​Forecasting new and early developing technology is like herding cats, they never seem to wind up where they were supposed to be, but despite the vast number of technical, process and potential unknowns, the folks at Trendforce are trying their hand at predicting the future for Micro-LEDs.  As Trendforce is based in Taiwan, and Taiwan is the locus for much of the LED industry, it makes sense that if anyone would want to wax optimistic about micro-LEDs it would Taiwanese analysts.  While they do not give much of their absolute data to us mere mortals, they do predict that annual revenue from Micro-LED Chips used in TV applications will be ~23m this year, growing slowly in 2022 (we ballpark at $50 -75m), a bit faster in 2023 (~$350m) and 2024 (~$900m), but spiking to $3.44b in 2025.
While the details are sparse, they do cite a number of challenges that need to be met in order to reach these goals, with the cost of the micro-LED chips themselves the biggest stumbling block, a logical conclusion given the almost 25m needed for a 4K TV and the necessity for almost absolute uniformity across such vast numbers, something the industry is unable to do presently.  As we have mentioned a number of times, the transfer process for such large numbers and such small devices, is also a major challenge, one that has spawned a number of competitive solutions that have yet to prove themselves effective, but they did mention one area in the development of micro-LEDs that tends to be overlooked, and that is testing. 
Standard practice for testing LEDs is photoluminescence, which allows larger LEDs to be ‘binned’ (classified) by their brightness, so systems such as LED backlights have uniformity.  However when LEDs drop to the sizes that will be needed for micro-LED displays that are within normal parameters (say 65” or 75”), it gets progressively harder to make those measurements and obviously more time consuming based on the number of micro-LEDs.  Without a repeatable and consistent ability to test micro-LEDs, particularly before they are transferred to a substrate, the technology will be unable to compete and the transfer process will become moot, so with open questions as to just these few challenges to the development of Micro-LEDs we see forecasts as laced with too many …but, if...’s to be of service.  Maybe it sells expensive reports, but while we commend those who can see into the future, we would rather see continuing updates on the progress of the challenges mentioned and those that will potentially appear when the first set are solved than forecasts that we know will have little relevance to actual commercial development and implementation. 

Note: As Micro-LEDs get more press, we thought it helpful to help generate some basic understanding of what micro-LEDs are and how they are made.  We will offer the series in relatively small chunks with the first below.  Stay tuned for more…
Micro-LEDs are being touted as a potential game-changing display technology, but not unlike other revolutionary display technologies, micro-LEDs will not come easily.  While they have some unique properties that make them well suited for displays, some of those same characteristics present potential problems that at best can take a significant time to conquer, and at worst, could keep the technology from ever being fully commercialized. That said, before we cast a negative pall over micro-LEDs, we also note that they have the potential to become a replacement display technology for almost any display application, once major stumbling blocks have been removed, so there is significant capital, time, and other resources being spent toward solving such problems, and similar challenges facing other display technologies were eventually addressed and met with solutions that enabled them to become major display technologies.
 In order to understand the potential of micro-LED technology, it is essential to understand what they are and how they are produced, with the understanding that few of these processes, materials, and equipment, are in either a final stage of development or have even met some consensus in the industry as to how they will be used or implemented, making the micro-LED space a fluid one, and not surprisingly, one that can change daily, weekly, or monthly.  There are some basics however that are.at least for now, inherent in understanding micro-LEDs, and we try to bring them to light below.
Micro-LED substrates
Micro-LEDs are similar to more typical LEDs used in LED lighting and as edge or direct-lit backlights used in LCD displays, however they are, as the name implies, considerably smaller.  There is no ‘official’ size that would make an LED be called a micro-LED, but on a general basis, standard LEDs range from about 5mm, which is the size of this ❿, down to .5mm or 500um.  Below 500um, such LEDs are called mini-LEDs, but once they go below 100um, they are generally called micro-LEDs.  For reference, a 5mm LED chip is .196” across, meaning about 5 could fit in a 1” line (no spaces) and 25 in a 1” square.  A 500um mini-LED is 0.019” inches wide, meaning 52 could fit in a 1” line, or 2,770 in a 1” square, and a 100um micro-LED is 0.039 inches across, with 254 in an inch and 64,516 in a 1” square.  Micro-LED displays used for near-eye applications such as AR or VR, can be as small as 2um, which is .0000787” wide, with 12,700 in an inch and 161.29m in a 1” square.

