When considering the trends in LEDs, we often divide our discussion in to categories like microLED, miniLED and Narrow Pixel Pitch (NPP) LED. Let me throw out the idea of discussing trends based upon the final display backplane technology. The four categories I suggest are silicon, transparent, PCB (printed circuit board) and flexible. Below are some initial thoughts on such a categorization that I welcome comments on.
NPP LED displays are traditional digital signage solutions that place LEDs on PCB substrates. These are typically passively driven using a multiplexed drive scheme. LED size and pitch vary, but NPP LED displays are typically those under 1.5 to 2 mm pixel pitches.
LEDs are tested and binned based on wavelength and light output. The popular Surface Mount Device (SMD) approach places a single red, green and blue LED in a single package which can then be assembled on the PCB using pick-and-place equipment. Small modules are assembled this way, with several modules combined into a cabinet. Multiple cabinets then form the final video wall display.
As NPP displays start to move to pixel pitches on 1.5 mm or less, some SMD limitations start to become evident – they can only be size-reduced so far. As a result, newer packaging techniques like flip chip and chip-on-board are being used to allow smaller LEDs to be packed more closely together. Secondly, the size of the LED emitter is shrinking. These so-called “miniLEDs” can create even denser LED displays – perhaps down to 0.7 to 0.6mm pitches.
There is no agreed upon definition of the size of the emitter for miniLEDs. I favor a range of 0.05mm (50 microns) to 0.3mm (300 microns) but others have suggested the range should be 100 to 500 microns. However, these smaller emitters and pixel pitches are pushing the limits of PCBs and the pick-and-place machines used t make the modules. For example, the line widths and geometries of PCB technology have limitations at these dimensions and the tolerances for placement of the LEDs may be too high to meet production needs with decent yields, depending on the equipment and processes used.
In addition to even more narrow pitch LED video walls, miniLEDs are also being considered for high-density backlights for an LCD displays. Such devices have already been commercialized as gaming monitors that can employ 10K to 25K LEDs in the backlight. The advantage of this is that the number of dimmable zones can be increased dramatically leading to high-performance HDR displays with reducing haloing. But control of all these LEDs with a passive drive scheme gets expensive as a lot of drivers are needed, driving up costs.
As a result, it seems some developers of miniLED backlights are moving to active matrix glass substrates. One huge advantage if this approach is that the backplane can be made in the LCD fabs and use conventional LCD drivers. But to be cost effective, traditional pick-and-place machines are not likely to cut it. Some sort of mass transfer approach will be needed. Many approaches have been proposed and developed – some that expand single pick-and-place equipment, some that use stamps to transfer dozen or hundreds at a time and some that are more selective transfer type. In a recent visit to LED maker Epistar, we learned that they have worked with Uniqarta to do 1K transfers per second of known good die to the glass substrate.
Such techniques can also be used to make transparent direct view displays such as the ones shown by X-Display (formerly X-Celeprint) and PlayNitride. X-Display transfers singulated red, green and blue microLEDs plus a driver IC onto a passive matrix backplane. They have developed a mass transfer process that they claim will soon hit 4-nines of quality. Their demo at DisplayWeek was a transparent full color display with 320×160 resolution, 3000 cd/m² and 40% transparency in a 4.6” x 2.3” format (70 PPI).
PlayNitride showed a borderless transparent 7.56” display using an LTPS active matrix backplane from Tianma. They also had a flexible backplane demo, perhaps on polyimide. They use a programable mass transfer head that can be used to transfer known good dies and to remove and replace bad die. Samsung has a relationship with PlayNitride as they were the source of the microLEDs used in Samsung’s 75” microLED TV unveiled at CES 2019.
Kyocera was showing a transparent 1.8” microLED display with 256×256 resolution and a 127 micro pixel pitch (200 PPI) with 20 micron emitters. It uses an LTPS backplane and RGB microLEDs made in cooperation with Glo. Brightness was specified at 3000 cd/m² and contrast at 1M:1.
Note that in the examples above, the size of the LED can vary from 20 microns (a microLED) to perhaps 75 x 125 microns, which is a miniLED. This is precisely why it may be better to look at the LED market from the substrate or process point of view rather that the LED size point of view.
But there will always be an exception to the rule. For example, there is a new quad RGB LED package that Aoto has demonstrated in a 1.5mm pitch videowall. I don’t know what substrate these 100-micron diameter LEDs are placed on, but I suspect PCB. The challenges I mentioned above apply to the placement of the red, next to the green, next to the blue LEDs with high accuracy (probably a micron or two). Solving this need is common for this class of application – and doing it quickly is the key to success.
Silicon substrates are the third category. These are required for the highest density and smallest emitter sizes. There are various ways to integrate the silicon and LED materials such as mating singlated LEDs displays to singulated silicon backplanes; growing GaN on silicon backplane wafers; and wafer-to-wafer bonding (GaN-on-silicon to silicon backplane). Plessey Semiconductor may the first to successfully demonstrate this last approach which may be able to lead to the lowest cost solution long term as the silicon wafers can scale to a 12” size.
Finally, I have added flexible substrates as a fourth category although you might argue this is part of the transparent category. I am separating it as the substrates don’t have to be continuous sheets like polyimide but could be strips of LEDs of various sizes.
All of these categories will have a lot of variations and different workflows within them. That is the nature of LED displays today – there are lots and lots of methods. It seems likely that certain methods will gain dominance, but their dominance may be confined to certain applications, sizes, pitches or substrates types.
Thinking of LED manufacturing by substrate type is merely a thought exercise that may or may not have value. It was simply a common theme that emerged while speaking with a number companies on a recent trip to Taiwan. (CC)
