Liquid crystal displays (LCDs) have been dominating flat-panel display markets for a long time, because extensive research has long been performed to drop the cost and enhance the image quality. Spontaneously, many liquid crystal (LC) modes have been proposed, such as in-plane switching (IPS) [1, 2], fringing field switching (FFS) [3-5], and vertical -alignment (VA) modes [6, 7]. Among these modes, the VA mode has been widely used in LCD television (TV), due to its advantageous electro-optic properties, such as fast response time and high contrast ratio in the normal direction. On the other hand, a few unsolved draw-backs remain, such as color shift and gamma distortion, which originate from the initially vertical alignment. Many works have been reported to overcome such drawbacks with their own particular features, e.g. multidomain vertical alignment (MVA), which has protrusions on the surface [8, 9]; advanced super-view (ASV) [10]; patterned vertical alignment (PVA), which utilizes an oblique electric field formed by electrodes of desired patterns [11-13]; and polymer-stabilized vertical alignment (PS-VA), in which a pretilt angle is achieved by a certain polymer morphology on the surface [14-16]. In all VA modes, vertically aligned LCs tilt down in four different azimuthal directions, to improve the viewing-angle property. However, retardation of the LC layer causes a color shift as the viewing angle changes from normal to oblique [17]. As a result, the viewing-angle property was further improved by dividing each pixel into two regions, to produce different voltages, which can give rise to different LC behavior in each region [18-22]. According to the driving method, the 8-domain mode can be classified according to many types:, including the capacitor-coupling type (C-C Type), which can generate different voltages for each domain by adding a capacitor [23], and the two-transistor type (T-T Type), which independently controls the voltages in desired regions [21, 24]. More recently, the charge-shared type (CS Type) was proposed, in which upper and lower adjacent pixels are charged simultaneously using one-gate two-data [22] or half-gate two-data [25], and also the resistivity-division type (RD Type) [26] was reported, in which a pixel is divided into high-pixel (high-tilt-angle zone) and low-pixel (low-tilt-angle zone), without extra sharing gate lines and additional data lines. However, all of the technologies mentioned above give rise to reduced aperture ratio, due to the additional data lines and transistors in the 8-domain mode.

To overcome these drawbacks, here we propose a novel method to form 8-domains by controlling the width and space of electrodes in both main regions and subregions, in combination with the polar anchoring strength.

The normalized transmission of light through the VA cell in which a vertically aligned uniaxial LC medium is driven by an oblique electric field under a crossed polarizer is given by

where ψ is the voltage-dependent angle between one of the transmittance axes of the crossed polarizers and the LC director; d is the cell gap; Δn_eff is the effective birefringence of the LC medium, dependent on voltage and viewing angle (θ = polar angle, ϕ = azimuthal angle); and λ is the wave-length of incident light. In the PS-VA mode, the LC director is initially vertically aligned, but an extra process such as polymer stabilization under an electric field is performed to give a pretilt angle of about 89° in four ϕs. In the voltage-off state, an excellent dark state is achieved, due to the very small value of Δn_eff. In the voltage-on state, ψ (V) should be 45° and dΔn_eff (V) should be λ/2, to achieve maximum transmittance in the normal direction. In addition, the LC director should reorient in four azimuthal directions, to minimize the viewing-angle dependence of the dΔn_eff in 4-domain PS-VA mode. Nevertheless, quite strong color shift of the image as displayed in the normal direction does occur for oblique viewing directions of the 4-domain device, because dΔn_eff changes abruptly as θ increases at a fixed ϕ[20]. To reduce such a color shift, an 8-domain VA mode is proposed, in which one pixel is divided into two regions and the LC director in each region has two different tilt angles, low-tilt angle θ_A and high-tilt angle θ_B, while keeping a 4-domain in each region, as shown in Fig. 1(a).

The reorientation of vertically aligned LCs is determined by competition between the elastic energy of the LC cell and the applied electric field. In two regions of the device, the same LC is filled so that the field response to the LC in each region is the same. On the other hand, the magnitude of the anchoring energy, especially the polar anchoring energy W_θ, can determine the tilt angle of the LC director at a given voltage, because the main deformation of LCs is associated with bending [27, 28] and also the field intensity determines the tilt angle of the LC director. In PS-VA mode, the pixel electrode is patterned like a fishbone [29] with electrode width w and distance l between electrodes, as shown in Fig. 1(b). When the l in region A is different than that in the region B, the average field intensity that the LCs experience is different; that is, the narrower l is, the higher the field intensity becomes, such that the tilt angle of the LC director in region A can be smaller than that in the region B, though the same voltage is applied to both regions.

