Learn about Micro-LED display technology in one article

The growth history of Micro-LED

LED technology has been developed for nearly three decades. It was initially used as a new type of solid-state lighting source. Although it was later applied to the display field, it was still just a behind-the-scenes hero - the backlight module. Today, LED is gradually moving from behind the scenes to the front stage, ushering in the most vigorous development period. Now it has appeared in various important occasions many times and plays an increasingly important role in the display field.

▲ Figure 1 Application of LED in ① Bird's Nest ② Water Cube ③ Shanghai World Expo

The reason why LED has become the focus of attention is mainly due to its many unique advantages. It is not only self-luminous, small in size, light in weight, and high in brightness, but also has the advantages of longer life, lower power consumption, faster response time, and stronger controllability. This makes LED have a wider range of applications and gives birth to more high-tech products.

▲ Figure 2 Large-size LED display (lower resolution)

▲ Figure 3 Comparison between 8×8 LED array and micro-LED array

Nowadays, large-size LED displays have been put into use in some advertisements or decorative walls. However, their pixel sizes are large, which directly affects the fineness of the displayed image. When the viewing distance is slightly closer, the display effect is unsatisfactory. At this time, micro-LED display came into being. It not only has all the advantages of LED, but also has obvious characteristics such as high resolution and portability.

There are two main trends in the development of micro-LED displays. One is the main focus of Sony, which is small-pitch, large-size, high-resolution indoor/outdoor displays. The other is the wearable devices (such as Apple Watch) that Apple is launching. The display part of such devices requires high resolution, strong portability, low power consumption, and high brightness, which are exactly the advantages of micro-LED.

Micro-LED display has been developed for more than a decade, during which many project teams around the world have released results and promoted the further development of related technologies. For example, in 2001, the Japanese Satoshi Takano team announced a set of micro-LED arrays they studied.

The array uses a passive drive method and wire bonding to connect pixels to the drive circuit. The three red, green and blue LED chips are placed on the same silicon reflector to achieve colorization through the RGB method. Although the array has achieved initial success, it also has shortcomings that cannot be ignored. Its resolution and reliability are still very low, and the forward conduction voltage of different LEDs varies greatly ^[1] .

In the same year, HX Jiang's team also made a passive matrix-driven 10×10 micro-LED array. This array innovatively used four common n-electrodes and 100 independent p-electrodes. It also adopted a complex layout design to optimize the wiring layout as much as possible. Although the display effect has improved to a certain extent, it has not solved the problem of low integration capability ^[2] .

▲ Figure 4 10×10 array wiring layout of HX Jiang’s team

Another outstanding achievement was announced by the Hong Kong University of Science and Technology team in 2006. They also adopted passive drive and flip-chip technology to integrate the Micro-LED array ^[3] . However, the forward conduction voltage of pixels in the same row also varies greatly, and when the number of pixels lit in the column is different, the brightness of the pixels will also be affected, and the brightness uniformity is not good enough .

▲ Figure 5: Hong Kong University of Science and Technology team’s achievements

In 2008, ZY Fan's team announced another passively driven 120×120 microarray, with a chip size of 3.2mm×3.2mm, a pixel size of 20×12μm, and a pixel spacing of 22μm. The size has been significantly optimized, but a large number of wires are still required, and the layout is still very complex ^[4] .

In the same year , the microarray published by Z. Gong's team still used passive matrix drive and flip-chip technology for integration. The team made a blue (470nm) micro-LED array and a UV micro-LED (370nm) array, and successfully excited green and red quantum dots through the UV LED array, proving the feasibility of quantum dot colorization. ^[5] 。

▲ Figure 6 UV micro-LED array

▲ Figure 7 Integration of Micro-LED array and Si-CMOS

In addition, in that year, BR Rae's team successfully integrated a Si-CMOS circuit that can provide a suitable electrical pulse signal for UV LEDs and integrated a SPAS (single photo avalanche diode) detector, which is mainly used in portable fluorescence lifetime readers. However, its driving ability is relatively weak and the operating voltage is very high. ^[6] 。

In 2009, the team led by ZJ Liu of the Hong Kong University of Science and Technology used UV micro-LED arrays to excite red, green and blue phosphors to obtain a full-color micro-LED display chip ^[7] . In 2010, the team used red, green and blue LED epitaxial wafers to prepare a 360 PPI micro-LED display chip ^[8] , and integrated the three chips together to realize the world's first full-color micro-LED projector without backlight source ^[9] .

▲ Figure 8 The world’s first full-color micro-LED projector without backlight

Later, the Hong Kong University of Science and Technology team led by ZJ Liu and the Sun Yat-sen University team jointly raised the resolution of micro-LED display to 1700 PPI, reduced the pixel pitch to 12 microns, and adopted passive site selection + flip-chip packaging technology ^[10] . At the same time, they also successfully prepared a WQVGA active site selection micro-LED display chip with a resolution of 846 PPI and integrated optical communication functions into the chip ^[11] .

▲Figure 9 1700 PPI micro-LED micro display chip

These are just some of the more important achievements in the history of micro-LED development. Since then, the exploration of micro-LED has continued to deepen, and more progress has been announced, including further reducing the size, improving the uniformity of brightness, etc. There are also many discussions about its driving method, preparation process and colorization, which will be introduced in the subsequent series.

