Coating weight reduction by a carrier layer for tensioned-web slot coating.

INTRODUCTION

Tensioned-web slot coating (TWSC) was developed as an efficient method for the production of magnetic media (1-3). This technique can be considered as an extension of conventional slot die coating, so it has the advantage of a pre-metered coating method, which implies the coating thickness is predetermined before the coating operation.

The difference between the standard slot die coating and TWSC is that the slot die is positioned against a rigid backing roller for the former, whereas the slot die is positioned against a freely moving substrate for the latter. Recently, it was found that owing to the special feature of TWSC, this method can deliver very thin coating. The minimum wet coating thickness is usually above 10 [micro]m for conventional slot die coating, but TWSC may provide a wet thickness of less than 1 [micro]m for dilute coating solutions (4). The requirement of very thin coating is necessary for advanced products such as key components of liquid crystal display panels. For example, antireflection, antiglare, or hard coat films are made through precision coating technologies and the required coated layers are very thin, the wet thickness may vary between 1 and 10 [micro]m, and some may have multilayer structures (5), (6).

Contrary to the fast development of TWSC in coating industry, the fundamental research works published in open literature are relatively rare. Feng (7) first examined the flow field in the coating bead region of TWSC theoretically; both equation of motion for liquid and membrane theory for the moving web were included in his analysis.

Lee and Scriven (8) attempted to determine the stable operating region of TWSC experimentally. Park et al. (9) developed a crude macroscopic model for TWSC. Both Fuchigami (10) and Carvalho (11) reported that the minimum wet thickness, which is defined as the lowest stable coating thickness of a coating solution at specific operating conditions, is proportional to the tension number defined as the ratio of fluid viscous force to web tension.

Gutoff and Cohen (12) pointed out several possible approaches to improve the performance of TWSC. More recently, Lin et al. (4) carried out an experimental study to examine almost all possible parameters on the minimum wet thickness of TWSC. They performed a dimensional analysis to divide all the parameters into three-dimensionless groups and obtained a universal correlation through a least square procedure. Lin et al. (13) developed a macroscopic model which can predict the minimum wet thickness with reasonable accuracy.

Previous researchers found that as the viscosity of a coating solution increases, the minimum wet thickness also goes up for both slot die coating (14) and TWSC (4). For many industrial applications, it is desirable to carry out a thin layer coating of a viscous solution, not only to meet product specifications but also to reduce the energy required for drying. There are several patents that reported the possibility of simultaneous multilayer coating for TWSC (15-17), but these patents were concerned with specific structures of products. Yu et al. (18) reported that the thickness of a viscous coating layer can be significantly reduced by using a thin and dilute carrier layer for conventional slot die coating. Lu et al. (19) examined two-layer slot die coating of poly(vinyl) alcohol (PVA) solutions. In our study, we shall examine the idea if it is possible to reduce the wet thickness of a viscous layer by introducing a thin and less viscous carrier layer for TWSC.

EXPERIMENTAL

Dilute PVA solutions from Chang-Chun Chemical (code BP2420) were used as test fluids; the average molecular weight of PVA is around 120,000. The viscosities of PVA solutions were measured by a Brookfield viscometer (LVDV-III+CP) and surface tension was evaluated by a surface tensiometer with a Wilhelmy slide balance (CBVP-A3; Kyowa Interface Science Co.). A surfactant FSO (Du Pont) may add to adjust the surface tension of the coating solutions.

Physical properties of test solutions are listed in Table 1. A previous study (20) found that the PVA solutions used in the present work may exhibit slight shear-thinning behavior if the concentration exceeds 1% and shear rate higher than 800 [s.sup.-1]. Since the concentrations of PVA are equal or less than 7% in the present work, we assumed the test solutions were Newtonian. Polyethylene-terephthalate (PET) films made by Nan-Ya Co. (code BH21) were purchased as substrates. The film surfaces were treated by corona discharge to have a better wetting ability; the film thickness was 75 [micro]m in the experiment.

TABLE 1. Physical properties of the PVA test solutions at room
temperature 25[degrees]C.