Figure 1. (a) Configuration of LC directors in on-state of the 4-domain and 8-domain VA modes, and (b) side-view of fishbone electrode structure, indicating that the tilt angle of the LC director in region A is smaller than in region B, due to difference in field intensity, though the same voltage is applied to both regions.

To test the feasibility of the proposed 8-domain PS-VA mode, we used a three-dimensional simulation program (TechWiz LCD, SANAYI system Co., Korea) to simulate the electro-optical properties of the proposed structures. The following conditions were set for the simulation: the negative dielectric anisotropy Δε = -4, and the birefringence Δn = 0.0816 at 589.3 nm. The surface pretilt angle was 89° (in experiment, this can be achieved by curing a reactive monomer in the PS-VA mode [15-17]), and d was 4 μm. Planar indium tin oxide (ITO) and fishbone-shaped ITO electrodes were used on the top and bottom substrates respectively. The pitch of the patterned electrode in each domain was kept constant while w and l in the fishbone-shaped electrodes varied, as shown in Fig. 2. For the 8-domain PS-VA mode, a pixel was divided into a main region and subregions, and the simulation was done in three sets: cases 1, 2, and 3, as shown in Table 1. In both cases 1 and 2 the same electrode structures were set as w (l) = 1.6 (4.4) μm in the main region and w (l) = 2.7 (3.3) μm in each subregion, as shown in Fig. 2(a), while the electrode structures in the main and subregions were interchanged in case 3, as shown in Fig. 2(b). For the anchoring conditions, in case 1 both W_θ and azimuthal W_φ anchoring energies were 10^-3N/m, representing strong anchoring conditions [30]. In case 2, the polar anchoring energy was set differently as W_θ = 10^-3N/m and 10^-4N/m in main and subregions respectively, while the azimuthal anchoring energy was the same in both regions, W_φ = 10^-3N/m. Finally, in case 3 the anchoring energies were also interchanged between main and subregions. In all cases the ratio of the main-region to sub-regions was 2:1.

Figure 2. Top view of fishbone-shaped electrode with different electrode width and distances between electrodes in two regions of a pixel: (a) Cases 1 and 2; (b) Case 3.

TABLE 1.. Simulation conditions for the proposed 8-domain PS-VA mode.

We first simulated the voltage-dependent transmittance (V-T) of the conventional 4-domain with respect to the electrode width and space, and to the magnitude of W_θ at fixed W_φ, as shown in Fig. 3. In experiment, the anchoring energies of the alignment layer can be controlled by the degree of polymerization of a monomer in the PS-VA mode, and also by mixing planar and vertical alignment materials in the same layer [17]. As shown in Fig. 3(a), the threshold voltage V_th, which is defined as a 10% transmittance change from a dark state, is 2.9 V and 3.0 V at W_θ = W_φ = 10^-3N/m with l = 3.3 μm and l = 4.4 μm respectively. The fact that V_th with l = 3.3 μm is 0.1 V lower than that with l = 4.4 μm indicates that the narrower electrode spacing can contribute to a stronger vertical component of an applied electric field, giving rise to lower V_th. The V-T curves at w = 2.7 μm and l = 3.3 μm with different magnitudes of W_θ also reveal that lower anchoring energy results in lower V_th and lower operating voltage V_op, as shown in Fig. 3(b).

Figure 3. Voltage-dependent transmittance curves (a) with respect to w and l at W_θ = W_φ = 10^-3N/m, and (b) with respect to magnitude of W_θ at w (l) = 2.7 (3.3) μm in the conventional 4-domain case.

After confirming that the V-T curves are strongly related to electrode structure and the magnitudes of anchoring energies, we changed W_θ = 10^-4N/m in the region with w (l) = 2.7 (3.3) μm, while keeping W_θ = 10^-3N/m in the region with w (l) = 1.6 (4.4) μm, and then V-T curves were calculated as shown in Fig. 4(a). Here the magnitude of W_φ was fixed. As expected, V_th is lowered by 0.4 V at W_θ = 10⁻⁴N/m in the region with w (l) = 2.7 (3.3) μm, and also higher transmittance is observed. Based on this result, we designed the following four different cases, including cases 1, 2, and 3 introduced in Table 1, and tested the electro-optic performance. One additional case is the conventional 4-domain structure with w (l) = 1.6 (4.4) μm and W_θ = 10^-3N/m. Figure 4(b) shows V-T curves for all four cases. The V-T curves of the conventional 4-domain case and case 1 exhibit similar behavior with V_th = 3 V, while the V-T curves of cases 2 and 3 show lower V_th of 2.8 V and 2.7 V respectively, and higher transmittance of 8.1% and 10.4% respectively, owing to relatively weak polar anchoring energy.

