The hottest improved silicon substrate LED address

  • Detail

Improved silicon substrate LED address high solid-state lighting cost

translation for reference

address quoted in this article:

improved silicon substrate light-emitting address high solid-state lighting cost

for today's high brightness, the mainstream technology is gallium nitride (GAN) on sapphire or silicon carbide (SIC) substrates. These materials are very popular because the resulting lamps are bright, efficient, and last for a long time. However, these chips are difficult to manufacture and package into usable devices, multiplied by the cost of using them as the final product of the optical engine. Although prices have plummeted in recent years, lighting is still quite expensive to buy than traditional alternatives. This initial cost is considered to be a major factor in slowing the acceptance of solid-state lighting (SSL)

the manufacturer is a pioneering group that has been trying to reduce the cost of high-power LEDs by replacing sapphire or SiC substrates with silicon (SI). Materials are usually used to make most electronic chips (chips). The key benefit is to supply chips at a very low price and take advantage of the LED manufacturing opportunity of the devalued 8-inch wafer factory. After the merger, these concepts can greatly reduce the price of LED and overcome the opposition of consumers

initially, the technical challenge was limited to the performance of Gan silicon LEDs, which made them unattractive for mainstream lighting. Now, some manufacturers, especially Toshiba, have introduced a new generation of these LEDs, which have greatly improved performance at a very competitive price, making them a viable alternative to conventional devices in many applications

this paper reviews the latest generation of commercial equipment for the development and description of silicon lined 3. The most commonly used is the LED with a surface roughness bottom adapted to the tolerance level

push back the cost of LED

although it has taken many years and millions of dollars of research and development funds, modern LED is a cost-effective alternative to traditional light sources, such as incandescent bulbs, fluorescent tubes, and halogen lamps. When mainstream lighting factors, such as the initial purchase price, energy consumption, and life considerations, determine the cost of ownership

a recent report 1 analyst McKinsey & Company concluded that by 2016 (depending on how fast led prices continue to fall), the return of an LED lamp (due to lower operating costs and longer service life) will offset the higher initial purchase price compared to compact fluorescent lamps (CFL) between 1.7 and 3.9 years old. In 2011, the corresponding calculation produced a figure of about 14 years. (Figure 1)

picture of the payback period of LED bulbs and CFL bulbs

Figure 1: the payback period of LED bulbs and CFL bulbs in the residential market (black lines represent the basic situation, and lighter lines represent faster led price erosion). (provided by McKinsey & Company)

unfortunately, the conclusion reached in the same report is that although the market share of LED in lighting applications is about 45%, by 2015, the price premium of LED lighting products is still high, and the initial purchase price is significant, which hinders decision-makers from taking into account the initial investment in general lighting applications

considering the replacement of LED lamps, such as Philips' 100 W (incandescent lamp), which is equivalent to PAR38 LED bulbs with a retail price of $22 $12, the 100 watt of the same company is equivalent to T2 wriggle CFL and a 100 W equivalent ecosmart $6, this silence may not be surprising when compared with halogen bulbs

A key factor in the cost of LED bulbs is the LED chip itself. Each LED bulb usually includes six, eight, or ten LED chips, each of which is manufactured from a complex wafer manufacturing process made of special materials, and then expensive packaged in an array of three or four step assembly operations. Using cheaper substitutes to replace both materials and manufacturing costs will greatly reduce the initial purchase price of LEDs and encourage more rapid popularization

substitution of silicon

in the electronic revolution, it is based on silicon; A stable, cheap, rich semiconductor can easily grow into crystals, pieces into pieces, and be subjected to CMOS process to put each wafer into thousands of IC. In addition, huge investment has been made, and the foundry has produced such chips in large quantities, reducing the unit cost by only cents

Recently, chip manufacturers have moved to a more efficient chip manufacturing process using 12 inch (300 mm) chips as raw materials, replacing the old 8 inch (200 mm) chips. As a result, there is an excess of 8-inch capacity around the world, some of which can be turned to LED manufacturing, significantly reducing the price of the final product

most contemporary LEDs are combined from Gan, which is suitable for emitting photons in the visible part of the spectrum and is composed of band gaps on sapphire substrates. GaN thin films are grown by a method called epitaxy, in which the active region of LED is established by continuous layer deposition on the substrate. One drawback is the lattice spacing of GaN (the unit distance between individual atoms in the crystal structure) and the mismatch between microscopic defects in the active region on the sapphire substrate. These defects, also known as penetrating dislocations, endanger the brightness and life of the two LEDs

silicon carbide is easy to operate, has Gan that is more closely matched than sapphire, reduces defect density, and improves the crystal structure of efficiency and longevity by at least one, sometimes two, number of amplitudes. (see the article in the technology zone. The improvement of materials and manufacturing improves the luminous efficiency.)

sapphire and silicon carbide are not only expensive to produce, but also difficult to reliably manufacture on a wafer with a diameter greater than 4 inches (100 mm). In addition to being cheaper and easier to use, 8-inch silicon wafers require only slightly longer processing than wafer sizes of 4 inches. The final result is that the factory can produce four times (the surface area of 8-inch wafers is four times that of 4-inch wafers (Figure 2), while cutting the material and processing costs at the same time

