[Dry goods] LED package device chip junction temperature test (on)

LED Semiconductor Lighting Network At present, LED is fully recognized by the industry for its excellent electro-optical conversion efficiency and light efficiency. But everyone knows that in addition to its own function - LED, there is an important issue that can not be ignored, that is, the heat of LED. After more than a decade of development, although the electro-optical conversion efficiency of LEDs has reached 40% to 60%, there is still a lot of energy that is emitted through the form of heat. Although people are wishful thinking that a certain light source can directly convert all of the electric energy into light energy, and want to remove the excess heat, the court can't do it.

The topic of this article is to let everyone know how to perform thermal testing of LEDs.

First of all, from the packaging point of view, the current development mode of LED is in line with the development of electronic packaging. The difference is that the functional electronic chip is replaced by a light-emitting LED chip , and the opaque packaging plastic is replaced by silica gel or Light-transmitting material such as epoxy resin. Therefore, it can be found that the LED heat conduction path inherited by the package is also consistent with most of the chip-type electronic components, and can be simplified as a one-dimensional heat dissipation path from the chip to the substrate.

Figure 1 LED one-dimensional heat conduction path

The heat of LED is mainly derived from its chip. It is believed that the cause of heat generation is from the non-radiative carrier composite, and the second is that the photon generated from the carrier is not effectively emitted. If the chip is only solid crystal on the support, there is no lens or fluorescent glue on it. At this time, the surface temperature of the chip can be observed by an infrared camera. Generally, this case can be obtained by infrared imaging of the surface temperature of the chip. It is important to note that the surface temperature of the illuminating surface cannot be measured with a thermocouple. Although thermocouples are very convenient thermometers, for an LED chip that is both luminescent and hot, the thermocouple will generate considerable error due to the absorption of optical radiation. The closer to the illuminating surface, the temperature error measured by the thermocouple. It will be bigger.

Figure 2 LED chip infrared image

But our question is, if it is a packaged chip, how do we measure the junction temperature of the chip? The above mentioned infrared thermal imager can only measure the temperature of the surface of the object (except for materials with strong infrared transmittance), and can not capture the junction temperature of the chip covered by the lens or fluorescent glue. In fact, because of the particularity of its diode, LED itself can be used as a sensor to characterize temperature. As shown in the standard JESD51-1, the terminal voltage of the diode changes with the temperature of the PN junction, and the terminal voltage and junction temperature are very close to linear changes! Since this is the case, then we can not use voltage to monitor the temperature of the PN junction inside the chip! In fact, JESD51-1 is a description of this feasible and accurate method.

Figure 3 K coefficient test example

Below we will briefly introduce the test methods mentioned in the JESD51-1 standard.

First, let's take the slope of the voltage-temperature curve mentioned above? V/? T is called the K coefficient of this semiconductor chip. Each chip has its own K-factor, which is determined by the chip's own PN junction and is one of its own characteristics. Figure 3 shows an example of a K-factor test.

The K-factor is measured by reading the voltage at different ambient temperatures with a small test current, Isssss (such as 1 mA), and we can assume that the actual temperature of the chip is equal to the ambient temperature at which it is given. Why does the test current Isssss use a small current? Because the current is large, the current will cause the chip to heat up. After the chip heats up, its actual temperature is quite different from the ambient temperature. The ambient temperature is a parameter that we can control. Therefore, our test current Isssss uses a small current.

After getting this K coefficient, can you know the junction temperature? Since there is a correspondence between voltage and temperature, then we can not read the normal working voltage of the LED to push the junction temperature of the chip? However, things are not that simple (unless your home's LED is working at a small current like 1mA). For a single-chip LED, the normal working current Idddds has now reached several hundred milliamperes, or even more than 1A. What we want to know is the junction temperature of the LED at this operating current. What should I do? Since there is a K factor, can we test it with a jump method?

For example, the LED is lit for a period of time under the operating current Idddds. After the LED is thermally balanced, the operating current is suddenly jumped to the test current Isssss, as long as the current jumps fast enough, as long as the speed of the collected voltage can keep up, Moreover, the test accuracy is still on the lever, then we can measure the voltage V1 at the moment after the jump into the test current. In order to use the K factor, we let the LED work under the test current until it is in thermal equilibrium with the ambient temperature. At this time, a voltage value V2 is measured. Since these two voltage values ​​are measured by Isssss, then the two voltages can be used to correspond to the temperature with the K factor. What is the value of the two voltage values? Dividing V by the value of K gives a change in temperature? T.

Figure 4 Variations in voltage and temperature of current transients

Smart readers can find this temperature change value? T is actually the difference between ambient temperature and junction temperature, so just need to put it? T plus the ambient temperature to get the junction temperature of the chip. Because the acquisition time is short enough, it can be considered that the heat has no time to dissipate immediately. The measured temperature corresponding to the first voltage value can be regarded as the corresponding temperature value of the LED under the operating current. This method is introduced in the JESD51-1 standard. There is a strict proof in it. The author simply describes this test method here.

The question is, since there is such an advanced test method, is there equipment that can meet the requirements of this high-speed switch and high precision? Of course, the standard setters in the JEDEC committee have specifically designed a test device that fully complies with this standard, that is, the industry-renowned T3Ste (r Thermal Transient Tester), whether it is a diode, a triode, or a FET. Or IGBTs, the junction temperature and thermal resistance of these semiconductor devices can be measured with the T3ster. Moreover, after the mathematical operation, the thermal resistance of each layer structure on the heat transfer path can be analyzed to find the heat dissipation bottleneck.