The average high-brightness (HB) LED converts only around 45% of the applied energy to visible photons, with the remainder generating heat. If this heat is not adequately dissipated from the LED, it will overheat and potentially cause catastrophic failure. Even if catastrophic failure does not occur, an elevated junction temperature in an LED can cause reduction in light output, changes in color, and/or a significant decrease in life expectancy.
This article shows how to calculate junction temperature and notes the importance of thermal resistance. It examines lower thermal resistance LED packaging alternatives such as chip-scale and chip-on-board (COB) designs, and looks at the factors that affect heatsink performance.
How heat is generated and impacts an LED
When sufficient voltage is properly applied across the P-N junction of an LED, current flows through the junction generating both light and heat. However, the average high-brightness (HB) LED converts only around 45% of the applied energy into light, with the rest generating heat.
Since the P-N junction is small, the heat-generation-rate per unit area is large: a 1 W, 1 mm2 LED can generate as much as 100 W/cm2. As the junction temperature increases, both the forward voltage and the lumen output of the LED decrease. In order to prolong their lifetime and maintain performance, the junction temperature of LEDs must be kept within the manufacturer’s specifications during operation.
As shown in Figure 1, at a constant operating current, the forward voltage decreases by approximately 20 mV for every 10°C rise in junction temperature. More specifically, the forward voltage at a constant current of 350 mA decreases by 0.17 V as the junction temperature of the LED increases from 25˚C to 80˚C.
Figure 1: High LED junction temperatures reduce forward voltage. (Source: Osram)
Similarly, the light output drops by 10% as the junction temperature increases from 25˚C to 80˚C as shown in Figure 2. If the LED produces 90 lumens at 25˚C, it would only produce 81 lumens at a junction temperature of 80˚C. Simply put, at constant operating current, the luminous efficacy decreases by about 1.8% for every 10°C rise in junction temperature.
Figure 2: Higher LED junction temperature decreases light output. (Source: Osram)
The dominant wavelength of an LED is the wavelength of the photons that an LED predominantly emits, which determines the LED’s color. For a monochromatic LED such as the red 626nm LED shown in Figure 3, the dominant wavelength increases with a higher junction temperature, altering the color.
Figure 3: Higher junction temperature shifts the dominant wavelength and thus changes the LED’s color. (Source: Osram)
Calculating junction temperature
The efficiency of solid-state lighting devices depends heavily on the junction temperature, which in turn, primarily depends upon three factors: the applied power; the thermal resistances between the LED junction and the ambient temperature; and the ambient temperature itself. The applied power determines how much heat is generated, while the thermal resistances and ambient conditions dictate how efficiently the heat can be dissipated.
The resistance of two important thermal paths affects junction temperature. The first is the resistance between the LED junction and the thermal contact at the bottom of the package. The second is the resistance from the thermal contact to ambient.
The temperature of the LED junction (TJ) is the sum of the ambient temperature (TA) and the product of the thermal resistance from junction to ambient (Rth j-a in the equation below) and Pd, the power dissipated (If x Vf). Thermal resistance is defined as the rise in temperature of a component per unit of power dissipated in units of °C/W.
The equation is:
Understanding the thermal path of the LED device from junction to ambient is essential when designing a lighting system to ensure maximum thermal performance. To simplify things, here we are just looking at the sum of the resistance between the LED junction and ambient, but in an actual LED illumination system, there would be a number of resistances defining the thermal path of the entire system.
Low thermal resistance allows the LEDs to be driven at higher currents in order to increase luminosity, without excessive risk of early failure due to overheating. The maximum junction temperature and the thermal resistance of an LED should be supplied in the manufacturer’s datasheet.
Packaging can help
The thermal resistance from the LED junction to the thermal contact at the bottom of the package is governed by the package’s design. Recognizing this, engineers have focused on developing more thermally efficient designs such as chip-scale package (CSP) devices and chip-on-board (COB) LEDs.
