Christian Geng, Joe Yang / AIXTRON Taiwan
Charlie Huang /AIXTRON AG, China office
Rainer Beccard /AIXTRON AG, Kackertstr. Germany
In the 20-year period, the compound semiconductor industry has undergone a series of changes, from research and development to technology orientation to application orientation, and eventually developed into market orientation. Twenty years ago, R&D activities told us about possible outcomes and thus driven the development of related technologies; today, the market is defined by a variety of possible applications, and production costs are becoming increasingly critical drivers.
From the perspective of LED applications, we can see the same trend: early LED applications are based on the characteristics of LEDs; today, depending on the application requirements, what characteristics LEDs must have. However, LEDs are competitive only when manufacturing costs can compete with traditional technologies. As a key device for manufacturing LEDs, Metalorganic Chemical Vapor Phase Deposition (MOCVD) must undergo continuous design evolution in order to meet higher and higher requirements.
For example, machine capacity is one of the developments. In 1988, the first commercial MOCVD machine was able to carry a 2-inch wafer and demonstrate the process capability to produce a specific component. In 1992, AIXTRON launched the first multi-chip planar reaction chamber Planetary Reactor for GaAs GaAs and Indium Phosphide InP series materials according to the patents authorized by Philips (this is the registered trademark of Ai Siqiang). ), and in 1996 introduced models for GaN GaN series materials, can accommodate six 2 å‹ wafers. This development has enabled the compound semiconductor industry to mass produce components such as LDs and LEDs, and the output is enough to make the industry thrive. Because Planetary Reactor achieves excellent uniformity across individual wafers and wafers, and the reproducibility of each operation is stable, compound semiconductor component manufacturers are able to meet the new challenges of this mature industry: yield .
As the industry grew, Ai Siqiang developed a larger capacity machine by continuously developing the principle of Close Coupled Showerhead (CCS) by Planetary Reactor Thomas Swan (acquired by Ai Siqiang in 1999). station. In response to the LED process beginning to use 4 å‹ substrate and the microwave industry using 6 å‹ substrate, this new device can also be designed to carry different sized substrates at the same time. However, MOCVD equipment has to meet not only the need to increase capacity, but also to continuously improve the yield. In order to further improve the performance of the equipment, it is necessary to have a deeper understanding of chemical and fluid mechanics. Therefore, Ai Siqiang has established A department specializing in Computational Fluid Dynamics (CFD) to study equipment and processes to simulate the effects of hardware and process parameters.
At present, the LED industry has entered a new stage and has begun to be applied to the automotive and display industries. The price of LED chips has continued to fall, and the importance of manufacturing costs has become increasingly prominent. In view of this, MOCVD equipment must be adapted to this demand, so The Ai Siqiang team began a series of improvement actions and launched new products in 2005:
‧ Integrated Concept Platform (IC), an ergonomic and Semi S2-compatible platform for use in Aussie and Thomas Swan's MOCVD systems with optimized design.
‧ Yield-Plus kit for Planetary Reactor, featuring a novel three-layer gas nozzle that significantly increases epitaxial yield.
‧ Sandwich Technology and Bead Blasting remove components from the chamber for cleaning and increase productivity.
‧ Maneuverable reaction chamber height for CCS systems.
These features can be upgraded on an installed system, and they are also included in the next-generation MOCVD system specification (can be standard or optional) for blue, green and UV LEDs for mass production of AlGaInN series materials :
‧ AIX 2800G4 HT (42x2â€) and ‧ CCS 30x2 "system" Crius
In order to develop these features, a systematic approach must be taken to understand production costs, as explained below:
Cost of Ownership (CoO)
Key factors affecting the production costs of MOCVD systems include:
Depreciation
2. Operating costs
3. Yield In-depth observation of these factors, we understand that they are affected by the following three parameters:
a. Capacity
b. Material deposition efficiency
c. Uniformity and reproducibility
Before delving into these parameters, we will explain the relationship between these parameters and the CoO factors:
1. The equipment depreciation cost per wafer is calculated by dividing the equipment cost (investment) by the total number of wafers produced during the depreciation period. The total number of wafers is proportional to the capacity:
Depreciation - Investment / Capacity
2. Operating costs can be divided into the following two categories:
‧ Plant facilities and manpower ‧ lead gas materials (precursors)
Plant facilities and personnel costs are inversely proportional to the capacity of the system, and the consumption of lead gas materials is inversely proportional to the deposition efficiency.
