm)..

2.3. Characterization

Single layers of Al₂O₃, parylene, and organic-inorganic dyads were characterized using various instruments. The film thickness and refractive index were determined using an ellipsometer (Film Sense, Model FS-1, version 1.69) with excellent thickness precision of < 0.001 nm. Measurements at an angle of incidence of 65° and the default refractive index were deduced at a single wavelength (λ) of 633 nm. The surface morphologies of the films were analyzed by field-emission scanning electron microscopy (JEOL, JSM-7610F) up to an accelerating voltage of 15 keV at room temperature with a secondary electron detector. The topography of the Al₂O₃ film’s surface of dyads was determined using atomic force microscopy (AFM) (XE-100, Park Systems) under ambient conditions in non-contact mode. The optical transmittance of the moisture barrier layer was measured using a ultraviolet–visible (UV-2600, Shimadzu) in the wavelength range of 300–1,400 nm. The WVTR of the manufactured barrier layer was measured using Technolox’s Deltaperm equipment at the condition of 40 °C, 90 % relative humidity (RH), and the cross-sectional shape of the dyad structure is confirmed with a focused ion beam scanning electron microscopy (FIB-SEM) (Helios 600i, FEI Co.). A solar module of 100 cm² applied with a moisture barrier layer with confirmed moisture barrier ability was tested for damp heat (85 °C/85 % RH) for 500 h using a solar simulator (940463A-1000, Newport Co.) and a temperature and humidity tester (C340, WEISSTECHNIK).

3.1. WVTR measurement

We investigated water permeability or WVTR as a function of film thickness to investigate the barrier properties of the Al₂O₃ layers on the PEN substrate. As the film thickness increased, the WVTR monotonically reduced and appeared to saturate for t ≥ 70 nm, as shown in Fig. 3(a). The barrier of a single Al₂O₃ film with t = 70 nm on PEN yielded a WVTR of 9.64 × 10⁻² g/m²/day. For comparison, Al₂O₃ (70 nm) layers were also deposited on ETFE and PET polymer substrates, and their WVTRs were obtained 1.79 and 2.06 × 10⁻² g/m²/day, respectively. The processing time for the 100 nm thick layer was 4.5 h which was 38 % longer than that for the 70 nm sample, whereas the performance improved by only 23 %. Thus, the 70 nm thick Al₂O₃ layer was optimized as an organic/inorganic dyad moisture barrier.

Figure 3. (a) WVTR on PEN, PET, and ETFE vs. thickness of Al₂O₃, (b) WVTR of 1-, 2-, and 3-dyads of various thicknesses of parylene, (c) FIB-SEM image of fabricated 2-dyad of Al₂O₃ (~70 nm) / parylene (~700 nm), and (d) a graph of WVTR vs. time of moisture exposure for the Al₂O₃ (70 nm)/PEN (125 μm). The mark is at 2.01 × 10⁻⁴ g/m²/day.

We investigated dyad samples in which the water permeabilities varied with a combination of organic and inorganic materials, by varying the polymer thickness for the 70 nm thick Al₂O₃ layer. In addition to the bare PEN substrate and Al₂O₃ (70 nm)/PEN, we prepared six dyads with parylene layers of different thicknesses; 1-dyad (500, 700, and 1,000 nm), 2-dyad (500 and 700 nm), and 3-dyad (700 nm). Figure 3(b) displays the WVTR results for all the barrier samples. As expected, the dyad exhibited significantly improved moisture protection and lower water permeability than the single Al₂O₃ (70 nm)/PEN. As shown in Fig. 3(d), the smallest WVTR = 1.99 × 10⁻⁴ g/m²/day was obtained from the 3-dyad barrier (700 nm polymer), which is nearly identical to that of the 2-dyad 2.01 × 10⁻⁴ g/m²/day. Figure 3(c) displays a FIB-SEM image of 2-dyad sample. As designed, the ALDgrown Al₂O₃ layer on PEN (125 μm) has a thickness of 72 nm for the first and second dyads. Although designed to be 700 nm, the parylene thicknesses were 810 and 765 nm in the first and second dyads, respectively, which were within the tolerance of deposition of 15 % of the equipment. To fabricate a flexible solar module, a 2-dyad barrier was used for encapsulation because of its effectiveness in prohibiting moisture ingression, along with the process complexity and functionality.