Author: Liu Zhaojun Zhang Ke

Three main technical means of colorization of Micro-LED display

Colorization of Micro-LED display is an important research direction. In today's severe trend of pursuing colorization and high resolution and high contrast ratio, major companies and research institutions around the world have proposed a variety of solutions and are constantly expanding. This article will discuss several major methods for realizing colorization of Micro-LED, including RGB three-color LED method, UV/blue light LED + luminescent medium method, and optical lens synthesis method.

1. RGB three-color LED method

The principle of RGB-LED full-color display is mainly based on the basic principle of color matching of the three primary colors (red, green, and blue). As we all know, the three primary colors of RGB can be combined to synthesize most colors in nature after a certain ratio. Similarly, by applying different currents to red, green, and blue LEDs, their brightness values ​​can be controlled, thereby realizing the combination of the three primary colors and achieving the effect of full-color display. This is the method commonly used in large LED screens ^[1] .

In the RGB color display method, each pixel contains three RGB tri-color LEDs. The P and N electrodes of the tri-color LEDs are generally connected to the circuit substrate by bonding or flip-chip bonding. The specific layout and connection method are shown in Figure 1 ^[2] .

After that, a dedicated LED full-color driver chip is used to drive each LED with pulse width modulation (PWM) current. The PWM current drive method can achieve digital dimming by setting the current effective period and duty cycle. For example, an 8-bit PWM full-color LED driver chip can achieve 2 ⁸ = 256 dimming effects for a single-color LED. So for a pixel containing a three-color LED, theoretically, 256*256*256=16,777,216 dimming effects can be achieved, that is, 16,777,216 colors can be displayed. The specific driving principle of full-color display is shown in Figure 2 ^[2] .

However, in fact, since the actual output current of the driver chip may be different from the theoretical current, each LED in a single pixel has a certain half-wave width (the narrower the half-peak width, the better the color rendering of the LED) and light decay phenomenon, which in turn causes deviation problems in the full-color display of LED pixels.

▲ Figure 1 Schematic diagram of single pixel layout for RGB full-color display

▲ Figure 2 Schematic diagram of RGB full-color display driving principle

2. UV/blue light LED+luminescent medium method

UV LED (ultraviolet LED) or blue LED + luminescent medium can be used to achieve full color. If UV micro-LED is used, red, green and blue luminescent mediums need to be stimulated to achieve RGB three-color ratio; if blue micro-LED is used, red and green luminescent mediums need to be matched, and so on. This technology was patented by Professor Liu Jimei and Professor Liu Zhaojun of the Hong Kong University of Science and Technology in 2009 and has been authorized (patent number: US 13/466,660, US 14/098,103).

Luminescent media can generally be divided into phosphors and quantum dots (QD ). Nanomaterial phosphors can emit light of a specific wavelength under the excitation of blue or ultraviolet LEDs. The light color is determined by the phosphor material and is simple and easy to use, which makes the phosphor coating method widely used in LED lighting and can be used as a traditional micro-LED colorization method.

Phosphor coating is usually applied to the sample surface by spin coating or dispensing after the micro-LED is integrated with the driving circuit. Figure 3 shows an application of the phosphor coating method, where (a) shows that a pixel unit contains four sub-pixels of red, green and blue, and (b) shows the color effect after the micro-LED is lit ^[3] .

This method is intuitive and easy to understand, but it has its shortcomings. First, the phosphor coating will absorb some energy, reducing the conversion rate; second, the size of the phosphor particles is relatively large, about 1-10 microns. As the size of micro-LED pixels continues to decrease, the phosphor coating becomes more and more uneven and affects the display quality. This gives quantum dot technology an opportunity to shine.

（a）                                 （b）

▲ Figure 3 Pixel design and display effect of phosphor colored micro-LED

Quantum dots, also known as nanocrystals, are nanoparticles composed of II-VI or III-V elements. The particle size of quantum dots is generally between 1 and 10 nm, which can be used for smaller micro-displays. Quantum dots also have electroluminescence and photoluminescence effects. They can emit fluorescence after being stimulated. The color of the light is determined by the material and size. Therefore, the wavelength of different light emission can be changed by adjusting the particle size of quantum dots.

The smaller the quantum dot particle size, the bluer the luminescent color; the larger the quantum dot, the redder the luminescent color. Quantum dots have diverse chemical compositions, and their luminescent colors can cover the entire visible range from blue to red. They also have high light absorption-luminescence efficiency, very narrow half-width, and wide absorption spectrum, so they have high color purity and saturation. They also have a simple structure, are thin, and can be rolled up, making them very suitable for micro-display applications ^[4] .

At present, spin coating and mist spraying technology are commonly used to develop quantum dot technology, that is, using a sprayer and airflow control to spray uniform and size-controlled quantum dots. The device and principle diagram are shown in Figure 4 ^[5] . It is coated on the UV/blue light LED to stimulate it to emit RGB three-color light, and then full colorization is achieved through color matching, as shown in Figure 5 ^[5] .

However, the main problem with the above technology is the uniformity of each color and the mutual influence between each color. Therefore, solving the separation of red, green and blue colors and the uniformity of each color has become one of the important difficulties in the application of quantum dot light-emitting diodes in micro-displays.

