Share: Exploring the causes of white light LED degradation

In recent years, countries around the world have become more and more concerned about energy issues such as environmental protection and energy conservation, which has indirectly affected investment and technology development in these fields. Among them, solar cells, lithium-ion batteries, SiC power transistors , and white light LEDs have attracted the most attention. It is generally believed that the above plans are expected to become high-growth fields in the future with the support of national scale.

White LEDs have expanded from mobile phones and LCD TV backlight modules to general lighting applications such as medical, automotive, and plant cultivation. Foreign companies have even launched affordable 60W white LED bulbs. This new generation of lighting sources using multiple white LEDs is rapidly replacing traditional fluorescent lamps and incandescent bulbs.

Regarding LCD TV backlight modules or large lighting, the use of a large number of white light LEDs must balance cost and performance. Japanese industry players generally believe that the target of 0.5 yen/lm and 200lm/W can be achieved in 2011, and the improvement of chip performance, phosphors, and packaging technology have always played a key role. Reliability is another important issue for white light LEDs, which includes the durability of single LEDs and the brightness distribution when multiple white light LEDs are lit at the same time. To overcome these problems, domestic and foreign manufacturers have actively launched technology development.

Regarding the durability of white light LEDs, i.e., the degradation of LEDs, it is generally believed that the degradation of the light beam, packaging, and chips over time is the main reason for the reduction in lifespan. However, in reality, these degradation factors are complicated, so it is very difficult to analyze the degradation pattern, especially since the lifespan of white light LEDs is very long and it is not easy to conduct degradation tests. Traditional degradation tests include: current acceleration test, temperature acceleration test, accelerated weathering test, etc. Next, this article will introduce the results of the "overvoltage degradation test" and the analysis results of white light LED degradation.

Analysis methods and evaluation items

Figure 1 shows the basic structure and degradation factors of typical white light LEDs for lighting; Table 1 shows the main evaluation items and analysis methods of GaN LEDs and related materials. Transmission electron microscope (TEM) can directly observe the displacement and defects based on the cross-sectional structure of the LED. The evaluation of the deflection, stress, composition, carrier concentration, and defects in fine parts is very important during degradation analysis, especially for the evaluation and analysis of carrier concentration and defects at the nanoscale. Generally, the following methods are used: Scanning probe microscope (SPM ) , Scanning spread resistance microscope (SSRM), Scanning capacitance microscope (SCM), and cathode luminescence (CL).

Regarding the evaluation of resin and phosphor structure, it is generally believed that the use of Fourier Transform Infrared Spectrometer (FT-IR), Solid-State Nuclear Magnetic Resonance (Solid-State NMR), and Raman spectroscopy can achieve the expected results.

Evaluation of chip degradation

Regarding the problem of GaN series components, since its defect density is 5 digits higher than that of GaAs series, and the defect and displacement problems are very serious, it is generally believed that the defects and displacement of LED chips have a direct and significant impact on the degradation, durability and other characteristics of LEDs. The traditional GaN single crystal film grown on a sapphire substrate has a strong compressive stress on the GaN film layer due to the great difference in the lattice constants between the sapphire substrate and GaN, which is also the main reason for the formation of defects and displacement. Recently, most of the industry has switched to using SiC single crystal wafers with similar lattice constants, or GaN single crystal wafers with the same lattice constants to grow films to produce low-defect, low-displacement high-quality GaN epitaxial.

There are two ways to obtain white light sources, namely, the pseudo-white light method of combining blue light LEDs with yellow phosphors, and the high color rendering white light method. The pseudo-white light method mainly combines blue light LED with yellow phosphor to form a pseudo-white light LED. Once the blue light generated by the blue light LED chip is absorbed by the yellow phosphor, the phosphor will produce yellow light, which is then mixed with the blue light that is not absorbed by the yellow phosphor to form the so-called pseudo-white light. The emission spectrum of this white light LED has two peaks, white light and blue light.

The high color rendering white light method mainly combines blue light LED with green and red phosphors to form a high color rendering white light LED. Once the blue light generated by the blue light LED is absorbed by the phosphor, the green phosphor will produce green light, and the red phosphor will produce red light. The green light is then mixed with the red light and the blue light not absorbed by the phosphor to form pseudo-white light. The emission spectrum of this white light LED has peaks in the red, blue and green fields, and the color reproducibility is also better than the above-mentioned pseudo-white light method.

The cross-sectional structure of a typical blue LED used in a pseudo-white light mode is shown in Figure 2. The light-emitting layer is composed of GaN-based compound semiconductor quantum wells with a film thickness of less than 100nm. When emitting light, defects and displacements will be formed, which is also one of the reasons for LED degradation.

