How to make the LED brighter?

Reducing the "Droop effect" and improving light extraction efficiency can help achieve better and brighter LEDs. In the past 10 years, LEDs have transformed solid-state lighting. LEDs have promoted their applications in general lighting due to their high efficiency and long life. The efficiency of LEDs continues to improve at an alarming rate, not only reducing the number of LEDs for a given application, but also reducing the cost of the hardware system, which in turn increases the adoption rate and reduces the cost. This increase in efficiency makes the high-brightness chip smaller and can generate an addressable matrix from a densely stacked array, which is very suitable for dynamic beam steering of car headlights. The inherent high-speed switching of InGaN LEDs makes them ideal for visible light communication or Li-Fi.

The wide penetration of LEDs into various markets is due to the significant increase in blue LED plug efficiency, white conversion efficiency, precise customization and the ability to control color points. In this article, the Lumileds team from San Jose, California discussed the technical details of LEDs and compared different structures to highlight opportunities for future improvements. In the following sections, we will introduce various aspects: white LED efficiency typical loss (range); high-power LED multilayer stack epitaxy considerations-internal quantum efficiency and attenuation measurement, the relationship between polarity and semi-polarity and non-polar GaN; Carrier diffusion and light extraction device-patterned substrate; wafer structure comparison.

Droop effect

With the increase in LED brightness requirements, the operating current density has increased, extending from the traditional 35 A/cm2 to more than 100 A/cm2. This change has a profound effect on epitaxy, because increasing the internal quantum efficiency at a density of 100 A/cm2 and increasing at a density of 10-20 A/cm2 have a significantly different focus.

At lower current densities, the increase in internal quantum efficiency comes from the increase in material quality, because indirect recombination dominates at low currents.

In sharp contrast, when LED driving is more difficult, the focus must be on the Droop effect. Today, the industry generally supports OJ composite as the main reason for the decline in the efficiency of the most advanced industrial equipment. As the carrier density in the quantum well increases, Ouge loss is significant at high driving currents, which enhances the possibility of the three-particle recombination process.

One option to reduce OJ recombination is to use more wells to introduce the active area, because this may reduce the carrier density of each of them, but the probability of success is not high. The asymmetry of electrons and the effective mass of holes cause the carrier density on the p side of the active region to be higher than the carrier density on the n side, and result in a change in carrier recombination. Therefore, the benefits of adding quantum wells may be small, or even none.

A better method is to use energy band structure engineering. This can promote better carrier distribution and ensure that the carrier density of each quantum well is low. To achieve this, the operating point of the equipment is higher on the efficiency curve.

Although active areas designed for low Droop can usually achieve uniform distribution of carriers in quantum wells, they come at the expense of material quality, and this increases non-radiative indirect recombination. Generally, an increase in indium content in a low Droop active area design will result in a decrease in material quality. Obviously, the best LED must combine resistance to sagging efficiency with high material quality to ensure low indirect recombination (see Figure 2).

Another option for avoiding the Droop effect is to increase the rate of radiation recombination through a larger superposition of electron and hole wave functions. Today's LEDs are produced on the c-plane and are plagued by internal electric fields, which pull away electrons and holes and damage radiation recombination. Improvements can be made by switching to semi-polar and non-polar substrates to reduce or eliminate the polarization-induced electric field in the active area. The advantages are not only limited to higher radiation recombination, but also reduce the Ogee recombination rate under higher driving current.

Realizing all these promises is not easy. Today, semi-polar and non-polar components are limited by a short non-radiative recombination life, and the substrate is very expensive and has no commercial use. In addition, despite advances in these alternative crystal orientations, they are chasing moving targets due to carrier expansion and improvements in material quality.

Improve light extraction efficiency

One way to optimize light extraction in modern high-power LEDs includes reducing the number of pump photon hops, that is, the number of round trips that the pump photons typically make in the wafer cavity before they leave, and cutting the pump absorption in the wafer cavity.