In addition, the current quantum dot technology is not mature enough, and there are still disadvantages such as poor material stability, high heat dissipation requirements, sealing, and short life. This greatly limits its application scope, but as the technology advances and matures, we expect quantum dots to have the opportunity to play a more important role.

▲ Figure 4 (a) High-precision atomization spray system (Aerosol jet technology) and (b) its schematic diagram.

▲ Figure 5 Schematic diagram of the red, green and blue primary color array produced using high-precision spraying technology

3. Optical lens synthesis method

Lens optical synthesis method refers to synthesizing RGB three-color micro-LEDs into full-color display through an optical prism (Trichroic Prism). The specific method is to package three red, green, and blue micro-LED arrays on three packaging boards respectively, and connect a control board and a three-color prism.

After that, the image signal can be transmitted through the driving panel, the brightness of the three-color micro-LED array can be adjusted to achieve colorization, and an optical projection lens can be added to achieve micro-projection. The physical picture and schematic diagram of the entire system are shown in Figure 6, and the display effect is shown in Figure 7 ^[6] .

▲ Figure 6 Prism optical synthesis method a), b) actual picture, c) principle diagram

▲ Figure 7 Display effect of prism optical synthesis method

author:

Liu Zhaojun, Peng Deng, Zhang Ke (Sun Yat-sen University)

Guo Haozhong, She Qingwei (National Chiao Tung University, Taiwan)

Decoding the three different driving methods of Micro-LED

Liu Zhaojun Zhang Ke

Micro-LED is a current-driven light-emitting device, and its driving mode generally has only two modes: passive addressing drive (PM: Passive Matrix, also known as passive addressing, passive addressing, passive driving, etc.) and active addressing drive (AM: Active Matrix, also known as active addressing, active addressing, active driving, etc.). This article will also analyze a "Semi-active" site selection drive method These modes have different driving principles and application features. The following will introduce their principles in detail through circuit diagrams .

The passive addressing drive mode connects the anode (P-electrode) of each column of LED pixels in the array to the column scan line (Data Current Source), and connects the cathode (N-electrode) of each row of LED pixels to the row scan line (Scan Line).

When a specific Y-th column scan line and X-th row scan line are selected, the LED pixel at the intersection (X, Y) will be lit. The entire screen can be scanned point by point at high speed in this way to realize the display image, as shown in Figure 1. [1,2] This scanning method has a simple structure and is relatively easy to implement.

But The disadvantages are that the connection is complicated (X+Y connections are required), the parasitic resistance and capacitance are large, resulting in low efficiency, the pixel light-emitting time is short (1 field/XY), resulting in low effective brightness, easy crosstalk between pixels, and high frequency requirements for scanning signals.

Another optimized passive addressing drive method is to add a latch in the column scanning part, which is used to store the column scanning signals (Y1, Y2... ... Yn) of all pixels in the Xth row at a certain moment in advance in the latch.

When the Xth row is selected, the above Y1-Yn signals are simultaneously loaded onto the pixels [3] . This driving method can reduce the column drive signal frequency and increase the brightness and quality of the display. However, it still cannot overcome the inherent defects of the passive site selection driving method: complex wiring, easy crosstalk, and the inability to save pixel selection signals . The active site selection driving method provides a good solution to the above difficulties.

In the active matrix drive circuit, each Micro-LED pixel has its own independent drive circuit, and the drive current is provided by the drive transistor. The basic active matrix drive circuit is a two-transistor single-capacitor (2T1C: 2 Transistor 1 Capacitor) circuit, as shown in Figure 2 [4] .

Figure 2 Active site selection drive method

At least two transistors are used in each pixel circuit to control the output current. T1 is a gate transistor, which is used to control the on or off of the pixel circuit. T2 is a driving transistor, which is connected to the voltage source and provides a stable current for the Micro-LED within a frame.

There is also a storage capacitor C1 in the circuit to store the data signal (Vdata). When the scanning signal pulse of the pixel unit ends, the storage capacitor can still maintain the voltage of the gate of the driving transistor T2, thereby continuously driving the Micro-LED pixel until the frame ends.

The 2T1C driving circuit is just a basic pixel circuit structure of active-selection Micro-LED. It is relatively simple and easy to implement . However, since its essence is a voltage-controlled current source (VCCS), and Micro-LED pixels are current-type devices, it will bring certain difficulties in controlling the display grayscale , which we will discuss in the later section "Colorization and Grayscale of Micro-LED".

Dr. Liu Zhaojun’s research group has proposed a 4T2C current proportional Micro-LED pixel circuit that uses a current controlled current source (CCCS) approach and has advantages in achieving grayscale [5] .

Another method worth mentioning is a “semi-active” location-selective driving method [6] . This driving method uses a single transistor as the driving circuit of the Micro-LED pixel (as shown in Figure 3), which can better avoid crosstalk between pixels.

Compared with passive site selection, active site selection has obvious advantages and is more suitable for current-driven light-emitting devices such as Micro-LED. The detailed analysis is as follows:

① Active site selection has stronger driving capability and can achieve larger area driving. However, the driving capability of passive site selection is affected by the driving performance of external integrated circuits, and the driving area and resolution are limited.

