Edge-lit LCD backlight unit for 2D local dimming

Local dimming technology has been highly desired for integration with liquid crystal displays (LCDs) in order to improve their contrast ratios (CRs) as well as to overcome power efficiency bottlenecks. In this paper, we propose and demonstrate a slim (~1 mm) edge-lit LCD backlight unit (BLU) capable of 2D local dimming. We designed a semipartitioned light guide plate (LGP) patterned with inverse-trapezoidal microstructures, which allows the ultra-slim BLU to function without prism sheets. Since light emitting diodes (LEDs) are placed in the middle of the LGP, the BLU can freely define illuminated areas and the whole BLU can be modularly expanded like a tile canvas. The fabricated BLU achieves uniformity in both local and global luminance distributions, as well as in high local dimming performance. Experimentally, the BLU increases the CR of the display up to two orders of magnitude compared to conventional BLUs. © 2018 Optical Society of America under the terms of the OSA Open Access Publishing Agreement

. Moreove blocks. These image quality Therefore micropatterne studies impro challenging to blocks non-ad In this pap 2D local illum previous studi illumination, located away eliminate pris out-couplers o The prop However, the prove the usef the design of of the fabricat 2. Design of Figure 1 show uses 0.6 mmboth a uniform the middle of proposed desi Fig. 1 side v couple r, the lateral s e two inherent p y and power eff , advanced ed ed LGPs [9,10]  ctures were pat l so that ligh in Fig. 1(b). T adjusting the s he side angle is eets [14,15]. isolators that prevent bright eamless lightin LEDs. The is trench, and a r en local block ng sidewall. T y a small amou apes up from diate area betw ck is limited. A fabrication ma ign to complet l e ) o tterned as ht can be Therefore, side angle s in [13]), optically t hot spot ng among solator is reflective ks and its Therefore, unt of the the LGP ween local A metallic argin, the tely cover The following section will explain how the proposed BLU can make a uniform spatial luminance distribution when all of LEDs are lit, and how the isolators can prevent leakage of the light to the neighboring local block, even though there is a channel through which the light pass.

Design of the 2D local dimming LGP
We describe the design procedure in the following sequence: the isolator itself (Fig. 2), a single local block (Fig. 3), and all the local blocks ( Fig. 4) of the BLU. LightTools, which is an optical ray-tracing simulator, was used for the simulation. Figure 2 shows the results of optical simulations which are designed to explain how the isolators work. The isolator should act as a barrier to light even though there is an optical channel. At the same time, the channel area should not be darker than the illuminating block area for seamless illumination. The key is that as the channel thickness decreases the light flux entering the channel decreases while the light extracting efficiency of the channel increases. This results in a continuous and uniform illumination from the prior illuminating block as well as exhaustion of all light in the channel region escaping only a small amount of light to the next illuminating block. This is how we can get the uniform luminance distribution in the channel region ( Fig. 2(a)-2(c)). The details are explained below.

