DTF Taiwan LED Supply Chain Technology Forum: Testing standards and implementation of production tests
DIGITIMES staff, Taipei [Thursday 21 October 2010]
Testing is of great importance as most problems occurring during the manufacturing process are related to ineffective testing before production. Jeff Lee, director of Chroma ATE's Integrated System Solution BU, in his analysis of the LED's light, electrical and heat characteristics, noted the key points of testing and introduced some testing methods adopted by the industry.
The interaction of light, electrical and heat characteristics
Lee began by analyzing the LED's light, electrical and heat characteristics that are interrelated to one another. The characteristics of light include light pattern, light quality and light intensity, and spectrum is the characteristic of light quality. The characteristics of light intensity include luminous flux, luminous intensity, radiation flux and radiation intensity. The electrical characteristics include AC, DC, silicon controlled rectifier (SCR), and electrostatic discharge (ESD). AC and DC include electrical voltage and current, and ESD involves human body model (HBM) and machine model (MM). The characteristics of heat include thermal resistance, heat capacity and heat conduction. We need to run tests in accordance with all these various characteristics.
Color rendering index (CRI) is the indicator of the quality of color, and incandescent bulbs have the best CRI at 100 among all light sources. The LED's CRI is not as high in conventional CRI calculation method but the difference is minor for human eyes.
Snell's law, a theory that refraction or total reflection occurs when light goes through different mediums, is another consideration in light testing. Measures have to be taken to avoid errors in testing, as in reality a luminous body would be affected by substances nearby.
Different chip structures also matter. Chip structures are often upgraded in order to elevate light extraction efficiency, such as truncated inverted pyramid (TIP) that increases the luminous efficacy of red light, or surface roughening and so on.
The space between chips also influences measurement. Testing of a multi-chip design would be different from a single-chip one, and the smaller the space between chips is, the higher the luminous intensity will be. The LED's light profile also matters, as the luminous intensity varies with different light profiles even though the luminous flux is the same. Therefore, re-calibration will be needed when testing emphasizes a particular factor. Light uniformity is another problem that needs to be taken care of in testing, because even if there is only one single-chip, the phosphor-excited light still may have uniformity problems and with more chips the differences will be bigger.
In terms of LED's electrical property, if uneven curves occur when analyzing LED I-V characteristic due to thyristor effect, forward bias has to be raised to a certain level for the conduction of light, and vice versa, the avalanche current should also be measured.
Static can cause damage to LEDs. Sometimes it is due to human contact, and sometimes static discharge from machines sends ESD pulse to the interior of the semiconductor and damages the insulation, which can cause an early disruption of negative voltage. Tests should be done to detect any damage; otherwise LED devices will suffer attenuation as time goes by.
In addition, as the LED changes with different materials due to the band gap, wavelength will be different as the band gap decreases with higher temperature.
Reference for measurement
The standard for luminous intensity measurement is CIE 127:2007. The included angle will be smaller when measured from a distance, and the specimen under testing will be closer to the point source. Different standards should be adopted for different light-emitting areas. It should also be noted that light is the strongest around the optical, instead of mechanical center of the specimen under testing.
The calculation of luminous flux, if done at every angle with a sensor, is very time-consuming and costly, so it can be calculated by using integrating sphere. As the distance and size of partial flux's opening will affect test results, there are strict rules in this regard. In addition, illuminance can be worked out with relative area figures when measuring radiant flux density and irradiance.
The measurement of electrical characteristics is under the influence of thyristor effect that was first used by the US military and applies to all PNPN semiconductors because LED phenomena are similar to thyristor's. The reference standards in use now are: MIL-STD-750E Method 4026.3: Forward recovery voltage and time; and Lumileds' Application Brif AB22 Test Conditions For P3 series TS AlInGaP LED Chips.
ESD requires various tests in accordance with the causes, and MM is used to simulate the electro-static discharge protective ability when the machine touches the components in the fabrication process. The standards include ESDA STM5.2, JEDEC EIA/JESD22-A115 and Q100-003-Rev-C. HBM simulates the components' electro-static discharge protective ability to human bodies, or the human bodies' to the components. CDM simulates the components' protective ability against ground discharge. All these have their own standards.
Jeff Lee, director of Chroma ATE's Integrated System Solution BU
Photo: Digitimes, October 2010