② Active site selection has better brightness uniformity and contrast . In the passive site selection method, due to the limited driving capability of the external driver integrated circuit, the brightness of each pixel is affected by the number of lit pixels in this column. Generally speaking, Micro-LED pixels in the same column share the driving current of one or more output pins of the external driver integrated circuit.

Therefore, when the number of pixels lit in the two columns is different, the driving current applied to each LED pixel will be different, and the brightness of different columns will vary greatly. This problem will be more serious in large-area display applications, such as LED TVs and LED large screens. At the same time, as the number of rows and columns increases, this problem will become more serious.

③ Active site selection It can achieve low power consumption and high efficiency . Large-area display applications require relatively large pixel density, so the electrode size must be reduced as much as possible, and the voltage required to drive the display will also increase greatly. A lot of power will be lost in the row and column scanning lines, resulting in low efficiency.

④ High independent controllability . In passive addressing, higher driving voltage will also bring a second trouble, namely crosstalk. That is to say, in passive addressing LED array, the driving current theoretically only passes through the selected LED pixel, but other surrounding pixels will be affected by the current pulse, which will eventually reduce the display quality. The active addressing method avoids this phenomenon well through the pixel circuit composed of the gate transistor and the drive transistor.

⑤Higher resolution . Active site selection drive is more suitable for high PPI and high resolution Micro-LED displays.

Although the third "semi-active" drive can better avoid crosstalk between pixels, it cannot fully achieve all the advantages of the active site selection drive method listed above because there is no storage capacitor in its pixel circuit and the driving current signal of each column needs to be modulated separately.

Taking the blue light Micro-LED epitaxially grown on a sapphire substrate as an example, there are four ways to connect the pixel and the driving transistor T2 as shown in Figure 4. However, since the LED epitaxial growth structure is p-type gallium nitride (GaN) on the surface and n-type gallium nitride on the bottom, as shown in Figure 5.

From the perspective of manufacturing process, it is more reasonable to connect the output end of the driving transistor to the p-electrode of the Micro-LED pixel, that is, (a) and (c) in Figure 4. In Figure 4 (a), the Micro-LED pixel is connected to the source of the N-type driving transistor. The non-uniformity of the electrical characteristics of the Micro-LED caused by the non-uniformity of epitaxial growth, manufacturing process, and device aging will directly affect the VGS of the driving transistor, resulting in non-uniformity of the displayed image.

The Micro-LED pixel in Figure 4 (c) is connected to the drain of the P-type driving transistor, which can avoid the above influence, and its current-voltage relationship is shown in Figure 6. Therefore, it is more appropriate to drive Micro-LED with a P-tube pixel circuit.

Figure 6 Current-voltage relationship between Micro-LED and driving transistor

The real technical difficulties of MicroLED

Until now, LEDs have not been used as direct light-emitting elements, i.e. pixels, in fine-pitch displays. This is due to a number of issues, including cost and manufacturing feasibility. However, the idea of ​​using MicroLEDs and sub-millimeter pixel pitches to produce displays dates back to the early days of LEDs.

Interest in developing displays based on MicroLEDs has grown significantly over the past five years, especially after Apple acquired Luxvue in 2014. Last October, Facebook acquired Oculus, an immersive virtual reality technology company; in May this year, Sharp acquired another MicroLED startup, eLux, and recently Google invested in Swedish MicroLED manufacturer Glo.

Given these acquisitions, it is clear that microLEDs are not just in the lab. So why are these big brands so interested in this technology? Because microLEDs can use independent red, green and blue sub-pixels as independently controllable light sources, which can form displays with high contrast, high speed and wide viewing angles.

In fact, MicroLED displays are much better than their OLED counterparts because MicroLEDs have a wider color gamut, higher brightness, lower power consumption, longer life, greater durability, and better environmental stability. In addition, as shown in Apple's recent patent documents, MicroLEDs can integrate sensors and circuits to achieve thin displays with embedded sensing functions, such as fingerprint recognition and gesture control.

Although MicroLEDs have not yet entered the market, they are not just ideas on paper. At the "International CES" in January 2012, Sony exhibited a 55-inch MicroLED display with 1920×1080 pixels, containing 6.2 million sub-pixels, each of which is an independently controllable MicroLED chip, which received strong media attention. However, Sony has not given a timetable for commercialization, and so far, no MicroLED TV has entered the market.

MicroLED is an inherently complex technology

Today, there is no universally accepted definition for MicroLED. However, in general, MicroLED is considered to be an LED chip with a total surface of less than 2500 mm2. This is equivalent to a square of 50mm×50mm, or a circular chip with a diameter of 55mm. According to this definition, microLED is already on the market today: Sony made another appearance in 2016 in the form of a large LED video wall with a small pitch, where the traditional LED package was replaced by MicroLED.

The technology to manufacture MicroLED displays involves everything: processing LED substrates into MicroLED arrays ready for pick-up and transfer to receiving substrates for integration into a non-uniformly integrated system: the display. The display in turn integrates LEDs, pixel driver transistors, optics, etc. Epiwafers can hold hundreds of millions of MicroLED chips.

There are two main options for realizing MicroLED displays. One is to pick up and transfer MicroLEDs individually or in groups onto a thin-film transistor drive matrix, similar to what is used in OLED displays; the other is to combine a complete monolithic array of hundreds of thousands of MicroLEDs using CMOS drive circuits.