Concentration of    Viscosity   Surface tension [sigma]     Density p
PVA (%)           [mu] (mPa s)           (mN/m)          (kg/[m.sup.3])

2                       7.4               39.2                1003
3                      18.7               38.4                1003
5                      74.4               57.2                1011
6                     210                 35.9                1015
7                     493                 34.3                1017

The pilot coater for the flow experiment is the same as the one reported previously (4), (13). The schematic of the coating and drying loop is displayed in Fig. 1; a roll of PET film that serves as a substrate is mounted on an unwinder D. Once the flow experiment starts, the PET substrate will pass a dancer roller E and then moves through two tension control rollers F and F'; the slot die C is set beneath the moving substrate between F and F'.

[FIGURE 1 OMITTED]

A coating solution in tank A is delivered by a pump B and then fills up the slot die C. The coating solution emanating from the slot die exit will coat on the moving PET substrate. The moving substrate with coated layers will enter an oven G and then is collected by a rewinder H after drying. The coating width is 0.06 m and the coating speed varies between 0.02 and 0.17 m/s. A CCD camera I. as shown in Fig. 1, was positioned above the moving web between the two rollers F and F', since both the coating solutions and PET film are transparent; the camera can catch the coating defects that appear on the film surface. The images of coated layers can be magnified and displayed in a monitor and then recorded for analysis. This flow visualization device was used in a previous study (4).

A two-layer slot die, as shown in Fig. 2a, was made for flow experiment. The geometry of the slot die lip is shown in Fig. 2b. The die geometry is symmetric to the center line: the slot gap is fixed to be 150 [micro]m for both layers. The receiving and departing angles are 10[degrees] for all die lip elements. The flow geometry of the two-layer coating is depicted in Fig. 2c; the position of the upstream meniscus is critical for stable coating.

[FIGURE 2 OMITTED]

The wrapping angle [theta] is a free parameter for adjusting the coating quality and thickness in real production situation. However, this angle was set before the flow experiment in the present study so that its effect can be systematically quantified. To determine the wrapping angle, the distance between the center of the tension control roller F and the central line of the slot die X as shown in Fig. 3a was measured first. Before the coating experiment, the two tension rollers F and F' were positioned such that the substrate just touched the slot die exit, and the substrate was kept in a horizontal position. Once the flow experiment started, the tension control rollers were lowered to a predetermined position as shown in Fig. 3b and the vertical distance between the initial and final position was measured as Y, and then the wrapping angle [theta] was computed as

[FIGURE 3 OMITTED]

[theta] = arc tan(Y/X). (1)

The parameter Y could be adjusted so that [theta] is equal to any specified wrapping angle in the flow experiment.

Since the thickness of the coated film on the substrate is thin and it is difficult to take online measurement, the average film thickness H of the coating layer was estimated as

H = [Q/VB], (2)

here Q is the total volumetric flow rate coated on the substrate. V is the coating speed, and B is the width of coating. Note that by taking the average coated film thickness, minor variations owing to edge effects that can not be detected by the CCD camera were neglected.

Values of the minimum wet thickness of the top or the bottom carrier layer had to be determined separately before the two-layer flow experiment to serve as references for comparison. The procedure for the determination of the minimum wet thickness of the viscous top or the dilute bottom carrier layer was the same as reported by Lin et al. (4) for single layer flow experiment. For a selected coating speed, the flow rate of the carrier layer was adjusted until a uniform and stable coated layer was observed, and then the flow rate was reduced gradually until coating defects appear. The corresponding coating thickness of this critical flow rate is defined as the minimum wet thickness at the specific coating speed. This procedure could be repeated until values of the minimum wet thickness over the operable range of coating speed were completely determined.

The two-layer flow experiment was carried out by establishing a stable and uniform bottom carrier layer slightly above its minimum wet thickness first at a low coating speed, and then the viscous top layer was delivered. Usually a high How rate of the top layer was required to establish stable coating, and then the flow rate of the top layer was reduced to a critical point at which a stable and uniform coating was no longer possible; this critical point determines the minimum wet thickness of the top layer with respect to the specific bottom carrier layer. The two-layer flow experiment started at a low coating speed and then the coating speed was increased gradually until the speed limit of the pilot coater was reached. Some trial-and-error steps were necessary to find a suitable flow rate for the top viscous layer. The quality of the two-layer coating could be observed and recorded on the coated film through the CCD camera.

