TechBlick will soon host its virtual microLED and QD event, and here is a good preview of some of the technologies and advances that will be detailed during the event. MicroLED-Info readers enjoy a special discount to this interesting upcoming industry event.
We highlight important advancements in MicroLED and/or QD displays in this article using technology slides. These advancements of the art will be presented at TechBlick’s 2-day global conference on “Mini- & Micro-LED Displays: Markets, Manufacturing Innovations, Applications, Promising Start-ups” taking place online in TechBlick’s ‘in-person virtual’ platform on 30 Nov - 1 Dec 2022.
The agenda includes the likes of Samsung, Sharp, AUO, Coherent, ASMPT, Komori, CEA, Micledi, 3D Micromac, Allows Semiconductors, and many more. Full agenda can be seen here.
What is the future market for microLEDs in terms of unit sales, application, chip type, backplane?
The slides below share some key forecasts and analysits by Jerry Kang from Omdia - one of the leading analysts in the field - who will present live online at TechBlick's microLED and QD event.
Slide 1 shows that the microLED market will experience substantial growth in the coming years. In 2021, the sales were expected to be just 0.12k units, reflecting the lack of technology maturity and the manufacturing high cost/low yield. In 2029, the figure is expected to reach 12.7M units per area. This is a transformational growth, and yet still represents just 0.3% of the total FPD market!
Slide 1 also shows the split of the market - in thousands of units - by application. The first to arrive were public displays owing to low technical barriers and much larger dies as well as ultra low PPI. Premium TV have also arrived and will grow. Here, the superior performance of color gamut, luminance, and contrast delivers value. Smart watches will be a drive of unit sales, as here the small size and lower PPI translate to lower technical barriers.
Slide 2 shows the evolution of the market by backplane technology. MicroLED on Silicon (on CMOS) is suited to high PPI AR or HUD devices. LTPS backplane is ideal for small and medium sized applications like smart watches due to high mobility of LTPS but LTPS does not scale to large areas due to non-uniformity of p-Si. Oxide TFTs can be scaled to Gen8 whilst PCBs will be used for ultra large displays such as signage.
Slide 2 also shows the the forecast split by chip type. The on-wafer option means growing the uLED array directly on the epiwafer. Here, pixel can be very small but overall size limited by wafer size (6-12inch). Flip chip means that microLEDs will have bottom electrodes so that they can be flip chip mounted onto the target substrate/backplane
Join us on 30Nov-1 Dec 2022 to learn more about the technology and market for microLED and Qunatum dots. You will hear from a fantastic lineup of speakers including Samsung, AUO, Sharp, ST Micro, Omdia, Yole, Coherent, Allos, etc.
QD-Si Image Sensors- beating InGaAs and SiGe in the NIR and SWIR regions with 1.62-2.2um pixel pitch
It took roughly 20 years of R&D to commercialize colloidal quantum dots (CQD) image sensors, which are the first commercial products in the marketplace to use CQDs in electro-active devices in contrast to all of the other current products that use CQDs in photoluminescence mode.
First, why quantum dots? The particle size of PbS QDs can be tuned to absorb thoughout the SWIR region spanning from 1000nm to 2500nm. This is shown in slide below, showing also that these QDs can also absorb in the NIR, visible, and UV regions at the same time.
Second, why quantum dots + silicon? Obviously the most advanced imaging technology is based on silicon. However, silicon is not sensitive to NIR and SWIR. As such, InGaAs and SiGe sensors have taken hold of this market. However, they are often expensive and their heterogenous integration with silicon read-out circuity (ROIC) can add to complexity and limit pixel sizes/pitches although advances in Cu-Cu bonding may change this.
As shown in the slide below, The QDs can be spin coated atop a 300mm silicon wafer. Single PbS QDs are formed into thin QD films with a ligand matrix. Out of this, a QF (quantum film) photodiode if formed with top and bottom electrodes (which must be transparent to a broad light spectrum). These QD photodiodes are formed atop the BEOL of a top-side illuminated image sensor. Cu vias are then used to connect the QD layer to the image sensor.
