Abstract
A transparent barrier structure consisting of silicon oxide 
stacks was deposited on a polycarbonate substrate at 80°C by plasma-enhanced chemical vapor deposition. Details of radio-frequency (rf) power effects on the 
and 
film properties in terms of etching rate, refractive index, internal stress, and water vapor transmission rate (WVTR) were investigated. It was found that the impermeability, flexibility, and optical property of the 
barrier films can be tailored by varying the rf power. A gradual decrease in the compressive internal stress of each stack film was designed to prevent the stress-induced cracks during the multilayer deposition process. The WVTR value of the optimum barrier structure (
+ 6 pairs of 
) can reduce to 
under a calcium test (100 days at 
, 40% relative humidity). After 5000 cyclic bending tests in a compressive mode, the WVTR value can keep below 
. The performance of the 
barrier stacks presented has high potential for future flexible electronics applications.
Export citation and abstract BibTeX RIS
Recently, transparent barrier coatings on polymer substrates have received much attention for liquid crystal displays, organic light-emitting-diode (OLED) displays, solar modules, and food packing applications.1, 2 An overview of the barrier requirements for different applications is shown in Fig. 1.3 Among them, the criteria of water vapor transmission rate (WVTR) and oxygen transmission rate (OTR) are most stringent for OLEDs applications. Currently, the OLEDs are fabricated onto glass substrates and then encapsulated with metal or glass lids. An OLED display on a plastic substrate is thinner and lighter compared with that on a glass substrate. By utilizing the substrate flexibility, displays of various forms can be fabricated including the so-called flexible display.4, 5 However, until now, the best performance of the single-layer barrier structure still shows in orders of magnitude short of the requirements for OLEDs. To improve the barrier performance, a variety of multilayered structures has been proposed to meet these requirements.6, 7 Besides impermeability, an ideal barrier film for flexible electronic applications should possess low deposition temperature, optimum film stress, good mechanical behavior, low surface roughness, high optical transparency, etc.
Figure 1. Barrier requirements for different applications.
In this study, we deposited the silicon oxide 
/silicon nitride 
stacks as the barrier structure instead of the organic/inorganic multilayered structure on the polycarbonate (PC) substrate to prevent adhesion problems which may occur between the organic and inorganic interfaces.8, 9 The 
and 
films were deposited by plasma-enhanced chemical vapor deposition (PECVD). PECVD is one of the techniques that allow industrial-scale deposition of high-quality barrier coatings with good uniformity and good adherence to the substrate. Furthermore, it is well known that one of the main advantages of PECVD is that applying ion bombardment to the surface is equivalent to increasing the surface temperature.10 This contributes to the densification of the film and improved adhesion, allowing the possibility of depositing high-quality coatings at low temperature. By varying the radio-frequency (rf) power level during the PECVD process, we can adjust the film density, refractive index, and stress value of the 
and 
films. Details of the 
and 
film properties in terms of the etching rate, refractive index, and internal stress were examined as a function of the rf power. The WVTR performance of various 
barrier structures is also discussed.
Experimental
To enhance the adhesion between 
and PC substrate, an inductively coupled plasma (ICP) system was employed for surface pretreatment. To generate the high-density plasma, a 13.56 MHz rf power was applied to the ICP coil, and another 13.56 MHz rf power was applied to the substrate to generate a self-bias voltage on the substrate. Typical pretreatment conditions were flow rate of 
for Ar, source power of 
, bias power of 
, and pressure of 
. After the Ar plasma pretreatment, the 
and 
films were deposited by PECVD in a parallel plate, capacitive-coupled-plasma 
reactor with an electrode spacing of 
. A gas showerhead (6 in. diam) served as the powered electrode and was isolated electrically from the reactor by a ceramic spacer. On the grounded electrode, a ceramic heater was kept at 
to heat the substrate to 
. The substrate temperature was measured using a thermocouple bead, electrically shielded from the plasma that contacted the back side of the substrate holder. 5% 
in Ar with 
or 
were used as source gases. During the growing process, the deposition parameters for the 
films were 
flow rate 
, rf power 
, pressure 
, and temperature 
; the deposition parameters of 
films were 
flow rate 
, rf power 
, pressure 
, and temperature 
to prevent any deformation of the PC substrates. The molded PC substrates were 
in thickness with a glass-transition temperature of 
and a surface roughness of 
.