RESULTS AND DISCUSSION

The operating window, or the stable and uniform coating region, for TWSC of single layer was investigated by Lin et al. (4): they found that different types of coating defects might appear outside the operating window. Leaking implies that too much coating solution is accumulated behind the upstream die lip and cannot be carried forward by the moving web; this defect occurs at high flow rate and low coating speed. On the other hand, spreading failure refers to the situation that coating solution cannot spread uniformly on the moving web, owing to low flow rate and high coating speed. Air entrainment would appear for the case with high flow rate and high coating speed, owing to the movement of the upstream meniscus as shown in Fig. 2c into the slot exit. Two additional coating defects were observed for the cases with a carrier layer. Figure 4 displays the photos of these two types of defects, i.e., edge withdrawal and rivulets; these photos were taken from the CCD camera positioned above the moving web. Figure 4a is a uniform coating; a stable coating of the bottom carrier layer is established first, and then the top layer is coated on the bottom layer. The viscous top layer will suppress the dilute bottom layer and the bottom layer is forced to expand in the cross-web direction; consequently the edge of the bottom carrier layer stays slightly beyond the slot exit as shown in Fig. 4a. As the coating speed increases, if the flow rate of the bottom layer is fixed and the coating thickness of the carrier layer decreases, the coating width of the carrier layer shrinks and the edge moves closer to the slot exit as shown in Fig. 4b. If this trend continues, the edge moves further toward the die center as shown in Fig. 4c, and the dilute solution cannot form an even bottom carrier layer and breaks into several unstable bands. On the other hand, if the viscous top layer becomes too thin, the coated top layer will break into many rivulets, as shown in Fig. 4d. Since the maximum coating speed of the pilot coater is limited to 0.17 m/s, this research was focused on studying the minimum wet thickness, not the maximum coating speed of TWSC.

[FIGURE 4 OMITTED]

The effect of viscosity for the carrier layer was examined on TWSC first. Figure 5 presents the values of the minimum wet thickness for a top viscous coating solution with viscosity [[mu].sub.t] = 74.4 mPa s. Two solutions with viscosity [[mu].sub.b] = 7.4 and [[mu].sub.b] = 18.7 mPa s were used as carrier layers. Note that the subscripts "t" and "b" denote top and bottom layers, respectively. Web tension and wrapping angle were fixed to be 500 N/m and 1[degrees], respectively. The effects of these two variables on the minimum wet thickness will be discussed later. These two parameters were also fixed for data from Figs. 6-9. The effects of these two variables on the minimum wet thickness will be discussed later. The experimental results indicate that using a carrier layer with low viscosity can effectively reduce the coating thickness of the top layer. Furthermore, the combined coating thickness for both layers is lower than that of the top layer alone. The reduction is more significant for the carrier layer with a lower viscosity. This finding is similar to that of conventional slot die coating studied by Yu et al. (18).

[FIGURE 5 OMITTED]

[FIGURE 6 OMITTED]

[FIGURE 7 OMITTED]

[FIGURE 8 OMITTED]

[FIGURE 9 OMITTED]

The effect of the carrier layer thickness on the viscous top layer is shown in Fig. 6; here the viscosities of the top and carrier layers are 74.4 and 7.4 mPa s, respectively. Two flow rates for the carrier layer were selected; the flow rate per unit coating width [q.sub.b1] is 0.012 [cm.sup.2]/s, and another [q.sub.b2] is 0.014 [cm.sup.2]/s. Once [q.sub.b1] or [q.sub.b2] was selected, the flow rate of the bottom layer was fixed in the experiment. The thickness of the carrier layer can be estimated from Eq. 2 and it was reduced proportionally as the coating speed increased. The results in Fig. 6 indicate that the case with higher flow rate [q.sub.b2] is more effective in terms of reducing the thickness of the top layer, and the combined thickness is smaller than the case with the lower carrier flow rate [q.sub.b1]. If the thickness of the carrier layer was too low, the viscous top layer tends to force the bottom carrier to withdraw, shrinking the coating width of the bottom layer, and then broke into few unstable bands as shown in Fig. 4d. Therefore, a thicker bottom carrier layer can better resist the suppressing of the top layer and is more effective in reducing the top layer thickness. The data in Fig. 6 also indicate that even the carrier layer is thicker; the combined two-layer thickness is still smaller than the case with a thinner carrier layer.