In addition to bringing silicon image sensor technology to the NIR/SWIR spectrum, the QD technology can also enable 100% fill factor and help shrink a complex global shutter pixel (which otherwise would need a larger space for the photodiode)
In this study, to be presented at TechBlick's Quantum Dot and microLED conference on 30Nov-1Dec, Jonathan Steckel from ST will outline the state-of-the-art, showing how 1.62-2.2um pixel pitches have been demonstrated on 300mm wafers. The quantum efficiency (QE) is >60% (940-1400nm)
Mass Transfer of uLEDs: Overcoming dimensional/manufacturing variations with magnetic head/stamp tech
MicroLED display technology requires massive parallel transfer technology. This is a complex technology as it is and will grow even more complex as displays with smaller dies and high PPIs are considered. This has been one of the frontiers of development in the uLED industry.
Many parallel transfer approaches have been proposed. Most are based on a type of stamp which picks up the microLED dies from the growth substrate and transfers them onto the target substrates, placing them at the right spot.
A critical challenge is how to overcome inevitable height and dimensional variations of uLEDs using standard elastomer-based microLED technology, which, if not managed properly, can adversely impact that all-important figure-of-merit: yield!
LuxNour Technologies Inc. is proposing a novel approach based on electromagnetic stamps which can - as shown in slide 1- tolerate tens of σ in uLED variation! This increases yield and eases the pressure on exact control of the microLED dimensions during the growth.
Makarem Hussein in slide one shows the structure of such an electromagnetic head, containing a bulk electromagnetic at the back, a non-magnetic dielectric element in between, and a pattern of high permeability materials (e.g., Ni) and openings. The high permeability areas shield the bulk magnetic, preventing its flux from protruding out. In contrast, the openings represent discontinuities in the shield, allowing the EM force to penetrate out. Slide 2 shows a close up of the structure as well as the resultant magnetic flux
In this approach, the microLEDs will also require a layer of metallization with a ferroelectric material. As shown in slide 2, when the magnetic field is on, the microLEDs - regardless of height variations - are picked up by EM force at the location of openings/discontinuities. When the field is off, the dies are released (or placed).
Slide 3 shows an example of a100mx100mm stamp on a 150mm Si wafer. Here, the high-permeability material is nickel. This stamp can handle 15um microLED dies with spacing of just 7.5um.
This is a very interesting technology with excellent potential. Of course, there is significant know-how and expertise and technology in EM stamp/head development (see patent: ). Furthermore, the microLED wafer manufacturers must adopt their metallization step to depost a ferrous material.
Makarem Hussein will join speakers from Samsung, Sharp, AUO, ST, Coherent and many others to discuss the present and future of microLED technology on 30 NOV - 2 Dec - see agenda here www.TechBlick.com/microLEDs
MicroLEDs, printed electronics and laser printing?
Holst has developed and advanced the so-called LIFT technology to enable even the laser transfer or laser printing of microcomponents such as microLEDs with high 1-um precision.
In the first slide you can see a comparison of the classical Laser Induced Forward Transfer (LIFT) vs the technique developed at Holst which is Volume-Controlled Laser Printing (VCLP).
In LIFT, the laser illumination causes a jetting of low to medium viscosity inks onto the acceptor or target substrate. This is not a young technology and some consider it the digitization of screen printing. In the example here you can see 200um dots of conductive paste printed with the LIFT process.
The VCLP is different. Here, the laser releases fine droplets onto the target/acceptor substrate. Given the volume control, better resolution is accessible. The example herein shows 40um dots of highly viscous solder paste printed using VCLP technology. Note that here the high-throughput deposition of ultrafine interconnects, such as conductive adhesives and solder pastes, is from a structured carrier plate covered with a proprietary permanent release coating.
An important feature of the VCLP technique is the control of heat flux. Without this the laser printing can result in blurred or poor definition. To manage the heat flux, Holst has developed a proprietary “permanent” stack for clean, fine interconnect printing. In the second slide, you can see the positive impact of this layer in achieving well defined high resolution laser printing.
What is amazing is that not only inks and pastes (also adhesive and solders) can be printed, but also microcomponents such as microLED dies. In the schematic in slide 3, you can see the concepts. Here, the microLEDs sit on the proprietary “permanent” stack and are then laser released across the print gap onto the acceptor/target substrate. The dies can be <10um with a <5um dicing street. This technique can reach >10M UPH (units per hour) with 1um assembly accuracy.