​Producing LEDs is usually done on a sapphire crystal substrates grown in a high temperature chamber using the Kyropoulos method, which uses high purity aluminum Oxide that is heated in a tungsten crucible with a ‘seed’ crystal on which the material ‘grows’ as the temperature is gradually decreased.  The resulting 'boule' which can be as large as 300 kg, is them sliced into wafers and polished.  While the sapphire crystal has a particular orientation, meaning the way in which the crystal structures are linked, the surface of the sapphire crystal must be free from imperfections, and given how small micro-LED are, even the smallest hairline crack or pit could ruin a large group of micro-LEDs.  Each wafer is microscopically examined before it is used to verify its usability and in some cases a microscopic map is made of the surface.
There are other potential substrates for LEDs, with SiC (Silicon Carbide), Silicon (Si), and Gallium Nitride (GaN) being most researched, as each has advantages and disadvantages (cost is a big factor with silicon being the cheapest and GaN being the most expensive), but for the purpose of this note, we are focused primarily on sapphire substrates.

Again in a typical setting, the wafer is polished by the wafer producer or outside service and then shipped to the LED producer.  The wafer is then examined again, cleaned, and loaded into a cassette that is used to automate the MOCVD (Metal Oxide Chemical Vapor Deposition) process.  The cassettes are loaded into a handler that sits between two MOCVD tools and loads them into the deposition chamber of each tool, which could be set-up to take eight 6” inch wafers or five 8” wafers.  We note that in terms of raw wafer surface area the 8 by 6” configuration is 20% larger which would imply higher throughput, but there are other considerations that also come into play in that decision, and there are also a number of single 8” wafer systems that clustered around a cassette handling system.. 
Once the wafer are placed in the MOCVD deposition chamber they are rotated individually at high speed (2,500 RPM) and depending on the particular MOCVD tool the entire susceptor plate (the metal wafer holder – Fig. 2) also spins.  In some tools the materials to be deposited enter the chamber through what is called a showerhead that is close to the wafers, with the spin allowing the wafers to be covered uniformly.  In other MOCVD tools the deposited materials are passed laterally across the spinning wafers in a configuration called ‘laminal flow’, which implies no turbulence in the materials, producing a uniform coating.  

​In the past the type of material flow corresponded to a particular manufacturer and was a selling point that meant choosing one tool vendor or another, but now Aixtron (ARXA.GR) produces tools in both configurations, which allows customers who prefer a particular configuration to vendor options.  Veeco’s (VECO) MOCVD tools tend to be laminar flow oriented.  As we have not seen any data that gives one configuration a distinct advantage over the other, it seems to be a personal preference of the customer, likely driven by what they already have.  It’s not a hard and fast rule but parts and service costs are always lower if the tools are similar, so there is some stickiness to the choice of vendors. There is also the choice of the orientation of the reactor (horizontal or vertical), again a preference of the customer.

Materials are released into the reaction chamber with a carrier gas (usually Nitrogen , Argon, or Hydrogen) where the wafers are heated to between 500®C and 1500®C, with some reactors being hot-walled, meaning the walls of the chamber (usually graphite or quartz) pick up the heat in the reaction chamber, while others are cold-walled, where coolant keeps the reactor walls at a lower temperature than the wafer surface, which means it accumulates less reactant materials.  For most LED type structures the materials that are sent into the reactor chamber are known as precursors, meaning they are not what is needed to create the final LED layers but are reactants that combine on the surface of the wafer to create the final structure. 
Typical precursors are Trimethylgallium or TMG, TrimethylIndium (TMI), and Trimethylaluminum (TMA), which supply the heavy metals, Arsine (AsH3) and Ammonia (NH3).  The materials react to each other on the surface of the wafer, hence the term Chemical Vapor Deposition and byproducts are pumped from the chamber.  We note that these are volatile and dangerous materials and are extremely hazardous to the environment, so the cost of disposing of such materials and agents used to clean the chamber must be included in any cost of ownership calculation.