Figure 3 is an example of the in-plane CL intensity distribution when a GaN single crystal film is made on a sapphire substrate. It can be seen from the figure that the luminescence between the GaN energy gap and the "yellow defect" luminescence light that causes defects can be found near 360nm and 560nm respectively. Figure 3 (a) is an image of a GaN single crystal film observed using a planar scanning electron microscope (SEM); Figure 3 (b) is the intensity distribution of light near 360nm; Figure 3 (c) is the CL spectrum distribution characteristics of the strong and weak parts of the luminescence line. Figure 3 (b) is a dark band with reduced luminescence intensity . In particular, near 360nm, the luminescence intensity between the energy gaps will decrease. At this time, if compared with the luminescence between the energy gaps, the luminescence intensity of the yellow defect near 560nm will become stronger.

The above results confirm that the crystallinity decreases in the bright part of the black spot, resulting in an increase in the probability of non-radiative migration and a significant decrease in the luminescence intensity at the bandgap end.

Figure 4 shows the acceleration voltage dependency of the CL intensity distribution when testing from the cross-sectional direction . In the figure, a light and dark pattern running through the film thickness can be observed. This shows that the pattern is clear when the voltage is reduced, and a high spatial resolution intensity distribution can be obtained.

The pattern of light and dark lines of CL intensity throughout the film thickness is consistent with the through-translocation period observed by the interlaced electron microscope (TEM) in Figure 5. Conversely, the 300nm periodic line pattern observed in the single crystal surface in Figure 3 reflects this through-translocation period, which confirms that the use of cathode luminescence (CL) can clearly observe the distribution of defects and translocations at the nanometer level.

Figure 6 shows the result of using a scanning diffusion impedance microscope to test the cross section of the blue LED in Figure 2 when voltage is applied to it. The scanning diffusion impedance microscope is based on a contact atomic force microscope (AFM) and is composed of a conductive probe and a large-scale amplification circuit. The scanning diffusion impedance microscope uses the feedback of the atomic force microscope in the contact sample surface mode to enhance the scanning analysis technique of the rotary arm probe touch pressure (increasing the load). Since it uses a highly conductive probe to detect the microcurrent formed at the contact position when the bias voltage is applied to the sample, it can accurately grasp the local impedance distribution on the sample surface.

According to the test results of the scanning diffusion impedance microscope in Figure 6, it is confirmed that the high and low impedance areas of the V-shaped pits in the p-type clad layer of the degraded LED have an increasing trend. Since the V-shaped pits are characteristic defects found in the InGaN quantum well structure, they are also called "V-shaped defects". From the comparison of Figures 6(a) and (b), it can be seen that the V-shaped defects will increase when overvoltage is applied.

Figure 7 shows the results of using the cathode luminescence method (CL) to test the silicon-doped GaN film on the sapphire substrate. The cathode luminescence method mainly observes the quantum well (hereinafter referred to as the active layer) and the light with wavelengths around 463nm and 360nm caused by the buffer layer between the sapphire substrate and the clad layer. The intensity distribution of the luminous light caused by the 463nm active layer is shown in Figure 7(a) and (b). Figures 7(a) and (b) are also the CL intensity distribution characteristics of the unpowered and degraded components; Figure 7(c) is the CL spectrum characteristics of the unpowered and degraded components.

According to Figures 7(a) and (b), the dark bands with reduced intensity tend to increase in degraded components. In other words, as the voltage is applied, the number of dark bands, translocations and V-shaped defects increases significantly. The reduced crystallinity increases the probability of non-radiative migration, which ultimately leads to reduced intensity. If you look closely at the spectrum in Figure 7(c), the CL spectrum of the severely degraded component shows that the luminescent light generated by the 463nm active layer is almost completely absent. In addition, the researchers also developed "cathodoluminescence" and "Raman spectrometer" using near-field light in response to the "Next-generation cathode luminescence (CL) and Raman spectrometer using near-field" plan proposed by the New Energy and Industrial Technology Development Organization (NEDO) of Japan.

The project also applied the ultraviolet resonant Raman effect and a specially shaped probe to develop an ultraviolet laser-excited near-field resonant Raman spectrometer, which successfully evaluated the stress of silicon with a spatial resolution of less than 100nm for the first time in the world. Currently, researchers are reviewing its application in the evaluation of compound semiconductors. Regarding the quantum well structure of InGaN, the use of the newly developed cathode luminescence spectrometer can surpass the spatial resolution limit of 100nm of the traditional cathode luminescence method and detect the composition changes around the "V-de fec t" inside the quantum well structure of InGaN with a spatial resolution of 10nm.

The main purpose of the above plan is to improve the resolution of the cathode luminescence method. In order to reduce the diameter of the electron beam, the researchers used a new Schottky emission (SE; Schottky Emission ) electron gun to create a high-resolution scanning electron microscope (SE-SEM; High-Resolution Schottky Emission-Scanning Electron Microscope).

The spectroscopic system combines an elliptical mirror and an optical fiber to perform cathode luminescence spectrum detection while scanning the electron beam. It uses a new spectroscopic system that is different from the traditional spectroscopic system. The new spectroscopic system uses an ultra-small parabolic mirror with a thickness of 6mm and a focal length of 2mm to drive the piezo stage to obtain the cathode luminescence spectrum using a non-scanning electron beam method. In other words, the cathode luminescence released by the sample is collected by the parabolic mirror and then detected by the detector. Since it only detects the cathode luminescence released at one point, the resolution of the above-mentioned newly developed equipment is greatly improved compared to the spectroscopic system combining an elliptical mirror and an optical fiber.

