In Part 1 of this report, featured in the October 2010 issue, we explored the emergence of the solid-state lighting (SSL) revolution. We also covered the basics of light-emitting diode (LED) chips and characteristics of their color to help you better understand this up-and-coming technology. Now it’s time to round out our discussion with an examination of how the chip is made, technical details of lumen operation/depreciation, and electronic control of the technology.

Manufacturing details

When it comes to producing the LED chip, a major issue for die makers is the lack of color uniformity due to the difficulty in controlling both the wavelength of the blue LED chips and distribution of the phosphor. Thus, after the white high-brightness LEDs (HBLEDs) are made, each one is tested for its correlated color temperature, light output, and forward voltage. Dies with similar characteristics are placed in a specific bin (or box) — a procedure known as “binning.” Essentially, the full production run of dies is separated into a number of bins, depending on the engineering parameters used.

The standard binning process applies a chart centered on the blackbody curve, using eight bins. A specific bin will contain a range of correlated color temperatures (CCTs), because of the difficulty in producing dies with identical characteristics. Some manufacturers offer bins with narrow ranges while others provide a wider range. Naturally, chips placed in a bin that accepts a wide range of CCTs are easier to obtain and more economical than those placed in a bin with a tighter range.

The most efficient LEDs have a high CCT, often above 5,000K, which is a cold bluish light. However, warm white LEDs have improved considerably and are available with a color rendering index (CRI) of 80 and CCT of 2,600K to 3,500K. Because LEDs produce high saturation colors and the LED source is highly controllable, color may become more important for lighting systems in the future.

Lumen depreciation

Every light source depreciates over time, meaning that light output diminishes as it operates. The lifetime for traditional sources is based on lamp failure, with mean lumen output measured as a percentage of rated lamp life, typically 40% for fluorescent and metal-halide lamps. In fluorescent lamps, the photochemical degradation of the phosphor coating and glass tube — and the accumulation of light-absorbing materials inside the lamp — reduce lumen output.

LEDs don’t burn out; they have a passive end of life, with the light output diminishing very slowly over time. The primary cause of lumen depreciation is heat generated at the LED junction and the system design. Therefore, the definition of “lifetime” for an LED is the number of hours required for the source output to decline to 70% of its initial output, according to LM-80.

Because the LED generates a lot of heat at the junction of the two semiconductor materials and doesn’t emit heat as infrared radiation (IR), this heat must be removed by conduction or convection. Unless this heat is removed, the junction temperature of the die will rise, resulting in permanent damage. Short of that, the continuous elevated operating temperatures can reduce light output, shorten lifetime, and decrease efficiency (lumens/watt). High temperatures can also lead to bond wire breakage/detachment. As a result, manufacturers state the maximum allowable junction temperature for that die. The data sheet of a typical HBLED may specify the die’s maximum junction temperature as 145°C, noting that the temperature should preferably remain below 90°C.

The importance of heat removal is seen in the cross section of a typical HBLED package, consisting of an emitter, a metal-core printed circuit board, and some type of external heat sink, as shown in Fig 1 (click here to see Fig. 1). The emitter holds the die, optics, encapsulant, and heat sink slug, which draws heat away from the die. Then, the LED package is bonded to an external heat sink made of high thermal conductivity material, such as copper or aluminum, to achieve what is called passive cooling. The thermal-conductive metal may be part of the fixture frame, and an LED downlight may contain up to 2 lb of aluminum for heat sink purposes.

Active cooling/ventilation methods include heat pipes or some type of electromechanical device, such as a piezio-electric device. One company offers a cooling system using an electromagnetically coupled diaphragm to pulse high-velocity jets of air through nozzles. As the air is forced out of the nozzle, a vortex is created, which entrains the surrounding air.

Getting the most lumens

HBLED manufacturers are constantly working to increase the luminous efficiency (lumens per watt) of their products, as they move toward the theoretical 200 lumen/watt efficiency. Presently, a limitation to boosting the brightness or intensity of the HBLED is extraction efficiency — the amount of photons that are able to escape from the die instead of remaining trapped inside the die by internal reflection.