Figure 4. Voltage-dependent transmittance curves (a) for two different sets of (w, l, W_θ) in the conventional 4-domain case, and (b) for the conventional 4-domain case with w (l) = 1.6 (4.4) μm and W_θ = 10^-3N/m, and the three-proposed 8-domain cases.

Figure 5 shows the pixel images in grayscale for all four cases. Cases 2 and 3 clearly exhibit a difference in transmittance between the two regions, and case 1 is relatively difficult to distinguish, especially at high transmittance. This indicates that the deviation of the tilt angle between two regions is greater in cases 2 and 3 than in case 1.

Figure 5. The luminance of pixels in several gray levels for (a) The conventional 4-domain case, (b) Case 1, (c) Case 2, and (d) Case 3.

Figure 6 shows the isoluminance curves at different gray-scale values in each case (T₁₀, T₅₀ and T₉₈; here the numbers in subscripts indicate relative transmittance with respect to maximal transmittance). Here the isoluminance curves are normalized to the maximal luminance in case 3 in the normal viewing direction, and the relative luminances 90%, 70%, 50%, and 30% are indicated by a solid line and the various dotted lines, as shown in the last column to the right. As shown in the first column for T₁₀ in Fig. 6, relative luminance of less than 30% exists within a polar angle of 40° in diagonal directions for the conventional 4-domain, cases 1 and 3, whereas it exists within a polar angle of 50° for case 2. In the third column for T98 in Fig. 6(c), remarkably, a relative luminance of 90% exists above a polar angle of 50° in every azimuthal direction in case 2. This indicates that case 2 shows the least luminance change, that is, wide viewing angle among all cases according to viewing angle changes.

Figure 6. Isoluminance curves at several gray values (T₁₀, T₅₀ and T₉₈) in (a) The conventional 4-domain case, (b) Case 1, (c) Case 2, and (d) Case 3.

In Fig. 7 are plotted the gamma curves, with a gamma correction factor following the power law γ = 2.2, for various viewing directions [31]. To simplify comparison, the azimuthal angle 0°, which is coincident with the transmittance axis of one of the polarizers, is fixed, while changing the polar angle from 0° to 60°. As clearly shown in the plots, the gamma curves in the conventional 4-domain case and cases 1 and 3 exhibit grayscale inversion upon changing the viewing direction. Unlike the others, case 2 shows the best characteristic with respect to gamma distortion among all cases. To investigate this quantitatively, the gamma distortion index (GDI) is calculated with the following expression [32]:

where L_i,on and L_i,off denote luminance corresponding to the ith gray level on-axis and off-axis respectively, and <> denotes the average for all cases of arbitrary gray levels. A smaller value of GDI means better off-axis image quality. The calculated GDI for polar angle 30° (60°) is 4.25 (14.17) in the conventional 4-domain case, 4.50 (15.71) in case 1, and 4.08 (13.84) in case 3. These results imply the image quality of case 3 is more improved than the two other cases, while case 1 is even worse than the conventional 4-domain case. Remarkably, although the only difference of case 2 from case 1 is W_θ of the sub-region, the GDI of case 2 is reduced to 3.28 (23% less than the conventional 4-domain case) and 13.10 (8.1% less for polar angles of 30° and 60° respectively. Therefore, case 2 shows huge reduction in gamma distortion, and would display the best image quality among the proposed cases. Images along normal and oblique viewing directions are calculated to visualize the effects of the reduced GDI in the proposed 8-domain VA device, as shown in Fig. 8. As clearly indicated, the original image in the normal direction is best perceived in case 2 for the viewing-angle direction with polar angle of 60° at azimuthal angle of 0°, as compared to the other cases. These simulated images also demonstrate the outstanding effect of the proposed 8-domain, which is optimized with respect to pixel structure and anchoring energy.

Figure 7. Viewing-angle-dependent gamma curves at several polar viewing angles and fixed azimuthal angle 0°: (a) The conventional 4-domain case, (b) Case 1, (c) Case 2, and (d) Case 3.

Figure 8. Calculated images at normal viewing angle and oblique viewing angle with polar angle of 60° and azimuthal angle of 0°: (a) The conventional 4-domain case, (b) Case 1, (c) Case 2, and (d) Case 3.

We proposed an 8-domain PS-VA mode in which the main-region and sub-region have different widths and spacings of slits in their electrodes, and also the anchoring energies in each region have different amplitudes. The optimization of both parameters gave rise to lower gamma distortion index at oblique viewing angles than that of the 4-domain PS-VA, while keeping both high contrast ratio in the normal direction and the same transmittance as that of the 4-domain PS-VA. We believe the proposed design can contribute to the development of a PS-VA mode with high resolution, high image quality, and low power consumption.