If the first of these two methods is adopted, assembling a 4K display requires picking, placing, and individually connecting 25 million MicroLED chips (assuming no pixel redundancy) to a transistor backplane. Using traditional pick-and-place equipment to manipulate such a small device, the processing speed is about 25,000 units per hour. This is too slow. It will take a month to assemble a single display.

To solve this problem, dozens of companies such as Apple and X-Celeprint have developed large-scale parallel gripping technology. They can process tens of thousands to millions of MicroLEDs at the same time. However, when the size of MicroLED is only 10μm, it is very challenging to process and place with sufficient precision.

There are also some issues to overcome related to LED chips. When their dimensions are very small, their performance is affected by sidewall effects associated with surface and internal defects such as open bonding, contamination and structural damage. These defects lead to accelerated non-radiative carrier recombination. Sidewall effects can extend to distances similar to the diffusion length of carriers (typically 1mm to 10mm): this is not important in conventional LEDs, which have edges of hundreds of microns, but it is very critical in MicroLEDs. In these devices, it can limit the efficiency of the entire volume of the chip.

Due to these defects, the peak efficiency of MicroLEDs is typically less than 10%, and when the device size is below 5mm, its peak efficiency may be less than 1%, which is far lower than the current best traditional blue-emitting "macro" LEDs, which can now produce external quantum peak efficiencies of more than 70%.

To make matters worse, MicroLEDs typically have to operate at very low current densities. They are typically driven below the 1-10 A cm-2 peak efficiency region because even at this low efficiency, LEDs are very bright. If a phone with MicroLEDs were running at its highest efficiency, its display would offer up to tens of thousands of nits of brightness, a level higher than the brighter phones currently on the market. The screen would be so bright that even the most daring users would not dare to look at it.

When LEDs operate at very low current densities, their efficiency is very low, preventing the technology from fulfilling its promise to cut energy consumption. Therefore, solving this problem has become a priority for MicroLED companies. Ways to improve efficiency include introducing new chip designs and improving manufacturing techniques. Both of these methods can reduce sidewall defects and keep electrical carriers away from the edges of the chip.

Developers of MicroLEDs also face challenges related to color conversion, light extraction and beam shaping.

Another requirement for modern displays is to eliminate dead or defective pixels. It is unlikely to achieve a 100% combined yield in epitaxy, chip manufacturing and transfer, so MicroLED display manufacturers must develop effective defect management strategies that can include pixel redundancy and single pixel repair, depending on the characteristics and cost of the display.

The fields where MicroLED is most easily realized at present

MicroLEDs are capable of being deployed in any display application, from the smallest to the largest. In many cases, they will be better than the ultimate combination of LCD and OLED displays. However, production feasibility and economic costs limit their use. However, detailed analysis shows that smart watches and other wearable products, such as microdisplays for AR/MR applications, best demonstrate the performance of MicroLED displays.

Among them, the implementation of MicroLED on smartwatches is the most likely, because smartwatches have a relatively small number of pixels and a mid-range pixel density, so the chip and assembly cost is efficient and is closest to the current state of MicroLED technology development. They have potential differentiated features, including the ability to extend battery life, reduce power consumption, and higher brightness, thereby providing good readability in outdoor environments.

If these displays begin to appear in large numbers, various sensors can be introduced in the plane of the front of the display, such as reading fingerprints and providing gesture recognition.

Another major opportunity for MicroLED is head-mounted displays for augmented reality (AR) and mixed reality (MR). In virtual reality, users wear a fully enclosed head-mounted display to isolate them from the outside world visually; while AR and MR applications overlay computer-generated images onto the real world.

MicroLED displays are made by cutting wafers into tiny devices.

and transferred to the transistor substrate using parallel pick and place technology

One of the requirements for these applications is that the overlay image be bright enough to compete with ambient light, especially in outdoor applications.

To meet these conditions, the display must be placed unobtrusively, using composite projection or waveguide optics with an optical efficiency of less than 10% to project the image to the eye. These requirements dictate that the brightness of the display range from 10,000 to 50,000 Nits, which is 10 to 50 times brighter than the best mobile phones on the market.

Today, MicroLED is the only candidate with the potential to provide these brightness levels while maintaining reasonable power consumption and compactness. Encouragingly, the same reasoning can be applied to head-up displays in automobiles and other environments, which can be considered a form of AR.

The market where MicroLED is trying to make an impact is the smartphone. Currently, OLED displays already offer very good performance at a very competitive cost. If MicroLED is also involved, the size of the sub-pixels must be reduced to a few microns, which makes it more difficult to provide acceptable efficiency.

The potential for success in TVs is even higher. In this case, the drawback is the relatively low pixel density, which is about 100 mm in a 4K, 55-inch TV. The low density hinders the efficiency of transfer technology, as thousands of chips need to be moved per cycle, compared to hundreds of thousands in a smartphone or smartwatch. To thrive in this market, alternative, high-efficiency assembly technologies need to be developed.

Who holds the core technology of Micro LED?

In the production process of Micro LED, due to the miniaturization of components, there are many problems that need to be overcome or improved, and the transfer technology in the process is the key to whether the product can be mass-produced and meet the standards of commercial products.