To address this problem, at least two companies have designed a die with an array of sub-wavelength microstructures known as photonic lattices. These nanoscale structures greatly increase the LED’s extraction efficiency by coaxing photons to exit in a highly directional manner, similar to a low-loss waveguide.

Another method of generating photons uses semiconducting crystals called quantum dots. These tiny crystals, just a few nanometers (billionth of a meter) in diameter, are a combination of zinc, cadmium, selenium, and sulfur atoms that are applied in a way analogous to making a PCB LED. A thin film of dots of different yet precise sizes, which acts like a coating of phosphor, is placed over the front of a PAR-shaped bulb fitted with several blue LEDs. The blue light excites the dots in the film, which, in turn, emits light of various wavelengths that in combination creates “white” light.

Electronic control

Fluorescent and HID lamps use a ballast to provide a starting voltage and to limit the electric current to the lamp. LEDs also need a current-limiting and voltage-control device, which is called a driver or sometimes a power supply. Most LED drivers use an AC-to-DC conversion system, which usually has losses of 14% to 18%. Typically, the driver converts the supply circuit power to the appropriate DC voltage (2VDC to 4VDC), and the required current, (generally 200mA to 1,000mA).

It’s common to look at the useful life of an LED in terms of L70 (the number of hours until the LED light output falls to 70% of the initial output); however, the life of the driver is equally important. Therefore, many manufacturers try to match the life of the two related components. LED makers usually specify which drivers are compatible with their products. Additionally, the components specified in a driver should be of high quality, since high ambient temperatures can cause capacitors within a circuit to dry out and short.

At least two manufacturers offer LEDs powered directly from an AC source, avoiding the conversion losses of a driver. For this operation, an LED array can consist of two strings of series-connected die — one string is illuminated during the positive half of the AC wave and the other during the negative half cycle.

LEDs are similar to CFLs when it comes to dimming. Their electronics often are incompatible with dimmers made for the incandescent source. Nevertheless, new designs will overcome this deficiency, and it’s possible for integrated circuitry to precisely control the brightness and color of the LED source.

The most economical dimming method is continuous analog current control. A small control voltage is applied to an analog dimming input pin on the driver IC, which, in turn, proportionally controls the HBLED current. This works well over the linear part of the curve but may not be very effective at the lower end where the output is minimal, because the eye’s perception of brightness in non-linear. In addition, HBLEDs usually have undesirable color shifts at very low current levels.

One of the eye’s transfer characteristics — response time — can be useful when applying the second dimming technique called pulse-width modulation (PWM). With PWM, the more expensive method, the chip is switched on for a given period each cycle and then turned off for the remainder of the cycle. By adjusting the relative duration of the pulse and the time between pulses, the total light output of the LED is reduced without any distracting flickering.

A light source will appear to operate continuously when it is modulated at a fast enough rate. This rate, known as the critical fusion frequency (CFF), varies based on intensity as well as physiological and other factors. A 300-Hz rate is usually satisfactory.

Working toward improvement

As mentioned in Part 1, the U.S. Department of Energy (DOE) is advancing HBLED acceptance through a number of programs and partnerships. In 2008, DOE launched an initiative called SSL Quality Advocates (QA), developed jointly with the NGLIA. QA is a voluntary program to ensure accurate reporting — not just in product labeling/packaging and literature, but also in press releases and manufacturers’ data sheets. At the core of the initiative is a lighting facts label that is attached to the fixture carton, providing information necessary to evaluate LED products, as shown in Fig. 2. The label lists five key parameters: lumens, watts, efficacy, CRI, and CCT.

Most immediately, the lighting facts label and other initiatives are focused on the estimated 2.2 million commercial buildings in the United States with antiquated lighting systems (inefficient fixtures, ballasts, lamps, and controls). Additionally, all federal buildings must convert to Energy Star lighting products by the end of 2013 in accordance with the Energy Independence and Security Act of 2007.