The minimum wet thickness of the coating solution that is used as the carrier layer goes up as the coating speed increases as shown in Fig. 7. Two specific flow rates [q.sub.b1] and [q.sub.b2] of this solution were selected for stable double-layer coating in the flow experiment at low coating speed. It is also interesting to note that the thickness of the carrier layer with double-layer coating may be even smaller than that of the same solution coated as a single layer. The data in Fig. 7 indicated that even values of the coating thickness corresponding to [q.sub.b1] and [q.sub.b2] for stable double-layer coating are higher than the minimum wet thickness at low coating speed; the wet thickness of this layer for the stable double-layer coating can be even smaller than that of the solution coated as a single layer at coating speed V higher than 0.12 [m/s]. This implies that the addition of a top viscous layer tends to stabilize the bottom layer at a high coating speed.

The effect of the top layer viscosity was examined with two coating solutions that have viscosities 210 and 493 mPa s, respectively. Comparing with the data in Fig. 6, the data in Fig. 8 indicate that as the viscosity of the top layer is increased from 74.4 to 210 mPa s, the minimum wet thickness of the top layer coated as a single layer increases substantially. However, when a bottom carrier layer is introduced, values of the combined two-layer thickness are lower than those in Fig. 6; this implies that a large viscosity ratio is helpful in reducing the top layer thickness. The trend is more significant with a top layer viscosity [[mu].sub.t] = 493 mPa s; the data in Fig. 9 indicate that the minimum wet thickness is above 100 [micro]m for the top layer coated alone. However, the thickness of the combined layer is less than 20 [micro]m. The reduction of the coated thickness is over 80%.

The effect of the wrapping angle on the minimum wet thickness of the top layer was examined; the result is similar to that observed by Lin et al. (4) on single layer coating; the larger the wrapping angle, the smaller the minimum wet thickness. Reducing the surface tension of a coating solution usually results in a smaller operating window, particularly the minimum wet thickness for slot die coating and TWSC (4). The same trend was observed if the surface tension of the top layer was reduced. It was found that if the surface tension of the top layer is decreased, the minimum wet thickness will go up; this effect is more significant as coating speed increases. Increasing the web tension can also reduce the minimum wet thickness. Experimental data of these variables on the minimum wet thickness can be found (21).

Previously, researchers (4), (10), (11) found that the minimum wet thickness of TWSC was proportional to the tension number for single layer operation. In addition to viscous force and web tension, the inertial force can also be important because a dilute bottom carrier is involved. A Reynolds number Re can be defined as

Re = [[[rho][V.sub.s.sup.2]]/[[mu]V/H]]. (3)

Here [V.sub.s] is the average velocity is the slot. Physically, Re represents the ratio of the fluid inertia impinging on the moving substrate to the viscous drag force to pull the coating solution. Re can be calculated for top and bottom layers in the flow experiment separately. It appears that the maximum value of Re in the present study is less than 0.1; therefore, the effect of inertial force is negligible.

We made an attempt to correlate the data of the minimum wet thickness of the double-layer coating by defining a tension number [T.sub.N] as

[T.sub.N] = [[[[mu].sub.eff]V]/T]. (4)

Here the effective viscosity [[mu].sub.eff] is defined as

[[mu].sub.eff] = [[[[mu].sub.u][H.sub.u] + [[mu].sub.b][H.sub.b]]/[[H.sub.u] + [H.sub.b]]]. (5)

The effects of viscosity and thickness for both layers are included in the definition of [[mu].sub.eff]. Previously, researchers (4), (10), (11) found that the minimum wet thickness H is proportional to [T.sub.N] as

H = a [([T.sub.N]).sup.b] (6)

where a and b are constants. The slope b represents the dependence of H on [T.sub.N].

The dependence of the top layer thickness [H.sub.u] on the tension number [T.sub.N] is displayed in Fig. 10; values of a, b, and confidence [R.sup.2] that were determined by a least square fitting procedure are given in Fig. 10 for different cases. The slope b varies between 0.79 and 1.15 and [R.sup.2] is close to unity. Previous studies (10), (11) reported that b varies between 0.4 and 0.8 for single layer operation; therefore, the top layer thickness has a stronger dependence on [T.sub.N] for double-layer operation. On the other hand, the combined thickness H can also be plotted against [T.sub.N] as shown in Fig. 11. Values of b given in Fig. 11 indicate that b varies between 0.3 and 0.026, which are smaller than those for [H.sub.u]. This implies that once a dilute carrier layer is introduced, the dependence on viscous drag is also reduced. The effect is more significant with a thicker carrier layer; values of b for the cases with higher flow rate [q.sub.b2] are much smaller than those with [q.sub.b1] at the same top layer viscosity.

