Going back a year or two, there were a number of significant challenges to the development of μLED technology, each of which was a potential showstopper. The main ones were:
• LEDs are of variable quality in terms of brightness and color. That makes it much harder to make displays as even and homogenous as viewers want and is the reason for the whole ‘binning’ process in large μLED display manufacturing
• A 4K display needs 25m μLEDs and an 8K TV needs 100m μLEDs. To get the cost down, the LEDs have to be <10μm. However, as the LEDs get smaller, they typically radiate less efficiently.
• The μLEDs must be transferred to a glass substrate
• Driving μLEDs is much trickier than driving OLEDs or LCDs
Micro LED technical bottlenecks such as chip efficiency; transfer yield, detection and repair must be resolved for product cost to be competitive. PlayNitride has launched RGB LED (flip chip) of 15μm x 30μm and single-color μLED (vertical chip) of 2.5μm x 2.5μm, said Li Yunli. Its mass transfer technology can achieve yields of 99.9 percent, still not sufficient for MP.
Eric Virey, an analyst from Yole Dévelopement gave a talk at the IMID event in Korea and will present at the OLED Summit. Virey presented solutions to this issues
• Variable Quality — Virey and Dr. Boris Kobrin of N-Tech Research both said that the variability issues are not solved, but the solution is close. The key is improving the homogeneity of the vapor deposition process used to make the LED wafers. Both analysts say that the MOCVD makers are confident that the next generation of tools will make LEDs that are much more consistent based on process improvements.
• Many companies are working on ways to analyze quality on the LED wafer and that could helpful in getting to production yields. For some time, companies have proposed a two step transfer process:
o First – Transfer the LEDs to 3 interposers, each with the correct number of LEDs for a single color
o Second – Transfer the LEDs on the transposer in bulk to the target substrate
• Transfer — Virey says that a 1.2mm square area of LEDs that could be moved to a 20 cm square interposer would then populate a TV. There would be up to 255 small areas from a 6-inch wafer and then the LEDs could be used to populate 46 4K TVs. Each TV would take 138 print steps in total.
• Efficiency Reduces with Size – μLEDs are very efficient in converting electricity into light, but very small sizes (>10 μm per side), the efficiency drops rapidly, due in part to edge issues that were not addressed in the past. PlayNitride (reportedly working with Samsung), Glō and X-Celeprint have all reported external quantum efficiency of 35 percent or more – good enough to make μLEDs the most efficient high-resolution display. Note that the EQE for OLEDs are ~20 percent. Virey claims that such a process would allow μLEDs to compete with the cost levels of OLEDs by around 2022.
Drivers — Macroblock, which supplies a lot of the drivers used in big LED walls say that the Samsung ‘Wall’ LED is using LTPS substrates as the driving technology. But tiling has always been much more expensive than a linear display and the Samsung Wall at IFA had visible bright and dark lines between some of the panels – a bit of an issue for a €300,000 or €400,000 display. Given that the costs of the μLED TVs are somewhat proportional to resolution, an equivalent 8K display would cost USD 700,000. OLED TV ASPs are expected to take a significant drop as explained below:
• 65-inch OLED TVs sell between USD 2500 and USD 3500.
• The use of printing should take down the cost by ~20 percent
• The use of Gen 10.5 would conservatively lower the cost by ~30 percent. A 65-inch panel built on a Gen 8.5 fab uses 63.4 percent of the glass, but 94.2 percent of the glass for a Gen 10.5, or a 50 percent difference.
• Therefore, a 50 to 70 percent reduction from technology is probable, while a conservative 10 percent would come from material cost reductions over a 4-5 year period.
Therefore, a 65-inch 8G OLED TV could have a price point of USD 1,200. Assuming a more reasonable ASP for a μLED consumer based TV of USD 100,000 today going down to USD 1200 in 2022, would mean an annual cost reduction of 33 percent and a total cost down of 99 percent. Virey said μLEDs are almost there and that the industry is now demonstrating solutions and are in the process of getting the costs and scale right – the last 20 percent of the problem. But the last 10% could take 10 years as it has with triple emitter blue, so getting to the 2022 target of costs down by 99 percent, seems like a stretch. – OLED Association