Each layer of the materials needed to create the LED structure is deposited through similar chemical reactions until the entire wafer is coated with what amounts to a stack of materials.  Once the deposition process is completed, the wafers are removed from the chamber and move to the etching process where resist, etch materials and masks are used to remove some of the layers to create individual LED structures.
More to come…

​Micro-LED TVs are a hot topic of conversation in the CE space, as we have noted many times, and while they are touted as the ‘next’ display technology, the technology comes with a number of issues that make predicting the Micro-LED takeover of the display space a bit challenging.  Among the technology’s biggest proponents is Samsung Electronics (005930.KS), who has been pushing the technology in the B2B space by producing a modular system that can be built into almost any vertical or horizontal configuration.  As with any new technology, the early iterations of such products are quite expensive and virtually impractical in all but the most unusual situations, but Samsung has faced similar challenges before and pushed on despite the technology’s trials and tribulations.
More to prove that it can be done, Samsung adapted this modular Micro-LED technology to what it calls a residential TV product, albeit one that costs in the neighborhood of $100,000, and promised to augment the 146” 2018 version with a smaller(!) 110” version, which it has done.  That said, the company also promised to release an even smaller version, a tiny 99” model in the 1st half of the year, which seems to have passed by.  As expected, a Samsung spokesperson said that the delay was due to the ‘high demand’ for the 110” product, but sources have indicated that Samsung has been having trouble adapting the technology to the ‘smaller’ size sets.
While this seems a bit odd considering a 99” TV screen is 86” by 46” and Micro-LED chips are less than 100um each, Samsung is trying to make a transition with the new ‘smaller’ sizes, which has presented additional difficulties.  A 4K TV has almost 8.4m pixels, each with a Red, Green, and Blue subpixel, and since each one of those sub-pixels is a Micro-LED it means that almost 25m very small LEDs have to be picked up from a wafer and moved to a substrate., a very time-consuming task using even the most advanced transfer technology.  Aside from the large numbers and the fact that red LEDs cannot be produced on the same wafer as blue and green LEDs, the inevitability of ‘bad’ Micro-LEDs in such a large array means that time must be taken to remove and replace a number of defective LEDs, adding to the cost of producing such a device.  In order to reduce the transfer time, Samsung has been looking for ways to produce all three color LEDs on a single wafer, which is a way to allow the transfer of 8.3m pixels containing red, green and blue micro-LEDs, rather than 24.9m single color micro-LEDs.
While Samsung is not the only one looking to make such a technology upgrade, the company does have it own LED production lines and experience in commercial LED production, but it seems that such a process is taking longer than expected to master and has delayed the ‘smaller’ models of Samsung’s micro-LED TV line, and will probably delay the 86” and 76” models that the company promise to release before the end of this year.  We certainly don’t fault Samsung for trying, although it is far better to under promise and over-deliver than to miss aggressive timelines, but we expect that if Samsung does miss its residential Micro-LED targets this year, we will see such TVs at various shows in 1H next year, and relatively early in the year, but as to when they will become available to consumers and at a cost that is at least just under nose-bleed level is likely more of a 2023 item; we hope.