According to the size of the display substrate, it can be roughly divided into two forms of transfer. The first is a small-size display substrate, which uses semiconductor process integration technology to directly bond the LED to the substrate. The representative manufacturer of technology is Taiwan Industrial Technology Research Institute. The second is for large-size (or unlimited size) display substrates. The pick-and-place technology is used to transfer the pixels on the Micro LED array to the backplane separately. Representative manufacturers are Apple (LuxVue), X-Celeprint, etc. Other manufacturers such as Sony, eLux, etc. also have related transfer technologies. Introduction to Micro LED related patents ♦ Taiwan Industrial Technology Research Institute

(A) Patent Name: Method for Transferring Light-Emitting Components and Light-Emitting Component Arrays
Announcement Number: TW I521690
Priority: US 61/511,137

This patent is about a method for transferring light-emitting components. The steps are to first form an arrangement of multiple LED arrays on a substrate 1, where one array is an LED of one color, for example, red light, green light, and blue light are each an array in FIG1.

​​The transfer process requires multiple welding steps to sequentially transfer the LEDs on substrate 1 to the predetermined positions of substrate 2. Therefore, as shown in FIG2, before each welding, a protective layer is first used to cover the LEDs that are not to be transferred, and then the conductive bumps of the LEDs to be transferred are connected to the pads of substrate 2. Finally, all the LEDs on substrate 1 are transferred to substrate 2.

Figure 1. Figure 3 of Patent TW I521690 (Source: TIPO)

Figure 2. Figure HJ of Patent TW I521690 (Source: TIPO)

This patent does not seem to specifically mention the size of the LED or words related to Micro LED, but its counterpart in the United States with the same priority mentioned that the light-emitting element is 1 to 100 microns, and the pitch can be adjusted according to the actual product needs, as shown in the text and table in Figure 3.

Figure 3. Patent US 14/583594 (Image source: USPTO)

(B) Patent Name: Method for Making Light-Emitting Element and Display
Announcement Number: TW I590433

This patent of the Taiwan Industrial Technology Research Institute is also related to the manufacturing technology of Micro LED, but its method is completely different from the previous one. First, an LED array is formed on a substrate, wherein a semiconductor epitaxial structure, a first electrode and a second electrode constitute a light-emitting diode chip, and the light-emitting element includes a light-emitting diode chip and a spherical extension electrode. After completion, the light-emitting element is removed from the substrate

Then, the light-emitting element is ejected through a nozzle. The friction between the light-emitting element and the nozzle causes the spherical extension electrode to carry an electrostatic charge, and the contact of the receiving substrate transmits an electrical signal through a circuit structure so that it also carries an electrostatic charge. In the embodiment of the specification, the spherical extension electrode carries a positive charge and the contact carries a negative charge.

As shown in FIG. 4, the light-emitting element is made to fall into the opening of the receiving substrate by, for example, shaking a sieve. Since the volume of the spherical extension electrode is larger than the volume of the light-emitting diode chip, during the falling process, the spherical extension electrode of the light-emitting element turns downward and falls into the hole to contact with all points.

Figure 4. Figures P, S, and T of Patent TW I590433 (Source: TIPO)

♦  Apple (LuxVue)

LuxVue was acquired by Apple in 2014. It has the most Micro LED-related patents among all manufacturers. In terms of transfer technology, it mainly uses electrostatic adsorption mass transfer technology.

Patent name: Micro device transfer head array
Announcement number: US 9548233 B2

In order to achieve better transfer efficiency, manufacturers using mass transfer technology continue to develop a variety of transfer heads. The special thing about Apple's patent is that its transfer head has a bipolar structure that can apply positive and negative voltages respectively. The

platform structure of the transfer head is separated in half by a dielectric layer to form a pair of silicon electrodes. When grabbing the LED on the substrate, one silicon electrode is positively charged and the other silicon electrode is negatively charged to pick up the target LED.

Fig. 5. Figs. 1B, 34, 35 of US 9548233 (Source: USPTO)

♦  X-Celeprint

Patent name: Micro device transfer head array
Publication number: US 2017-0048976 A1

X-Celeprint's mass transfer technology Micro-Transfer-Printing (μTP) uses a stamping head to apply pressure on the LED, using Van der Waals force to make the LED adhere to the stamping head, then pick it up from the source substrate, move it to the predetermined position on the target substrate, and then press the stamping head together with the LED to the target substrate to insert the connecting column on the LED into the backplane contact pad to complete the LED transfer.

Fig. 6. Figs. 5-6 of patent US2017-0048976 (Source: USPTO)

♦ eLux

According to reports, Hon Hai will acquire Micro LED startup eLux, which has two points worth noting in terms of patents. First, its transfer technology is different from the mainstream market, and second, its patents applied for in the United States use the CIP method to connect Sharp and its own patents in large quantities (as shown in Figure 8).

Patent name: System and Method for the Fluidic Assembly of Emissive Displays
Publication number: 2017-0133558 A1

eLux's transfer technology uses a brush bucket to roll on the substrate, and the liquid suspension contains LEDs, which then allows the LEDs to fall into the corresponding wells on the substrate.