These two key characteristics (the number of pump photon reflections and pump absorption) are significantly different in two common architectures: flip chip and thin film. Thin-film designs provide smaller source sizes, and they are preferred in highly oriented applications, while flip-chip designs connect directly to the board without the use of interposers. The common point of the two is high current density and low thermal resistance, both of which can realize high-density arrays.

In addition to these two designs, there is also a third, which is a variant of flip chip: it redirects photons through the top side of the die by blocking the sides of the sapphire substrate. The advantages of this design include: smaller source size and stricter angular radiation pattern; more effective coupling efficiency; greater design flexibility.

From the perspective of photon jumping, the two flip-chip designs have a strong dependence on the thickness of sapphire and are better than thin-film structures. With a flip-chip structure, the sapphire needs to be thick enough to prevent a large number of photons from jumping-for example, at least 100 mm for a 1 nanometer 2 chip.

The flip-chip structure has two characteristics that can significantly reduce the number of jumps, thereby facilitating light extraction. The first is that due to the high refractive index of sapphire, the refractive index contrast of the GaN escape surface associated with the thin film is reduced. The second is that once light enters the sapphire cavity, it can propagate out through the sidewalls, thereby reducing the scattering to the GaN area. For a typical sapphire thickness, sidewall radiation may account for 30% to 40% of the extraction efficiency.

Generally speaking, the number of photon bounces depends on the angular direction of the photon emission in the active area, and is the most at the angle close to the glancing angle. But the relationship between angle and photon bounce is not simple, because the valley curve appears between 15° and 40°. This feature can be seen in all three LED designs and is related to the complex transmission characteristics of the patterned sapphire surface interface. Note that for higher photon emission angles, the average number of photon jumps suddenly rises, consistent with the critical angle of the GaN-sapphire or GaN-siloxane interface. The side coating of the wafer has a significant effect on the number of photon jumps.

For flip chips without side coatings, in contrast to GaN-silicon resins, the number of reflections increases rapidly at higher angles near the critical angle of GaN sapphire. This is consistent with our understanding, because any internal reflection on the top surface of the sapphire-silicone will have a second chance to escape from the sidewall of the sapphire. The side coating of the flip chip brings about a huge change, which leads to an increase in backscattering into GaN, which is followed by an increase in rebound at a lower angle near the critical angle of GaN silicon.

The extraction efficiency of different types of designs can be explained with the following figure. For flip-chip chips, when the thickness of sapphire reaches a relative thickness of about 0.25, the extraction efficiency can be quickly improved, and then it will level off. Side coating does not improve the extraction efficiency. When a poorly reflective coating is used in combination with a high sapphire thickness, the extraction efficiency may decrease.

In order to obtain full efficiency, a five-sided luminous flip chip is better, because the sapphire cavity can reduce the interaction between the backscattered light and the damaged area of ??the chip. However, the net reflectance gain of the thin film design may only be significant for relatively high sapphire thicknesses. Generally, it must be much higher than 0.1, consistent with the dependence of the number of rebounds.

Our method of improving light extraction is mainly to reduce pump absorption. For flip chip, when circulating pump radiation propagates in the wafer cavity, its attenuation is usually 7% per round trip. On average, 8 photon jumps can increase the extraction efficiency by about 85%.

The biggest cause of this absorption is the GaN-Ag interface. One way to solve this weakness is to switch to a composite structure by inserting a thick enough low-index oxide layer between the metal and the semiconductor. The choice of SiO 2 will prevent the incidence and metallization interaction within the critical cone angle of approximately 40°. According to our simulation test, the reflector loss contribution can be reduced from 50% to only 20%.

The excellent current spreading is also produced by the composite structure, because it is possible to ensure that most of the current injected into the active area is far away from the n-GaN vias. This is particularly advantageous under high drive conditions.

Another measure to increase light coupling by reducing the number of photon jumps is to optimize the scattering characteristics associated with the patterned sapphire substrate. If pure sapphire is used, it will cause adverse effects in two aspects. First, within the angle range of the maximum incident radiation, the light transmittance of the exit surface will decrease. Second, the cancellation of guided modes will be reduced because the light is reflected by the specular surface instead of being diffracted.