​At the request of parent Samsung Electronics, display affiliate Samsung Display (pvt) has been developing an LTPS backplane designed for mini-LED displays.  Samsung Electronics has been offering a line of 8K and 4K mini-LED TVs and plans to expand their offerings later this year with larger size offerings up to 98”.  The backplane for such large mini-LED TVs has been built on PCB circuit boards but as Samsung Electronics is also developing smaller mini-LED models where the mini-LEDs are packed more closely together, mounting on PCB boards using typical production methodology limits the LED density.  Since the number of mini-LEDs in a 4K display is over 24m, and each is driven by its own circuitry, it becomes necessary to use thin-film technology as mini-LED display sizes get smaller.
SDC is expected to produce these TFT backplanes at it Gen 4 line in Cheonan, Korea, where it had previously produced rigid OLED panels.  That production has been moved to its A2 Gen 6 line, which is more efficient, leaving A1 in question as to usage.  As we noted last week (see our note 07/09/21) Samsung Display has been shifting production as it expands its OLED notebook line and offers production of same to outside customers, and has been toying with the idea of closing A1, which is its oldest OLED fab (2007).  While we expect, if the project (code name “M-Project”) is successful, at least a portion of the A1 capacity will be put into use, and using the existing Gen 4 equipment, the lines would be able to produce the 9.7” mini-LED modules that Samsung Electronics assembles into larger mini-LED backlights. 
Given that micro-LED TFT circuitry is more complex that of standard backlights, Samsung Display will have to change the process for TFT production, including increasing the number of mask steps, but by using SDC Samsung Electronics is better able to tailor the supply of mini-LED modules to its own sales targets and production.  Originally Samsung had been negotiation with Taiwan based AU Optronics (AUOTY), but decided that it could not have the same level of control under an agreement with AUO.  According to Korean press, at the time Samsung Electronics made the request to SDC (the SDC team was formed in April), Samsung Electronics also informed SDC that it was becoming more open to SDC’s QD/OLED project, which had been seen as questionable in Samsung’s eyes earlier.  This likely has encouraged SDC to work toward Samsung’s mini-LED TFT production request as it gives SDC some encouragement that parent Samsung will be a major customer of the new technology.  Funny how that works…

While Mini-LEDs are just beginning to establish a foothold in the display space, Samsung Electronics (005930.KS) seems to be pushing ahead with its micro-LED plans, but before we go further, we want to make sure readers understand that the two LED types mentioned thus far, mini-LEDs and micro-LEDs are very different, despite the fact that both based on light-emitting diodes (LED).  Mini-LEDs are small LEDs that are used as a backlight for LCD displays.  As liquid crystal material acts as a ‘gate’, allowing light to pass or not pass to the color filter, without a backlight, LCD displays cannot operate.  Mini-LEDs are smaller than typical backlights and provide more regional control over where the light is coming from, but they only operate in the sphere of LCD displays.
Micro-LEDs, while they are even smaller versions of typical LEDs, do not operate as part of an LCD display.  In fact they are more like OLEDs in that they are self-emitting, producing light that you see directly.  They are not ‘gated’ but turn on and off individually.  Further, while mini-LEDs might number in the thousands in an LCD display, the number of micro-LEDs is in the millions, with three for each pixel.  That comes to 24.75m micro-LEDs for a 4K display and 99.53m LEDs for an 8K display, and logic says that as the display gets smaller, all of those small LEDs are packed more closely, so producing a micro-LED TV gets ‘easier’ as the display gets larger. 
We note also that in mini-LED backlights while there are a few thousand LEDs, they are usually configured in ‘strings’, with a ‘string’ being a number of LEDs that operate as one, meaning they go on and off together creating a ‘zone’.  Each zone needs driver circuitry to operate, so the complexity of the backlight circuitry is dependent on the number of zones rather than the number of LEDs.  In micro-LED displays, each sub-pixel has its own control circuitry, so the TFT (thin-film transistor) complexity is far greater for micro-LEDs (many millions vs. a few hundred) than for mini-LEDs, vastly increasing the chance for bad components, increasing power consumption, and increasing heat, all of which are issues facing micro-LED display development, and we didn’t even mention the elephant in the room.
The elephant in the room here is logistics.  Moving hundreds of LED dies from a wafer to a substrate for mini-LED backlights presents some issues and tools have been developed to handle such a task, such as the Pixalux™ from Kulicke & Soffa (KLIC), but going from moving a few thousand small LEDs to moving many millions of LEDs that can be as small as 20um (0.000787”) is a completely different ballgame.  As an aside, assuming a 0.99999% transfer success rate for a 4K micro-LED screen still presents 248 bad micro-LEDs on average, each of which has to be removed and replaced, aside from testing each micro-LED for quality.
So who would spend billions of dollars to develop such a difficult display technology?  Almost everyone, because LCD technology has some very limiting restrictions and while LCD displays will be around for many more years, self-emitting light sources are the ‘way of the future’.  OLED displays (self-emitting) have become an integral part of the display industry and display manufacturers are always looking for technologies that will allow them to reduce the complexity of display production, and self-emitting materials eliminate the need for a backlight and in some cases the need for a color filter.  While the challenges facing micro-LED production seem quite daunting, the challenges facing OLED display production were also formidable but have been overcome and eventually micro-LED displays will become a part of the display infrastructure.
Of course timing is everything in the CE space, so there are a few companies that have been pushing ahead in the development of micro-LED technology, led, not surprisingly by Samsung Electronics, who has the capabilities to produce both semiconductors and displays, and is expanding its plans for micro-LED commercialization.  Samsung originally came out with a modular product called ‘The Wall’, which is a relatively low (960 x 540) resolution RGB Micro-LED system that has a pixel pitch (distance between LEDs, center to center) of 0.84mm and has no visible border around each module or cabinet, as a configuration of 12 modules is called.  This allows the cabinets to be stacked in almost any configuration, but we note this is a big system, with the cabinets about 3” deep and generating the equivalent heat to ~six 60W incandescent light bulbs, so the potential customer tends to be a commercial venue .