CSP technology eliminates the traditional sub-mount to directly attach the LED die to the pc board (Figure 4). Until recently, CSPs were not popular for LEDs because of the difficulty of extracting heat from such tiny devices. However, increases in efficacy and a higher tolerance to temperature have addressed that issue.
Figure 4: The many advantages of CSP technology include lower thermal resistance. (Source: Samsung Semiconductor)
There is no standard definition for a CSP, but the industry generally considers a “chip-scale package LED” to be any device that is of equal size, or up to 20 percent larger than the light-emitting area of the LED. CSPs have lower thermal resistance than conventional LEDs, primarily due to the metal-to-metal interface between the CSP LED and the pc-board heatsink surface.
For example, Samsung Semiconductor’s SCP8RT78HPL1R0S06E has a package thermal resistance of just 2°C/W. Samsung’s CSP technology scales down the size of a conventional LED package by combining flip-chip technology with phosphor coating technology, thereby eliminating the need for metal wires and plastic molds.
In the COB approach, manufacturers package a number of die directly onto a substrate. The low thermal resistance of the Bridgelux Vero and V series LED arrays which range from 1.6°C/W to as low as 0.25°C/W, is enabled through an LED die structure in which the thermal and electrical paths are separated.
Attaching an LED to a clean, flat, smooth heatsink is necessary for good thermal transfer. The use of a thermal interface material (TIM) between the LED and the heatsink is also required for good thermal transfer. LED supplier Cree claims the back of its ceramic substrate CX family of LEDs (such as the CXA1304-0000-000C00A427F) is ten times smoother than the back of the aluminum substrate often used on other COB LEDs.
To determine the flatness of a heatsink, Cree suggests using a razorblade as a straight edge to look for any gaps between the razorblade edge and heatsink (Figure 5.)
Figure 5: Checking heatsink flatness. (Source: Cree)
Thermal interface materials and heatsinks
A typical LED illumination system has multiple HB LED packages which are attached to a substrate and mounted to a heatsink. Because LEDs do not emit heat radiation like traditional incandescent bulbs, the heat generated by them must be conducted away through the substrate. Traditional thermal substrates include two types of ceramics: Al2O3 (aluminum oxide or alumina) and AlN (aluminum nitride). During assembly, the bottom surface of the substrate should be in full contact with the mounting surface of a heatsink. A thermal interface material (TIM) is used between the LED and the heatsink to fill in small voids and air gaps to help conduct heat. If there is space between the LED and the heatsink, the thermal path will not be as efficient. TIMs can come in the form of adhesives, greases, gels, pads, solder alloys and epoxies.
The heatsink is the last integral part of the thermal stack. Heatsinks transfer heat away from the LED, helping to keep the junction temperature within acceptable limits. Designers should consider the heatsink's surface, surface area, aerodynamics, thermal transfer, and mounting.
Heatsinks work in three ways: conduction (heat transfer from a solid medium to another solid), convection (heat transfer from a solid to a moving fluid, usually air), or radiation (heat transfer from two bodies at different surface temperatures). Heatsinks are usually made of a metal such as aluminum or copper with numerous fins to increase their surface area (Table 1).
||Thermal Conductivity (W/mK)
|Air (at 0°C)
Table 1: Thermal conductivity of common heatsink materials and air (Source: Bridgelux)
Passive or active cooling methods can be implemented with the heatsink to help cool it. As a general rule of thumb, there should be 10 square inches of heatsink surface area for every watt of power to be dissipated.
The majority of LED failure mechanisms are temperature dependent. Even if an elevated junction temperature in an LED does not result in failure, it can cause light output reduction, changes in color and/or a significant decrease in life expectancy. This article touched on how to calculate junction temperature and noted the importance of thermal resistance. It also discussed lower thermal resistance LED packaging alternatives such as chip-scale and chip-on-board (COB) designs, and examined the factors affecting heatsink performance.