Figure 7. Figs. 5-6 of patent US2017-0048976 (Source: USPTO)

Figure 8. eLux US patent status (Image source: USPTO)

Author: Wang Li

Excimer lasers improve Micro-LED manufacturing process

Original: Rainer Paetzel

Micro-LEDs (µLEDs) based on inorganic III-V semiconductors such as GaN hold great promise for making displays that are far more electrical efficient, brighter, more pixel dense, more durable, and more versatile than existing technologies. However, the transition from current LED devices (~200 µm) to µLEDs (~20 µm) will require technological innovation, especially in the assembly of µLED displays. This article shows how excimer lasers can be used to address two of the most challenging aspects of this process.

Laser Lift Off (LLO)

Most current LED manufacturing processes use sapphire wafers as substrates for MOCVD crystal growth due to their relatively low lattice mismatch and cost. However, sapphire is not an ideal carrier material for finished GaN LEDs due to its poor thermal and electrical conductivity, which limits the light flux that can be extracted. As a result, the process of producing high-brightness GaN LEDs requires a final step to bond the device to a final or temporary carrier and then separate the device from the "sacrificial" sapphire substrate. For µLEDs, the sapphire substrate must obviously be removed in order to manufacture the small, thin devices that make up flexible displays.

Figure 1. Schematic diagram of the process of removing a sapphire substrate by laser lift-off technology. a) Device crystal grows and attaches to the carrier substrate. b) Laser beam penetrates the sapphire substrate. c) Sapphire substrate is removed.

Laser lift-off using an excimer laser is the most common method for removing sapphire substrates. During the process, high-intensity laser pulses penetrate the sapphire substrate (which can be penetrated by an excimer laser beam with a wavelength of 248 nm) and directly illuminate the LED wafer. At the same time, the GaN layer absorbs a large amount of UV light and a thin layer decomposes into gallium and nitrogen. The resulting gas pressure pushes the device away from the substrate, separating the device from the substrate with almost no force on the device. The gallium can be washed off with water or dilute hydrochloric acid to keep the surface of the device clean.

In addition to wavelength, another important characteristic of excimer lasers is that the pulses are short (about 10-20 ns), which helps to suppress heat diffusion and minimize the thermal load on the device. In addition, the laser output of the excimer laser can form a thin and long beam with uniform energy distribution along two axes (flat-top beam). (Figure 2) For example, the energy uniformity of the 155 mm x ~0.5 mm beam provided by Coherent's UVblade system is better than 2% standard deviation (sigma). In this way, all processing areas will receive the same and optimal energy flux, thus avoiding the problem of energy overshoot or excessive thermal load during processing, which often occurs in other laser processes with Gaussian energy intensity distribution.

Figure 2. 155 mm laser beam profile of the UVblade (248 nm), including the minor axis (SA) and major axis (LA).

Note that the two axis scales differ by two orders of magnitude.

Excimer LLO is essentially a single-pulse process, so the laser beam uniformity and stability requirements are extremely high. Laser manufacturer Coherent has developed products that meet this demand, which provide excellent pulse stability (e.g. < 1% rms), which can greatly improve process control during processing and help users increase process window.

FIGURE 3. UVblade LLO system with LEAP excimer laser and beam optics.

During the operation, the excimer laser beam is scanned across the substrate, separating the devices by irradiating the entire processing area. If the focus is on high throughput, the beam is adjusted to completely cover the sapphire wafer (2", 4", or 6") in a single scan. This method requires a moderate laser intensity (e.g. 50 to 100 W). The stress in the film caused by the mismatch in the effective thermal expansion coefficient is uniformly relieved, further reducing the impact on the device. Therefore, this 248 nm method is the most common method to achieve LLO.

Another LLO strategy is to use a smaller beam and raster scan across the wafer. For example, Coherent has a UVblade system that produces a beam 26 mm long and 0.5 mm wide, which can cover a 2" wafer in just two passes. This typical system requires only 30 W of laser power at 248 nm. Raster scanning methods require controlled overlap of the individual shots in the scan direction, as well as overlap between scans.

Laser-induced forward transfer (LIFT)

Assembling high-resolution displays containing millions of µLED chips presents unique challenges. Here too, 248 nm excimer lasers are ideal for precisely lifting GaN off its original carrier. The nitrogen gas generated expands and creates mechanical forces on the µLED structure, pushing the chip from the original carrier to the receiving substrate. By combining a large cross-section beam, a mask, and projection optics, up to 1000 chips can be transferred in parallel with just one laser shot.

Another way to do this is to pre-assemble the µLEDs on a temporary carrier wafer or tape using a polymer adhesive. These adhesives are very UV-absorbent. When irradiated with an excimer laser, the adhesive undergoes a photochemical decomposition reaction, separating from the µLED chip and generating a force that pushes the chip toward the receiving substrate. The energy intensity required to irradiate the polymer tape or adhesive may be only one-twentieth to one-fifth of the energy required for LLO. This means that very high processing speeds can be achieved with only moderate laser intensities.

Figure 4. Schematic diagram of the µLED assembly process using LLO and LIFT.

In summary, excimer lasers, which have performed well in the fields of excimer laser annealing (ELA) for display processing and laser lift-off (LLO) for high-brightness LEDs, have also shown great potential in the emerging µLED field. Excimer lasers have the characteristics of short UV wavelength, short pulse, high energy, and high power, which make them extremely compatible with III-V materials commonly used in LED manufacturing. In particular, 248 nm excimer lasers can break the performance limitations of the 266 nm or 213 nm solid-state lasers currently used in this application field. This can promote the realization of high-productivity and cost-effective process strategies.