The final slide shows examples of mini as well as microLED transferred using this technique. The mini LEDs are 125x125x80 um3 and the microLEDs are 60x60x10 um3.
This are incredible results and points towards a new high throughput laser printing technique able to print finelines of inks as well as highly viscous pastes, and mini and micro components.
Monolithic chip-scale integration of QD color converters with GaN microLED chips
Manufacturing in particular mass transfer and repair remain the biggest challenges in realization of microLED displays. To simplify the process, many propose to transfer only the blue (or even a UV) LED and achieve RGB by depositing color converters such as quantum dots. This requires an additional deposition step. But what if one could achieve chip-scale monolithic integration of QDs and GaN microLED chips? Indeed, Saphlux proposes just this and will present this on 30Nov-1Dec 2022 at TechBlick's microLED and QD event.
Saphlux has taken this approach one step further. Instead of post-transfer deposition of QDs, their technology enables chip-scale integration of quantum dots and microLEDs. In this technique, a nano-porous structure is directly formed inside LEDs to serve as a natural vessel for in-situ QD integration. The effective light path can be extended by nano-porous structure to boost the overall efficiency due to the strong scattering effect. The reliability of quantum dots is - it is claimed- also improved greatly because of the high thermal conductivity of gallium nitride material.
Importantly, this technology claims to be able to integrate red, green, and blue pixels monolithically into a single chip to greatly reduce the complexity and cost of Micro-LED display manufacturing.
In this technique, nanopores in LEDs can be created by dipping the material in an acidic solution and applying a bias voltage, which drives electro-chemical etching of n-type GaN. By adjusting the etching voltage, one can change both the porosity and the size of the nanopores. Then the blue LED are bonded with exposed nanopores GaN to a current-driver panel and red/green QDs are selectively loaded to achieve Micro-LED full-color conversion.
Hybridization or monolithic integration of GaN microLEDs on Si CMOS drivers: technology review
microLEDs can be directly integrated with CMOS drivers (instead of usual TFT backplane) enabling high-PPI displays suited to AR/MR glasses, metaverse, and even in some cases large-area displays. The key technological challenge for this is the hybridization and/or monolithic integration of GaN microleds and CMOS. This is no easy feat as it involves heterogenous hybridization or integration of two different material systems: GaN and Si.
Over the years, several technologies have been proposed by companies and institutions to hybridize the two parts. They are ranging from hybridization techniques to full monolithic 3D integration. These options – explored from over the last decade from 2011 to 2022- are depicted in the slide below, which offers a clear categorization of the techniques, e.g., hybridization vs monolithic integration, direct vs indirect bonding, align vs non-aligned, etc.
In this presentation, François Templier from CEA-Leti will review these techniques and explain challenges for their fabrication at TechBlick's specialist event on microLEDs on 30NOV-1Dec . Some examples of solutions will be given, such as microtube technology and recent results with hybrid bonding.
R2R grown GaN LEDs on metal foils instead of expensive sapphire wafers?
R2R grown GaN LEDs and perhaps even GaN/AlGaN HEMT transistors on metal foils instead of expensive small-area sapphire substrates? This could be a breakthrough technology, bringing the robust and efficient inorganic LED technology to large-areas. In microLED displays, it could mean monolithic integration, leading to mobile-sized and large displays manufactured without a wafer-to-substrate transfer step.
As shown below, iBeam Materials is developing such technology. It first planarizes a rough metal foil and then uses an ion beam to form a nm-thick layer with aligned grains. This 'template' then acts as the growth substrate in lieu of, say, a sapphire wafer.
As seen below, this technology has already been used to demonstrate a functional GaN LED as well as a GaN/AlGaN HEMT. In July 2021 (when the results were presented at TechBlick) the PL was upto 70% of normal LEDs. However, a direct comparison is not yet fair as the standard approach benefits from decades and decades of accumulated know-how and production expertise.
Currently, the LEDs are still not done in a R2R fashion, although the 'template' can be R2R manufactured on 20"-wide substrate. The next step of development will involve demonstrating a R2R MOCVD GaN growth. The R2R production of the template is not the bottleneck, but the growth of a thick (5um or so) GaN LED
Finally, Vladimir Matias argues that this technology has the potential to lower cost of production by a factor x25. A detailed cost analysis is shown below, demonstrating the technical milestones which need to be achieved to enable this cost roadmap.