The film thickness, deposition rate, and etching rate were measured directly on the coated PC, or on small pieces of Si wafer as reference substrates using a Tencor-KLA (P-10) profilometer. The refractive index and transmittance data were measured by a thin-film measurement system (model: 1280, N&K Technol. Inc.). The film stress in the 
and 
coatings on 
Si wafers was determined using a Tencor FLX-2320 laser profilometer by applying the Stoney formula.11 It is said that the same tendency of stress variation can be observed in the films on PC substrates and Si wafers.12 A Wyko NT-1100 (Veeco) optical profiler was used to obtain a large fields of view over the 
and 
films. Measurements of WVTR and OTR permeation were carried out on a 10 
active sample area, at 
and standard pressure, using MOCON "Permeation W3/61" and "Ox-tran 2/61" instruments, respectively. Zero percent relative humidity (RH) was used for the OTR measurements. Frequent calibrations were performed with a standard PET film sample supplied by MOCON. Because the low permeation measurement through a polymer substrate is very critical, there is still no definite commercial method for measuring a WVTR below 
. Therefore, a calcium test was used to assess the ultra high barrier properties of polymer substrates. Figure 2 shows a typical experimental setup of the calcium test used in this work. The barrier-coated PC substrate was cut into 
, and a 50 nm thick calcium layer and a 100 nm thick aluminum protective layer were then deposited on the coated polymer substrate. Finally, the epoxy was used to seal the aluminum layer with the glass substrate. The calcium test samples were observed using optical microscopy at 
, 40% RH. The bending test was carried out with a homemade machine. Samples were bent to a cylinder of radius 
under compressive stress with the speed of 100 times per minute.
Figure 2. Experimental setup of calcium test used in this work.
Results and Discussion
The WVTR and OTR values of 
and 
with different film thicknesses on PC substrates are presented in Fig. 3a and 3b. The 
and 
films provide excellent barrier performance even when they are extremely thin. The 
and 
films show great oxygen impermeability while the 
films present higher moisture resistance than the 
films. Note that the diffusion mechanism for water vapor is more complex than that of oxygen.13 Water molecules interact with 
bonds and cause stress instability of the 
films.14 However, the 
films show better optical transparency than 
films. A combination of 
and 
films may act as an excellent barrier structure on polymer substrates. To meet the requirements of impermeability and optical transparency, film thickness of 70 and 
was chosen for 
and 
films, respectively.15 The 
and 
films are expected to provide a defect-decoupling effect because of the difference in chemical surface.16, 17 The calcium test results (0 and 
) of 
, 
, and 
on PC are shown in Fig. 4. By observing the percentage of color changing area, one could calculate WVTR values effectively.18, 19 The results confirm that water vapor is easier to transmit through the 
films than the 
films. In addition, the result of 
on PC shows less reactive area than the single inorganic layer, which may be due to the defect-decoupling phenomena between 
and 
film. In this work, we deposited 
instead of 
stack on the PC substrate to prevent the undesirable oxygen atoms reacting with the underlying PC substrate during the deposition process.20
Figure 3. WVTR and OTR values of PECVD (a) 
and (b) 
film on PC substrates as functions of film thickness.
Figure 4. The calcium test results of 
, 
, and 
on PC substrate.
Figure 5 shows the influence of the RF power on the etching rate and refractive index of 
and 
films. The etching rate of the 
film initially decreases from 9 to 
when the rf power increases from 30 to 
. Further increasing the rf power (i.e., above 
) makes the 
etch rate increases. The etching rate of the 
film decrease from 6.9 to 
when the rf power increases from 30 to 
. For the refractive index results, both 
and 
films become near stoichiometry with increasing the rf deposition power. For 
films, the refractive index behavior could be due to that the plasma energy at the low rf power level is too low to generate enough 
.21 To gain an insight into the etch rate results, the relationship between the internal stress of the films and the rf deposition during the PECVD process was investigated. The film stress is critical to the mechanical properties and flexibility of the barrier films and can be calculated by measuring the difference in the substrate curvatures before and after deposition. As shown in Fig. 6, the internal stress of 
films becomes more compressive with increasing the rf power from 30 to 
. Further increasing the rf power appears to lower the compressive stress of the 
film. The phenomenon indicates stress relaxation processes occur with increasing rf power after 
. These relaxation processes could result from the formation of nanoscopic cracks in the barrier films.22, 23 The internal stress of 
films becomes more compressive from 
to 
with increasing the rf power from 30 to 
. It is well known that higher compressive stress in the coatings is beneficial to the film density, barrier performance and adhesion.6 Therefore, the film density will increase with increasing the compressive stress level and the results of the etch rate and film stress show great dependence. However, higher internal stress tends to cause various forms of failures. Based on these results, the film density, refractive index and stress value of the 
and 
films can be optimized by varying the rf power density.
Figure 5. Etch rate and refractive index of PECVD (a) 
and (b) 
film on PC substrates as functions of rf power.
Figure 6. Internal stress of 
and 
film deposited on Si wafer as a function of rf power.