Analysis of the isolators
The forward-traveling light flux is proportional to the local thickness of the LGP when the thickness decreases abruptly. Figure 2(a) shows the relative flux of light entering the channel. The graph tells us that the light flux entering the channel is exactly proportional to the channel thickness. In other words, if the channel is thin enough, the isolator acts as a light barrier for local dimming.
Meanwhile, the light traveling in the channel region meets the surface of the LGP more frequently in a thinner channel as shown in Fig. 2(b). In the simulation, the pattern density and the input light flux were constant. Figure 2(b) shows that the light extracting efficiency is inversely proportional to the channel thickness. Figures 2(a) and 2(b) show that the thinner channel reduces the light that enters the channel and extracts most of that light.
Because the total amount of the illuminating light flux in the channel can be expressed as a product of input light flux and the light extraction efficiency, the spatial luminance is independent of the channel thickness as shown in Fig. 2(c). Therefore, the spatial luminance level can be maintained in the channel area even though there is much lower input light flux.
The entered light flux to the thin channel is rapidly reduced while passing through the channel. As shown in Fig. 2(d), the survived light rapidly decreases in the thinner channel because the thinner LGP results in a higher light extracting efficiency as mentioned above.
In other words, even though the channel area is bright, the light passing to the next local block can be dramatically reduced because i) only a small amount of light can enter the channel, ii) all the light entering to the channel is dissipated out (upward) that a small amount of light is escaped to the next local block, and iii) the recovered (increased) LGP thickness in the next local block reduces light extracting efficiency.
These simulations considered only the light with an incident angle of 25-35 ° because the inverse-trapezoidal microstructure used here can extract only this light in the normal direction (Lambertian light sources are used in all simulations except the simulation results in Fig. 2). In the simulation model, the inverse-trapezoidal microstructures are patterned in a hexagonal array with a 60 µm-pitch on the channel region, and the pattern is 5 cm away from the LEDs (far enough away to spread the light in the BLU). The dimensions of the structure are the same as in Fig. 1(b).  ation and optim for a uniform n in Fig. 3(a). hin the isolator multi-blocked L ous simulation LGP. This inc incident to th n density in the eal pattern den pitch of the mic erse-trapezoida a multi-blocke in Fig. 1 because there is no leakage from the previous blocks. Using the generated pattern distribution, we could get about 85.5% in spatial luminance uniformity, which is measured by the 9-point measuring method, and this value is comparable to a commercial level [15]. The optimization result is strongly tied to simulation set-up such as mesh size, LED distribution, the shape of the local block, and so on, so there is a chance to increase optical efficiency further with more fine tuning. Figure 5 shows the spatial luminance distribution of the proposed LGP. Each block can operate individually as shown in Fig. 5(a), where only one of the nine local blocks is turned on in sequence. The blocks in the first row are slightly brighter than the other rows. This is because the pattern must be denser in the first row in order to achieve uniform luminance when all LEDs are turned on, as mentioned in the previous section. Figure 5(b) shows a single LGP when only the center block is lit. Although some light enters into the neighboring blocks through the channels in the isolators, about 65% of all light was extracted in the target block. The confinement performance can further be engineered by controlling the channel thickness, width, and maximum pattern density more precisely. For example, an LGP with the same thickness and thinner channels (100 µm) shows that the isolation performance increased to 78.5% in other simulations.

Optical characteristics of the designed LGP
According to the simulation, the isolator region in the center block is about 30% darker than the main block area (Fig. 5(b)). This is because the pattern distribution is optimized in the simulation for the case that all LEDs are turned on. When the next block is also turned on, the luminance in the isolator region reached to 93% of that of the main block area.
The angular luminance distribution of the LGP was also investigated via simulations. Figure 6 shows the angular luminance of the LGP at several points without any additional films. When all LEDs are turned on, vertical directionality was observed at every point on the LGP (Fig. 6(a)-(c)). In the case of the single block operation, vertical luminance directionality was also maintained in the body area as shown in Fig. 6(d). However, at the isolators, the angular luminance distribution was asymmetric (Figs. 6(e)-6(g)). This is because of the relative position between the light sources and the measured points. This simulation confirmed that the LGP results in vertical illumination in the entire area including the channel region, and the local illumination is not affected by the angular luminance distribution.

Fabrication and demonstration of the proposed BLU
We fabricated a prototype of an LGP for demonstration. The fabrication process of the LGP is depicted in Fig. 7. We first made three components individually: patterned top sheet, reflective barrier sheet, and body sheets (Figs. 7(a)-7(c), respectively). The patterned top sheet was fabricated by using 3D diffuser lithography and PDMS replication process [11,16]. The PDMS was chosen as an LGP material for this particular process considering its high elastic characteristic which helps in replicating the overhang-shaped microstructure from a mold. If other properties are required as an LGP material, other fabrication processes can be used which were proposed previously [17][18][19]. The reflective barrier was made with a 50 µmthick PET film by using thermal evaporation and conventional photolithography process. Ag is suitable to use as a reflective layer because of its high reflectivity (96% in measurement), and the thickness of the layer was 200 nm; it is enough to screen the light from LEDs [20]. Flat and thick PDMS sheets were used as the bottom sheets. Finally, these three layers are bonded by liquid PDMS (as a glue) and cured. LGP, which ha Fig. 8(b) show P can define th e LGP and the shows a SEM se-trapezoidal m e slope was we on. hown in Fig. 9 diffuser sheet uniformity of th ig. 9(a)) and th ely, and the av mer case ( The angul and diffuser s area in a singl and 9(d), resp performance entire thickne simulation re distribution ch measured cha Figure 10 test image of 10(a)). The si 10(b) and 10( with the conv display the im Because th relies entirely one LEDs in t Fig. 10(b), the 9. Optical character nance distribution is illuminated, a ctively, of the LGP lar luminance sheets was also le block illumi pectively. As w without any a ess of the BLU esults in Fig.  haracteristic