South Korea's KIMM develops new technology for mass production of Micro LEDs Reel-to-reel transfer process

The Korea Institute of Machinery and Materials (KIMM), under the Ministry of Science, ICT and Future Planning, announced on July 24 that it has developed the world's first Micro LED panel manufacturing technology using the "roll transfer process".

The institute's Nano Applied Mechanics Team developed the "Micro LED panel" production technology using the roll transfer process, which tripled the luminous efficiency and reduced power consumption by 50%. Using this research result, it is expected that Micro LED display manufacturing will be 10,000 times faster than manufacturing traditional LED displays.

The roll-to-roll transfer process is a patented technology of the Korea Institute of Machinery and Materials. The TFT element is picked up and placed on the required substrate, and the LED element is picked up and placed on the substrate with the TFT element, thereby completing the active matrix Micro LED panel that combines the two elements.

With the reduction of production steps, the production speed is greatly improved. The current die bonder used to manufacture traditional LED displays can mount 1 to 10 LEDs per second on the substrate, but with the roll transfer technology, more than 10,000 LEDs can be transferred per second. It takes more than 30 days to produce a full HD 2-megapixel 100-inch digital signage through current methods, but the roll-to-roll transfer process can complete the entire process in one hour and greatly reduce processing costs.

X-Celeprint produces Micro LED arrays μTP technology

μTP technology was first developed by John A. Rogers and others from Illinois University in the United States, who used sacrificial layer wet etching and PDMS transfer technology to transfer Micro LEDs to flexible substrates or glass substrates to produce Micro LED arrays. The technology was spun out to Semprius in 2006, and X-Celeprint obtained the Semprius technology license in 2013 and officially began operations in early 2014.

μTP technology, in simple terms, uses an elastic stamp combined with a high-precision motion-controlled print head to selectively pick up an array of micro-components and print them onto a target substrate .

Specifically, a microchip is first made on a "source" wafer, and then "released" by removing the sacrificial layer under the semiconductor circuit to detach the microchip from the original substrate. Subsequently, a microstructure elastic stamp that matches the "source" wafer is used to pick up the microchip and transfer it to the target substrate.

This technology can selectively adjust the adhesion between the elastic stamp and the transferred device by changing the speed of the print head, thereby accurately controlling the assembly process. When the stamp moves faster, the adhesion increases, causing the transferred component to detach from the source substrate; on the contrary, when the stamp is far away from the bonding interface and moves slower, the adhesion becomes very small, the printed component will detach from the stamp, and then be transferred to the target substrate.

The stamp mentioned above can be customized to pick up and print multiple devices at a time, thereby efficiently transferring thousands of devices in a short period of time, so this process can achieve large-scale parallel processing.

Micro-transfer process flow: Figure 1: The elastic stamp approaches the wafer; Figure 2: The elastic stamp picks up the chip; Figure 3: The elastic stamp approaches the target substrate; Figure 4: The stamp "prints" (places) the chip on the target substrate

According to X-celeprint, the technology has been verified in many "printable" micro-devices, including lasers, LEDs, solar cells and integrated circuits of various materials (silicon, gallium arsenide, indium phosphide, gallium nitride and dielectric films including diamond).

GaAs-based red microLED printing case

In most cases, the semiconductor device to be transferred is first released from a "source" wafer, a method that utilizes a sacrificial layer beneath the device layer.

The structure of silicon-on-insulator (SOI) wafers is to prepare a 5-micron-thick single-crystal silicon layer on a 1-micron-thick oxide layer (Box: Barrier Oxide). Then, various devices and integrated circuits are prepared on the single-crystal silicon layer using standard SOI transistor processing technology. It is not difficult to see that the oxide layer of the SOI wafer can be used as a natural sacrificial layer, so it will be a very convenient and readily available "source" wafer .

Briefly introduce SOI processing technology :

First, according to the CMOS process standard, the single crystal silicon layer on the surface of the SOI wafer is patterned using photolithography and etching processes to expose the Box layer below. Then the patterned single crystal silicon is packaged and protected. The BOx layer under the device is removed by hydrofluoric acid etching. During this process, the ILD and wiring layers are protected and will not be damaged.

When the Box layer under the device is completely removed, the device will be completely separated from the wafer and fixed in place by the tether in the device layer. During the transfer, the tether can be broken or cut in a controlled manner.

Gallium nitride transistors are made on Si wafers (111). Reactive ion etching (RIE) will pass through the device layer through the through hole and down to the silicon substrate to separate the individual devices. A silicon dioxide mask is used in this step. The silicon nitride layer is deposited by plasma enhanced chemical vapor deposition (PECVD). The silicon nitride layer can not only passivate the sidewalls of the device, but can also be used to form anchor and tether structures.

Before printing the GaN chip, a layer of semiconductor thin film resin is applied on the COMS wafer. After the micro-transfer is completed, the bottom layer resin is solidified, and then the titanium tungsten and aluminum metal stacks are sputtered and deposited, and then the thickness is reduced by wet etching to finally form the device connection.