[FIGURE 10 OMITTED]

[FIGURE 11 OMITTED]

The economy of adding a carrier layer can be evaluated by computing the reduction of drying load with double-layer coating. For a single layer coating, if the wet coating thickness is [beta] and the solid content of the coating solution is S%, the drying load of this layer, [eta], can be defined as

[eta] = [beta](1 - [S/100]), (7)

where [eta] represents the thickness of an equivalent pure solvent layer that has to be evaporated by drying.

For the case with a carrier layer, the drying load of the double-layer coating, [delta], is defined as

[delta] = [[beta].sub.1](1 - [[S.sub.t]/100]) + [[beta].sub.b](1 - [[S.sub.b]/100]). (8)

[delta] stands for the thickness of the combined two solvent layers that have to be heated and removed from the wet coating film.

The reduction of the drying load with a carrier layer can be defined as

[PHI] = (1 - [[delta]/[eta]]) x 100%. (9)

[PHI] represents the energy that is saved by introducing a carrier layer.

Table 2 lists values of [PHI] calculated for different cases presented in Figs. 6, 8, and 9; the data clearly indicate a significant reduction of drying load is possible with a carrier layer, and 50-90% of the energy can be saved. The results of the case with top layer viscosity [mu] = 493 mPa s are extremely impressive; over 90% of the drying energy required can be saved. This demonstrates the advantage of introducing a dilute carrier layer for a viscous layer of TWSC. For practical production consideration, the upper layer is normally coated to give a desired dry coverage, and there are certain limitations to the maximum concentration attainable. However, up to that maximum, as the wet coverage decreases and the concentration increases, the viscosity of the upper layer would increase. This, as shown, would permit a still lower total wet coverage.

TABLE 2. The reduction of drying load for the double-layer coaling.

                                   Reduction of drying load (%)

Coating   Bottom layer flow   Top layer 1  Top layer 2  Top layer 3
speed    rate ([cm.sup.2]/s)    [mu]t =      [mu]t =      [mu]t =
(m/s)                         74.4 mPa s    210 mPa s    493 mPa s

0.13         qb1 = 0.012          46.8         66.6         92.1
0.13         qb2 = 0.014          50.3         70.0         92.6
0.14         qb1 = 0.012          43.5         68.1         91.9
0.14         qb2 = 0.014          47.3         71.7         92.7
0.16         qb1 = 0.012          42.6         69.8         90.7
0.16         qb2 = 0.014          46.2         72.5         92.6
             Average              46.2         69.8         92.1

Yu et al. (18) found that introducing a carrier layer can effectively reduce the minimum wet thickness of a viscous layer; the present study also found the same conclusion for TWSC. The wet thickness of the double-layer coating is less than 20 [micro]m for the top layer viscosity as high as 493 mPa s; this is much smaller than that of conventional slot die coating.

For conventional slot die coating, the coating gap is preset before the operation. However, owing to possible collision between the slot die and the backroller, the coating gap cannot be as small as that one wishes to reduce the coating thickness. On the other hand, the coating gap with the flow channel between the coating solution and the moving web is formed based on the balance of all forces, the die lip shape should be properly designed to avoid scratching the web, and a parallel or slightly converging flow channel is preferred to deliver the stable coating flow with the possible lowest minimum wet thickness. This is the advantage of TWSC over the conventional slot die coating.

CONCLUSIONS

The possibility of introducing a dilute carrier layer to reduce the wet thickness of a viscous layer for TWSC was examined experimentally. The flow experiment was carried out in a pilot coater with PVA solutions as test fluids and PET film as moving substrate.

The wet thickness ratio of the top to bottom layer has to be properly selected to obtain stable and uniform double-layer coating. If the bottom layer is too thin, edge withdrawal of the bottom layer would appear and the coating width of the bottom layer will shrink and then break into few unstable bands. If the top layer is too thin, it will break into rivulets.

It was found that the wet thickness of a viscous layer can be reduced substantially by a dilute layer. The reduction is more significant with a higher top/bottom viscosity ratio. The wet thickness of the viscous layer can be reduced over 50% by introducing a dilute bottom carrier layer.