Through the use of elliptical mirrors, high-resolution scanning electron microscopes can detect Raman spectra and photoluminescence in addition to cathode luminescence. Figure 8 is a structural diagram of the near-field spectroscopic system in the newly developed high-resolution scanning electron microscope. Then the researchers used the newly developed cathode luminescence spectroscopic system to detect the cathode luminescence spectrum of the InGaN single quantum well structure (SQW; Single Quantum Well) film layer made on GaN 2μm/sapphire, that is, In0.02Ga0.98N7nm/In0.20Ga0.80N3nm/In0.20Ga0.98N7nm. It is worth mentioning that the above film layer is a common component of typical blue light LEDs and the single quantum well structure film layer of InGaN has a V-defect that has a significant impact on the yield and durability of the component.

Figure 9(a) is a high-magnification scanning electron microscope image of V-defect; Figure 9(b) is the detection result of the cathode luminescence line spectrum distribution near V-defect during 5nm step detection. The cathode luminescence line spectrum distribution detection in Figure 9(b) is mainly carried out along the scanning electron microscope image line AB in Figure 9(a). The luminescence lines with wavelengths of 364nm and 448nm observed in Figure 9(a) are classified as the luminescence between the GaN and InGaN quantum well layers of each buffer layer. In addition, the broad luminescence lines observed near 560nm are mainly caused by the GaN defects in the buffer layer of the yellow defect. Since the luminescence line with a wavelength of 400nm can be observed on the slope of the V-defect, the researchers believe that this luminescence line reflects the change in the In component of the InGaN single quantum well structure on the slope of the V-defect.

Figure 10 shows the evaluation results of the peak wavelength, intensity, and half-value width of the luminescent line generated by the InGaN single quantum well structure film layer near 448nm. As shown in Figure 10(a), the closer to the bottom of the V-defect, the shorter the peak wavelength moves, and the less the In component. Conversely, at the bottom of the V-defect, the peak wavelength moves to the longer wavelength, which means that the In component at the bottom of the V-defect is very rich.

FIG11 is an evaluation result of the peak wavelength, intensity, and half-value width of the luminescence line generated by the InGaN single quantum well structure film layer near 400nm. FIG11(a) shows that the slope intensity of the V-defect becomes stronger, and the deterioration of the composition is mainly concentrated on the slope of the V-defect.

Figure 12 shows the evaluation results of the peak wavelength, intensity, and half-value width of the luminous line of the GaN buffer layer (2μm). From Figure 12(a), it can be seen that the peak position first moves to the short wavelength end, and the closer to the bottom of the V-defect, the more it moves to the long wavelength end. The wavelength moves to the long wavelength end mainly due to the diffusion of In into the GaN layer. As for the movement to the short wavelength end, it cannot be explained by the change of In composition. In particular, the change of the peak wavelength near the V-defect is mainly caused by the local oxidation of silicon inside the silicon component, which is similar to the stress change.

FIG13 approximates the V-defect in a space of w. Then, based on the structure of the InGaN film stacked on the GaN film layer at intervals of w, the stress generated at the edge of the InGaN film is calculated. The result is shown in FIG13(b). The black diamonds in the figure are measured data, and the white diamonds are calculated data. It can be seen from the figure that the measured results are very consistent with the calculated results.

Based on the above data, the researchers proposed the following occurrence modes for the formation mechanism of V-defect, which are: (1) In order to relieve the stress at the interface between the InGaN single quantum well structure film layer and the GaN stacking layer, In will diffuse into the GaN film layer to stabilize the through-translocation. (2) When the InGaN single quantum well structure film layer continues to grow, in order to bury the In quantity deficiency inside the through-translocation InGaN single quantum well structure film layer, a compositional degradation layer of the InGaN single quantum well structure film layer will appear. (3) The InGaN single quantum well structure film layer continues to grow. In order to ensure the In content of the compositional degradation layer inside the InGaN single quantum well structure film layer, the InGaN single quantum well structure film layer with In quantity deficiency must continue to grow. Therefore, the thickness of the compositional degradation layer of the InGaN single quantum well structure film layer will increase, resulting in an increase in the intensity of the luminous line generated by the compositional degradation layer near 400nm, and finally forming the so-called "V-defect". In other words, the degradation of LED chips is mainly caused by through-translocation and the proliferation of V-defect.

Conclusion

The above introduces the analysis results of the overvoltage degradation test. The main reason for the degradation of white light LED chips is the proliferation of defects. In addition, the degradation of resin and phosphor must also be taken into consideration. At present, foreign companies are conducting temperature acceleration tests on white light LED bulbs, which are divided into two categories: light degradation and thermal degradation. The degradation mechanism of phosphors is analyzed in detail. It is generally believed that with the mastery of the degradation mechanism, it will be positively helpful to extend the life of white light LEDs in the future.