Operating cost -1 / (capacity * deposition efficiency)
3. Yield should be the most important factor affecting CoO. In each calculation formula, it must be multiplied by the production capacity. If the yield is low, the depreciation and operating costs will increase.
Depreciation - Investment / (Capacity * Yield)
Operating cost -1 / (capacity * yield * deposition efficiency)
Closely related to yield is the uniformity of wafer characteristics, such as film thickness, composition, and impurity concentration, which must be achieved on a wafer, between all epitaxially deposited wafers, and between different epitaxial deposits. Claim. The uniformity problem in the MOCVD process will be explained in more depth later.
The new three-layer gas nozzle design in the yield upgrade kit addresses the three CoO parameters in several ways. It introduces all reactive gases into the horizontal direction in a lateral manner, reducing the need to precisely adjust the gas nozzles. The best uniformity is achieved; the organometallic vapor and hydride are injected from different layers, the uppermost layer is the carrier gas and ammonia gas, which is the biggest difference from the two-layer nozzle. The newly added upper gas can avoid the process gas. The vortex is formed and returned to the nozzle so that the nozzle can be kept clean and free of deposition, and can be used as another parameter for fine-tuning uniformity.
In addition, the nozzle is completely water-cooled, so that the temperature of the nozzle and its vicinity is low, which prevents the reaction gas from reacting prematurely, thereby preventing reactants from being deposited in the nozzle and nearby, and greatly improving the deposition efficiency of the leading gas material. For example, in the GaN epitaxial process, ammonia consumption can be reduced to 25%.
Another valuable feature is the increased deposition efficiency of trimethyl indium (TMIn), which allows the epitaxial temperature of InGaN to be increased by 50 degrees Celsius, thereby improving the material properties of the InGaN epitaxial layer and thereby increasing the brightness of the LED. The computational fluid dynamics model was applied to the development of this nozzle, allowing us to optimize its size to ensure that the gas flow in the chamber is a perfect laminar flow, avoiding the occurrence of eddy currents.
In addition to significantly improved yield and deposition efficiency, lower temperature gas nozzles can significantly reduce thermal stresses in the components of the reaction chamber, which can have a positive impact on component life and, of course, help reduce operating costs.
Increasing the chamber capacity itself can reduce production costs. The capacity of the 42x2†AIX 2800G4 HT is 1.75 times that of the previous generation and 30x2â€. The Crius is 1.58 times that of the previous generation, but will the price of the machine increase proportionally with the capacity? And won't! Because the increase in equipment investment is much lower than the increase in capacity, the depreciation of each wafer is reduced even lower, and the cost of parts and consumables is also reduced.
The plant facilities and labor costs of the main machine are the same as those of the small machine, so the operating costs of the main machine are reduced by 75% and 58% respectively.
Capacity Throughput
In the GaN LED epitaxial process, high temperature baking is usually used during the epitaxial growth to remove deposits on the graphite stage. Although stable process conditions can be maintained, it is time consuming and thus affects the MOCVD machine. The number of wafers that are shipped. For the Planetary Reactor, Ai Siqiang developed a sandwich graphite stage, the surface of which can be easily replaced with a new component, so the LED epitaxial process can continue without the need for high temperature baking steps, the deposited components can be taken The step of cleaning outside the chamber saves the time for high-temperature baking, and the entire production cycle can be reduced by 2 hours, so that the productivity can be increased by about 30%. In addition, the high-temperature baking can be reduced, so that the service life of the graphite stage can be improved. And prolonged, reducing production costs.