3.2. Surface topography of barrier

As stated above, an Al₂O₃ layer and a parylene layer were alternately laminated to prevent defects such as pinholes and cracks that may occur in the inorganic Al₂O₃ layer from continuously growing and forming a moisture permeation path [12,41–43]. As shown in Fig. 3(c), the 2-dyad structure confirmed that the individual parylene and Al₂O₃ layers formed a uniform layer without breaking the boundaries between them. However, as the physical vapor-deposited layers replicate the surface topology of the substrate, the high points in the underlying substrate cannot be smoothed using ALD-grown Al₂O₃ thin films. Surface features also introduce mechanical damage during the coating process, leading to defects in the deposited inorganic layers. Thus, it is important to place an organic polymer layer on the substrate (e.g., PEN or ETFE) to provide a smooth surface for the deposition of successive inorganic layers, increasing the probability that the oxide layer can effectively inhibit moisture or oxygen penetration. The required initial polymer layer thickness varies with the substrate depending on the surface roughness.

The surface topography of the bare substrates, Al₂O₃ film of the single barrier, and 1-dyad barrier layers as fabricated were obtained on the PEN substrate, as shown in Figs. 4(a), 4(c), and 4(e), and on the ETFE substrate, as in Figs. 4(b), 4(d), and 4(f), which were measured using AFM. Figure 4 shows the AFM images of the specimen scanned in an area of 1 μm × 1 μm. Figures 4(a) and 4(b) show the surface morphologies of bare PEN and ETFE. The PEN surface mostly appeared flat with micro-undulations of even height, resulting in a small root-mean-square (RMS) surface roughness of 1.45 nm. In comparison, the ETFE showed a wavy surface of large bumps and dips that caused a large RMS roughness of 4.95 nm. The surface features of the samples with both substrates showed little improvement in the surface topology when the surface was covered with a layer of inorganic Al₂O₃ films deposited using ALD, as shown in Figures 4(c) and 4(d). RMS surface roughness was 1.46 and 4.62 nm for PEN and ETFE, respectively. For the 1-dyad samples in which the initial parylene layer was deposited, the surface topology displayed a large granular surface shape for both substrates. The surface of 1-dyad on PEN showed a striking difference with a large increase in the RMS surface roughness to 5.43 nm in which the micro-undulation of the surface disappeared. By contrast, the sample on ETFE became smooth on the surface without significantly changing the RMS surface roughness from 4.49 nm.

Figure 4. AFM measured surface topology of barrier on polymer substrates: on PEN (a) bare surface, (c) Al₂O₃ single layer, and (e) 1-dyad of parylene (700 nm); on ETFE (b) bare surface, (d) Al₂O₃ single layer, and (f) 1-dyad of parylene (700 nm).

3.3. Optical transmittance of barrier

When encapsulating a flexible solar cell, the optical transmittance of the n-dyad barrier is important for minimizing the loss of lightconversion efficiency to the possible extent in the module. The optical transmittances of the barrier layers on PEN (125 μm) were investigated, and the results are shown in Fig. 5. For PV module encapsulation, we used flexible GaAs solar cells (Fig. 2), which mainly absorb visible light in the wavelength range of 380–780 nm [44]. The average visible transmittance (AVT) within this range was compared among the barrier structures: 85.7, 88.4, 88.9, and 86.2 % for the bare PEN substrate, Al₂O₃/PEN, 1-dyad, and 2-dyad, respectively.

Figure 5. Optical transmission spectra of barriers; bare PEN (black), Al₂O₃ on PEN (red), 1-dyad (green), and 2-dyad (blue).