The effects of surface tension of the top layer, wrapping angle, and web tension on double-layer coating were also examined. Increasing surface tension, wrapping angle, and web tension can reduce the minimum wet thickness of the top layer; however, the most critical factor on the reduction of coating thickness is the viscosity ratio of the top to the bottom layer. A tension number that includes an effective viscosity was defined and the minimum wet thickness of the top layer was found to be proportional to this modified tension number.

The advantage of introducing a bottom carrier layer can be demonstrated by calculating the reduction of drying load of the double-layer coating. It was found that the drying load can be reduced 50-90% by introducing a bottom carrier layer. The larger the viscosity ratio, the higher will be the reduction. This study also found that the minimum wet thickness for a top viscous layer combined with a dilute carrier layer is much smaller than that of double-layer slot die coating.

ACKNOWLEDGMENT

The authors thank Prof. Carlos Tiu of Monash University, Australia, for help with comments and suggestions.

REFERENCES

(1.) N. Chino, K. Tanaka, Y. Hiraki, H. Chikamasa, and N. Shibata, U.S. Patent 4,717,603 (1988).

(2.) N. Shibata and T. Sato, U.S. Patent 5,108,795 (1992).

(3.) S. Takahashi and N. Shibata, U.S. Patent 5,202,164 (1993).

(4.) F.H. Lin, C.M. Liu, T.J. Liu, and P.Y. Wu, Polym. Eng. Sci., 47, 841 (2007).

(5.) K. Nakamura and N. Matsunaga, U.S. Patent 6,731,363 B2 (2004).

(6.) M. Miyatake, T. Masuda, and T. Shigematsu, U.S. Patent 6,773,121 B2 (2004).

(7.) J.Q. Feng, AlChE J., 44, 2137 (1998).

(8.) H. Lee and L.E. Scriven. "Tensioned-web Slot Coating: Study of the Coating Window," Presented at the 11th International Society of Coating Science and Technology Conference, Minneapolis. MN (2002).

(9.) E. Park, L.E. Scriven, and M.S. Carvalho, "Physics of Coating Tensioned-web over Slot Die," Presented at the 12th International Society of Coating and Science Technology Conference, Rochester, NY (2004).

(10.) S. Fuchigami, "Challenges & Opportunities in Flat Panel Display Industry," Presented at the First Coating Workshop, Hsinchu, Taiwan (2005).

(11.) M.S. Carvalho, "Modeling Coating Process: Fundamental Understanding Leading to Higher Productivity," Presented at the First Coating Workshop, Hsinchu, Taiwan (2005).

(12.) E.B. Gutoff and E.D. Cohen, Coating and Drying Defects, 2nd ed., Wiley, New Jersey (2006).

(13.) F.H. Lin, CM. Liu, T.J. Liu, and P.Y. Wu, Polym. Eng. Sci., 48, 307 (2008).

(14.) K.Y. Lee and T.J. Liu, Chem. Eng. Sci., 47, 1703 (1992).

(15.) E. Chin, T. Numata, and M. Sugawa, U.S. Patent 5,743,722 (1998).

(16.) M. Tomaru, T. Mandai, and H. Takekuma, U.S. Patent 6,548,117 B2 (2003).

(17.) M. Tomaru, T. Mandai, and H. Takekuma, U.S. Patent 6,730,359 B2 (2004).

(18.) W.J. Yu, T.J. Liu, and T.A. Yu, Chem. Eng. Sci., 50, 917 (1995).

(19.) S.Y. Lu, Y.P. Lin, and T.J. Liu, Polym, Eng. Sci., 41, 1823 (2001).

(20.) V. Chu, M.Z. Tsai, T.J. Liu, and C. Tiu, J. Appl. Polym. Sci. (in press).

(21.) H.C. Tsai, Analysis on Double Layer Tensioned-web Slot Coating, M.S. Thesis, National Tsing Hua University, Hsinchu, Taiwan (2008).

Han-Chih Tsai, Hsien-Ming Chang, Ta-Jo Liu

Department of Chemical Engineering, National Tsing Hua University, Hsinchu, Taiwan 30043, Republic of China

Correspondence to: Ta-Jo Liu: e-mail: [e-mail]

Contract grant sponsor: National Science Council. ROC; contract grant number: NSC96-2221-E-007-090-MY3.

DOI 10.1002/pen.21406