Another way to increase capacity is to increase the growth rate of the thick GaN film layer in the LED structure. There are two important points to be aware of to achieve this goal: First, the necessary condition is that the design of the reaction chamber must allow high concentration of lead gas to enter without A chemical reaction will be generated in advance. In the Planetary Reactor, the organometallic gas and the hydride gas enter the reaction chamber separately, and the mixing time is delayed because of the laminar flow. In addition, maintaining the inlet at a low temperature can also reduce the reaction, and secondly, The development of epitaxial processes allows material quality to be maintained at high growth rates, such as lower growth rates and higher pressures to reduce defect density during GaN island growth bonding, then lowering the pressure and increasing the growth rate to 6 mm/h. .
Spending less time to grow the GaN layer allows the production cycle time to be shortened, and depending on the LED structure, it can increase the throughput by up to 25%.
Uniformity
The yield is in fact a convolution of the uniformity distribution and specification. Today's wavelength distribution specifications are much stricter than a few years ago. Take the backlight of a flat panel display as an example: white LEDs use blue LEDs with yellow light. Phosphor, its high fluorescence conversion efficiency only occurs in the narrow blue spectrum region. If it is a backlight composed of red, blue and green light, it needs to use green LED, and the human eye is very sensitive to the color change of the green band. Therefore, the specifications of green LEDs are very strict.
When we talk about uniformity, we must know that the wavelength distribution of the produced LED crystals accumulates the wavelength distribution of all the wafers, and the wavelength distribution on the wafer has a considerable relationship with the temperature distribution of the stage and the MOCVD gas flow. The wavelength distribution between the wafers is determined by the temperature uniformity of the wafer tray and the geometry of the reaction chamber, and the repeatability of each epitaxial growth is closely related to the consistency of the boundary conditions in the cavity.
Mass Production Information Feedback <br> One of the most important tasks for equipment suppliers is to continuously monitor and improve the production value of the equipment. There are several models to achieve this challenge – for example, setting up tests at their own plants. The laboratory or the partner's equipment is used to simulate the mass production status.
ASI's use of a unique model allows us to work very closely with our customers – the Manufacturing Oriented Research Laboratory (MORL), which is developed at the R&D labs at Axon and Thomas Swan headquarters. Test the MOCVD equipment and process, and then work closely with the customer in the second phase to complete the entire research and development work. The second phase of the work is carried out by Ai Siqiang's local experts and customers, complete planning, execution and monitoring. The first phase of R&D focuses on hardware evaluation and basic process validation, while the second phase focuses on component validation, mass production optimization, and fine-tuning of specific component processes. On the one hand, this model can ensure that the development of the process meets the needs of customers. On the other hand, Ai Siqiang can focus all the expertise on the new development of the equipment. The benefit of this model for customers and Aisiqiang is obvious: customers can Accelerate the R&D cycle and optimize mass production technology, and Ai Siqiang can collect first-hand information on MOCVD performance under real mass production conditions to continuously improve the machine. MORL plays a driving and absolutely neutral role in this model.
Today, Ai Siqiang Group's MOCVD system will go all out more than ever to meet the needs of mass production. Our development methods and products are reflecting the mature evolution of the compound semiconductor industry in these 20 years.
For the LED backlighiting keypads, the graphic overlay lighted by the backside LEDs which assembled on the flex or rigid circuit layer. By this way, it is very simple and also economical. But it couldn't get a even lighting effect on surface. For this backlighting method, the LEDs can be assembled under centered buttons also can be assembled beside of the buttons or other areas. For the products need each button be backlighted, usually design the LEDs under the center of the buttons, then use the center holed metal domes and print transparent color on the graphic overlay. By this way the center button will be much brighter than sides of it. But this way is easy to achieve and also its cost is the lowest. This is suitable for those products no very high requirement for the evenness of the backlighting.
LED Backlighting Membrane Switch
LED Backlighting Membrane Switch,Tactile LED Membrane Switch,LED Membrane Switch,LED Backlighting Panel Keypad
Dongguan LuPhi Electronics Technology Co., Ltd. , http://www.luphiprint.com