Samsung promotes a number of ‘typical’ configurations for ‘The Wall’, ranging from 110” (2.4m x 1.4m) with 2K resolution, to an astounding 583” (12.9m x 7.3m) 8K system, but even given its relatively low resolution it is a bit expensive (Samsung does not quote prices) and all systems require installation by a Samsung engineer, but estimates for even the smaller configurations are in the hundreds of thousands of dollars. 

​That said, in late 2019 Samsung decided to move micro-LED technology to the retail level and announced ‘The Wall’ for consumers.  Current models for consumers are thinner and pre-configured to allow for simple installation but are limited to only one size for mass production.  Samsung currently produces a 110” version at its production facility in Vietnam, which sells for ~170mwon, or ~$150,000 but Samsung has plans to expand that facility this quarter and will be developing smaller models between 76” and 99”, that will move the micro-LED product from a limited market to what will likely be the top end of the premium TV market.
In order to develop such products for mass production, Samsung has been building a supply chain specific to micro-LEDs and will be using China’s Sanan Optoelectronics (600703.CH) as a micro-LED supplier for the 110” and 99” versions, but will shift to PlayNitride (pvt) for the 88” and 76” micro-LEDs. In 2018 Samsung signed a long-term micro-LED development agreement with Sanan, who recently completed a $1.8b mini/micro-LED production plant in Hubei and is an investor in PlayNitride.  Sanan has also been a mini-LED supplier to both Apple (AAPL) and TCL (000100.CH).  AU Optronics (AUOTY) is said to be the supplier of the control circuitry for the line, and has been working with Apple to develop its own micro-LED technology, although with LED supplier Ennostar (3714.TT) primarily.
While we expect Samsung’s micro-LED TVs will remain in the rarified atmosphere of TVs that cost five figures for some time, Samsung’s commitment to the technology is telling, and reflects, to a degree, the general Samsung philosophy that LCD technology is heading into its twilight.  Samsung affiliate Samsung Display (pvt) has been reducing its exposure to LCD large panel production, and if it were not for the massive price increases seen over the last year in large panel LCD prices, would likely have been out of that business at the end of last year, leaving parent Samsung Electronics to procure large panel LCD displays on the open market and focusing itself further on developing small panel OLED and a new large panel OLED/QD technology. 
In the case of micro-LEDs, Samsung Electronics can choose to directly build or control as much of this relatively new supply chain as it wants, and with a potentially very large customer in the wings (Apple), they seem willing to step up their commitment to at least this early stage of the technology.  There are many hurdles that must be met to make micro-LEDs a viable commercial display technology and there is no guarantee that such efforts will be successful, but Samsung has made such ‘blind’ commitments before that have proved ultimately successful and it looks like they are heading in the same direction now.  Whether this will ultimately wind up generating a new technology and supply chain remains to be seen, but having Samsung pushing the technology and commercialization envelope is certainly a plus, and having another CE behemoth watching those moves, gives those in the new supply chain a bit of encouragement.