Figure 7 shows the WVTR values as a function of the number of 
stacks. One stack of 
deposited at rf power 
shows better moisture resistance than the stacks deposited at rf power 
or 
. Further increasing the stack number of 
deposited at rf power 
results in a slightly degraded WVTR value. On the other hand, increasing the stack number of 
deposited at rf power 
results in a slightly lower WVTR value. These observations show that the WVTR cannot decrease to a minimum value with simply repeating the optimum single stack. The increase of the stack layers would cause the decreasing of radius curvature which induces the strain to grow. Once the strain is equal to or greater than the critical strain, the multilayer coating will fail. Therefore, the 
stacks deposited at higher rf power show less moisture resistance when the stack number increases. On the contrary, the WVTR value of the four 
stacks deposited at rf power 
on a PC substrate can reduce to 
. Figures 8a and 8b show the curvature of single and four 
stacks deposited at rf power 
. The coatings eventually failed with decreasing radius curvature. However, the decreasing rate from the WVTR value of 
stacks deposited at rf power 
is too slow to meet the OLED requirement. According to these results, the rf power used in each stack was tuned to achieve low compressive stress and low permeability at the same time.
Figure 7. Calcium test results of 
stacks deposited at different rf power as a function of the number of stacks.
Figure 8. Curvature of (a) single and (b) four 
stacks deposited at rf power of 80/180 W.
There are internal stress and external stress in a thin film. Internal stress can be introduced in a thin film due to the temperature gradient and growth disorder. Once the process temperature and the substrate have been selected, it is not easy to control the thermal stress. Therefore, we made the compressive internal stress of the films decrease stack by stack by controlling the rf power. As shown in Fig. 9, the internal stress 
of the 
and 
films in the barrier structure reduced gradually. For 
films, the rf power conditions used for deposition were 
, 
, and 
, respectively. For 
films, the rf power conditions used for deposition were 
, 
, and 
, respectively. Hence, the release rate of strain energy which strongly depends on the internal stress of the film would not exceed the fracture energy of the film. Under this situation, we can prevent stress-induced cracks of the barrier structure during the film deposition process. Besides the internal stress, external stress also cause strain in the barrier films. When a composite film is bent by an external force, the strain ε is given by24
where 
is the distance from the neutral axis and 
is the radius curvature of the neutral axis. The strain ε will increase when the 
value increases. Hence, the outer stacks need to withstand the higher strain. By decreasing the rf power applied, we can avoid cracks generation from the outer barrier layers. At last, we deposited the 
film on the top of the barrier layer in case of reaction between moisture and 
film. Calcium test results of the optimum barrier structure (
pairs) without bending and after bending in a compressive mode for 5000 times are shown in Figs. 10a and 10b. The WVTR value of the optimum barrier structure can reduce to 
and can keep below 
after the cyclic bending. Moreover, the average optical transmittance of the barrier structure was greater than 87% in the visible light region as shown in Fig. 11. It is believed that, the impermeability, flexibility, and optical transparency of the multilayer barrier structure have high potential for future flexible display applications.
Figure 9. An optimum design of transparent barrier films of 
stacks on flexible polymer substrates.
Figure 10. (Color online) Calcium test results of the optimum barrier structure (a) without and (b) after bending in a compressive mode for 5000 times.
Figure 11. Transmittance of PC, 
on PC and 
pairs on PC as a function of wavelength.
Conclusions
In this study, the 
stacks were deposited on flexible PC substrates as gas barrier coatings instead of the organic/inorganic multilayer to prevent the adhesion problems which frequently encounter between organic and inorganic layers. Details of the 
and 
film properties in terms of the etching rate, refractive index, and internal stress were examined as a function of the rf power. Although the 
and 
films seem to become denser and near stoichiometry with increasing the PECVD deposition power, the higher internal stress may cause the coatings to fail. This phenomenon becomes more serious with increasing stack number and a compromise should be made between the impermeability and the mechanical properties. By adjusting the rf power level, we made the compressive internal stress of the films decrease stack by stack to prevent stress-induced cracks and achieve low permeability at the same time. In addition to the internal stress, we also discussed the influence of the external stress on the barrier films. By decreasing the rf deposition power used for each stack, the cracks generation from the outer barrier layers can be avoided when the external stress is applied. The WVTR value of the optimum barrier structure (
pairs) can reduce to 
measured by a calcium test (100 days at 
, 40% RH). After the cyclic bending test (5000 times) in a compressive mode, the WVTR value can still keep below 
. The high performance of the presented 
barrier stacks shows the suitability for future flexible electronics applications.
Acknowledgment
This work has been financially supported by the National Science Council of the Republic of China under contract no. NSC 93-2616-E-005-008.
National Chung Hsing University assisted in meeting the publication costs of this article.