The incident light passed through a dyad structure of both organic (parylene) and inorganic (Al₂O₃) layers and a thick PEN substrate, as shown in Fig. 3(c). The structure is the same as that of an antireflective coating [45,46]. Thus, it can reduce the reflectance of the incident light while increasing the light transmittance. Therefore, when the Al₂O₃ layer was deposited with a thickness of approximately 70 nm, the reflectance decreased and the average light transmittance increased by 2.7 %. When 70 nm thick parylene was deposited to form a 1-dyad, a wiggle in the transmittance occurred due to light interference owing to the characteristics of the 1-dyad structure, and the AVT slightly increased by 0.6 %. Likewise, the 2-dyad sample showed light transmission with increased interference, owing to a thicker layer with two successive 1-dyads. The reduced AVT of 2.5 % was caused by light absorption owing to the thicker total layer in the 2-dyad structure.

3.4. DH test

According to the encapsulation design, as shown in Fig. 2, we fabricated a flexible PV module, as explained in Section 2.2, and investigated the effectiveness of the 2-dyad moisture barrier using the DH test. The PV module of 100 cm² was placed inside the temperature and humidity tester and was tested following the standard procedure of IEC61646 in the environmental condition of 85 °C temperature and 85 % RH. Figure 6 shows photographs of the PV module and the measured I-V curves before and after the DH test. Table II lists the device characteristics of the PV modules.

Figure 6. Photos of PV module (a) before and (b) after DH test and (c) corresponding I-V curves of PV module; before test (black) and after DH 500 h (red). (d) Equivalent circuit of PV module of two solar cells connected in series.

Table II. Change of device parameters of PV module in the DH test..

Parameter	DH (0 h)	DH (500 h)	Change w.r.t. DH (0 h)
Voc (V)	4.43	4.36	−0.16 %
Isc (mA)	3.26	3.32	0.18 %
Fill factor (%)	78.70	76.30	−3.00 %
Efficiency (%)	18.77	18.25	−2.80 %

Total cell area: 0.605 cm²..

After DH testing for 500 h, the PV module exhibited a light-conversion efficiency of η = 18.25 %, which is a reduction of 2.8 %. Additionally, the other parameters showed slight changes, within 3.0 %, compared with those of the initial module. The fill factor appears to be most influenced by the change in efficiency, which is believed to be induced by the resistive damage in the microstrip line circuit of the PCB platform. Figure 6(b) shows that the solar cells were not affected by moisture during the test. However, the PCB platform appeared to be affected by the edges toward the center area where the solar cells were located. This trend could be attributed to the lower WVTR of the edge sealant (butyl rubber), such that moisture passes through the sides over a long period, whereas the top (2-dyad moisture barrier) and bottom layers of the module were very successful in preventing moisture ingression. The DH test concluded that the 2-dyad structure of Al₂O₃ (70 nm)−parylene (700 nm) layers is effective in defending against moisture penetration and will be a good encapsulating layer for fabricating a flexible solar module.

We investigated various moisture barrier layers of Al₂O₃ films and six dyad barriers coupled with parylene C with different thicknesses: 1-dyad (500, 700, and 1,000 nm), 2-dyad (500 and 700 nm), and 3-dyad (700 nm) on PEN and ETFE substrates. A single Al₂O₃ film on PEN showed an optimal thickness of 70 nm as a moisture barrier with WVTR = 9.64 × 10⁻² g/m²/day, and the capability of prohibiting water penetration in the single Al₂O₃ (70 nm) on PEN was superior to the film on the ETFE substrate. For dyad barriers composed of Al₂O₃−parylene pairs, the smallest WVTR of 1.99 × 10⁻⁴ g/m²/day was obtained from the 3-dyad barrier (700 nm polymer), which is nearly identical to that of 2-dyad, WVTR of 2.01 × 10⁻⁴ g/m²/day. Using the 2-dyad barrier, the flexible PV module was encapsulated and its effectiveness was evaluated in DH (85 °C/85 % RH) for 500 h. After the DH test for 500 h, the PV module revealed a light conversion efficiency of η = 18.25 %, which is a reduction of only 2.8 % compared with the initial module. In conclusion, the 2-dyad structure of the Al₂O₃ (70 nm)−parylene (700 nm) layers is effective in defending against moisture penetration and will be a good encapsulating layer for fabricating a flexible solar module.

The authors would like to express their gratitude to Dr. H.G. Kang at the Korea Advanced Nano Fab Center (KANC) for providing us with flexible GaAs solar cells and Prof. J.Y. Yang at Kunsan National University for providing us with the edge sealant material for encapsulation.