2-, 4-, 6-, and 8-inch wafer comparison image

figure 2:2-, 4-, 6-comparison, and 8-inch wafer

however, due to material and financial resources, switching to silicon as a substrate for LED poses a serious technical challenge. Most importantly, the crystal structure of silicon is a worse mismatch of Gan than sapphire. Worse, silicon has a very different coefficient of thermal expansion of GaN. These two factors lead to the incorporation of severe tensile stress into the wafer that leads to microcracks when the wafer is cooled during fabrication. Crack led function is poor if at all. Worse, silicon is a good absorber of photons that should escape and contribute to the brightness of LEDs. In this way, the light extraction of silicon led from early gallium nitride is one-third of similar devices based on sapphire in a quarter (see the article in the technology zone to push silicon substrate into the mainstream of LED lighting?)

some start-ups adhere to their development plans, and at the same time, the brightness of sapphire or silicon carbide led on gallium nitride, which is still backward in Silicon Luminescence on gallium nitride, has good effect and long service life. Today's test-bed equipment does not suffer from the poor performance of early equipment, and can be carried out at a small part of the cost of conventional LED

the second generation silicon

such as conventional LED has been steadily improved (while reducing the number of LEDs required to match the output of a single incandescent lamp or fluorescent bulb). A new market has opened up medium-range led. Midrange chips cannot meet the performance of today's high-end devices, but provide reasonable performance (for example, before providing chips with the highest specification whose brightness and life are equivalent to twoorthree years) at the budgeted price (see the article in the technology zone, medium power LEDs can provide cheaper alternative lighting applications)

The development of the mid-range market has opened up opportunities for silicon chips on gallium nitride. Today's silicon substrate LED performance can easily match the mid-range gallium nitride on sapphire or silicon carbide, while weakening the price of the latter

Toshiba (originally a joint venture with Purui, but later bought shares in joint venture partners) is one of the leading suppliers of silicon LEDs on gallium nitride

Toshiba is understandably shy about how it solves the technical problem of lattice and thermal mismatch between GaN and silicon, and has made some public statements. But before being acquired, Puri also showed that the problem of tensile strain is to use a proprietary solution buffer layer (between GaN and silicon)

more relevant information about Toshiba research is available in scientific papers. An example published in February, 2006 introduces how Toshiba researchers inhibit crack generation, which usually puzzles gallium nitride epitaxial silicon to use cubic silicon carbide as the interlayer. The lattice constant of cubic SiC is about midway between GaN and silicon, helping to alleviate the stress that would otherwise accumulate and cause cracking between the adjacent layers in GaN and silicon

researchers report that a 1 micron layer on the top of a conventional 8-inch silicon wafer on silicon carbide is sufficient to suppress cracking in the active GaN layer. Although slightly more expensive than depositing gallium nitride into bare chips, this process is still much cheaper than manufacturing sapphire or SiC wafers, because it is still based on cheap silicon manufacturing processes

Toshiba also claims that its silicon on gallium nitride process is suitable for the production of large volume emission single LED chips directly from the wafer without going through the traditional LED assembly process (Figure 3). The advantages of this technology are higher cost savings and a single LED may become increasingly popular on chip board (COB) array competition - products, including multiple traditional LEDs pre assembled into a single unit (see the rise of PCB LED modules on chips in the article in the technology zone)

Toshiba's gallium nitride on silicon processing image

Figure 3: Toshiba's gallium nitride on silicon process enables the packaging roll emitter led to be cut directly from the chip without going through a traditional packaging process3. (provided by Toshiba)

existing products

Toshiba has launched a series of silicon on gallium nitride products for the first time. At the end of 2012, tl1f1 1 watt LED provides 112 lumens (112 lumens/watt, at 2.9 volts and 350 ma current) as a cold white (5000 K) device

ten months later, the company announced the revised range (tl1l3 family) - the latest commercial products are available in bulk - it provides 135 lumens (135 lumens/watt, 2.85 V, 350 MA). Then, in early 2015, the company released the tl1l4 family, which claimed to represent a sample volume that jumped 60% from the performance of the previous generation of silicon devices on gallium nitride. The top range is 1 watt, cold white light (5000 K, CRI) 70) products, providing 160 lumens (160 lumens/watt, 2.8V, 350 MA). Other variants can span the temperature range of 2700 to 6500 K. The chip is packed in a 3.5 mm package (Figure 4)

in Si led Toshiba tl1l4 family gallium nitride image

Figure 4: tl1l4 in Si series LED gallium nitride promises 160 lumens from 3.5 to 3.5 mm package

the performance of this tl1l4 family is comparable to that of high-end conventional products, such as XLamp xm-l2 (155 lumens/watt, 2.85 V, 700 MA) of Cree and OSLON square (163 lumens/watt, 3.05 V, 700 MA) of OSLON. Toshiba's products actually provide mid-range equipment that is more competitive than the price, such as the XLamp mx-3s (85 lumens/W, 10.7 V, 115 MA) of Cree and the Luxeon 3535l (121 lumens/watt, 3.05 V, 100 mA) of Philips Lumileds, which have better performance

greater brightness (in reducing the cost of efficacy) to overcome the technical bottleneck of bio based chemical fiber and raw material industrialization; Bio based synthetic fiber needs to break through the industrialized preparation technology of bio based synthetic fiber raw materials. L1l4 products can operate with 1 a forward current or even 1.5 A as long as the junction temperature of the chip is kept below 150 ℃. Toshiba explained that the performance level of tl1l4 series makes it suitable for mainstream lighting applications

Copyright © 2011 JIN SHI