218
10 Reliability Analysis of Group III Nitride LEDs Devices
10.3 LED System Reliability
With the rapid development of semiconductor lighting technology and the increasing
number of semiconductor lighting products, the reliability of LED products has
gradually become one of the most important technical bottlenecks that hinder the
development of semiconductor lighting industry and the application of large-scale
markets. The semiconductor lighting is expected to achieve its inherent high efficiency and high reliability. However, LED lighting systems not only include devices,
but involve multiple links from chips, devices, light source modules, drivers, various
assembly interface processes to lighting systems. In this section, the LED system reliability framework was introduced. The framework includes failure trees, Bayesian
networks, and Markov chains. These models are used to predict the life of semiconductor lighting systems, while showing the failure modes of key components. The
failure modes of key components were showed at the same time.
10.3.1 LED System Reliability
The lifetime of lamp-level products not only depends on the LED, but also the
reliability of other parts inside the lamp. Thus, when the reliability of a lamp was
analyzed, the lifetime and reliability of not only the LED but also the remaining
components should be considered [23]. Here, a lamp-level LED product is divided
into the following major components: LED light source (including LED chips, packages, bonding wires), LED power drivers, optical materials (including phosphors,
paste caps, etc.), machinery, and heat dissipation modules.
The failure modes of the lighting products mainly include two types: the sudden
failure and the optical degradation. It is generally considered that the time, at which
the product suffers the sudden failure or the luminous flux of the product reduces to
70% or 50%, is the value of the lifetime of the product.
The relationship between the main components and the system reliability is shown
in Fig. 10.6, where more “★” means the stronger interaction.
It can be seen that there are many factors involved in the reliability of LED
system products, which made the failure mechanism complicated due to the interplay
among these factors. Therefore, methods such as failure tree, Bayesian network and
Markov chain need to be introduced to analyze prediction models of LED system
products. The concepts of failure trees, Bayesian networks, and Markov chains will
be introduced next.
Failure tree (FT) is a system analysis method. The maximum number of trusted
methods can be found according to the environment and operation by using FT. The
system will show an unexpected state due to the influence of these methods.
As shown in Fig. 10.7, the system reliability consisting of three partial failures
(A1, A2, and A3) is considered. From the characteristics of the system, the system
10 Reliability Analysis of Group III Nitride LEDs Devices
10.3 LED System Reliability
With the rapid development of semiconductor lighting technology and the increasing
number of semiconductor lighting products, the reliability of LED products has
gradually become one of the most important technical bottlenecks that hinder the
development of semiconductor lighting industry and the application of large-scale
markets. The semiconductor lighting is expected to achieve its inherent high efficiency and high reliability. However, LED lighting systems not only include devices,
but involve multiple links from chips, devices, light source modules, drivers, various
assembly interface processes to lighting systems. In this section, the LED system reliability framework was introduced. The framework includes failure trees, Bayesian
networks, and Markov chains. These models are used to predict the life of semiconductor lighting systems, while showing the failure modes of key components. The
failure modes of key components were showed at the same time.
10.3.1 LED System Reliability
The lifetime of lamp-level products not only depends on the LED, but also the
reliability of other parts inside the lamp. Thus, when the reliability of a lamp was
analyzed, the lifetime and reliability of not only the LED but also the remaining
components should be considered [23]. Here, a lamp-level LED product is divided
into the following major components: LED light source (including LED chips, packages, bonding wires), LED power drivers, optical materials (including phosphors,
paste caps, etc.), machinery, and heat dissipation modules.
The failure modes of the lighting products mainly include two types: the sudden
failure and the optical degradation. It is generally considered that the time, at which
the product suffers the sudden failure or the luminous flux of the product reduces to
70% or 50%, is the value of the lifetime of the product.
The relationship between the main components and the system reliability is shown
in Fig. 10.6, where more “★” means the stronger interaction.
It can be seen that there are many factors involved in the reliability of LED
system products, which made the failure mechanism complicated due to the interplay
among these factors. Therefore, methods such as failure tree, Bayesian network and
Markov chain need to be introduced to analyze prediction models of LED system
products. The concepts of failure trees, Bayesian networks, and Markov chains will
be introduced next.
Failure tree (FT) is a system analysis method. The maximum number of trusted
methods can be found according to the environment and operation by using FT. The
system will show an unexpected state due to the influence of these methods.
As shown in Fig. 10.7, the system reliability consisting of three partial failures
(A1, A2, and A3) is considered. From the characteristics of the system, the system