Technical Insight
Lasers light the road ahead
Laser-pumped phosphors create more compact and efficient headlamps that double driver visibility
BY ABDELMALEK HANAFI AND HELMUT ERDL FROM BMW
Take your car out after sunset, and for every mile you go, you are about five times as likely to be involved in an accident as you would be when going for a drive during the day. Why? Well, it is partly that you are tired and are more likely to be hit by a drunk driver − but probably the biggest factor is that you don't get as good a view of the road ahead.
To try and improve visibility, while reducing the number of accidents, headlamp designers are building units that are not only more powerful "“ so that the driver can see further "“ but produce a better emission profile. The aim is to give a clear view of the road ahead, while not dazzling on-coming traffic, pedestrians and wildlife (see Figure 1).
If an emission profile is to meet the requirements for European low-beam lighting, it must generate a visibility of up to about 100 m along the driver's lane. Such a beam must also contain a narrow peak of very high luminous intensity, which is also termed an asymmetric hot-spot. Another feature is the cut-off line, incorporated to avoid blinding the on-coming traffic. In general, the different illumination profiles produced by a headlight must also conform to international regulations, a set of rules that also stipulate the characteristics of dynamic illumination patterns that can help the driver negotiate a winding road and spot pedestrians and animals (see Figure 2).
The light source
There are also requirements for the phosphor, which has a conversion efficiency influenced by composition and temperature. If the phosphor gets too hot "“ this could result from heating by a laser spot size below 100 µm "“ it can quench luminescence.
The capability of the headlight is governed by the light source. Ideally, it should be powerful enough to enable the driver to see a long way ahead, while being easy to house and package. The quality of the beam produced not only depends on the brightness and luminous flux of the source, but also on the likes of its spectral profile.
Figure 1. The illumination produced by the headlamp varies in intensity, so that it allows the driver to see road markings and pick out pedestrians, but avoids causing excessive glare to on-coming traffic and wild life. Taken from J. Damasky, "Lichttechnische Entwicklung von Anforderungen an Kraftfahrzeugscheinwerfer", Ph.D. Dissertation, University of Darmstadt (1995)
For most of the twentieth century, the incandescent bulb has been the only option for the light source in a headlamp. But in the last decades designers have been able to consider halogen and Xenon lamps, and more recently even solid-state sources.
At BMW, we integrated the first xenon-based illumination system into our cars in 1991. Compared to predecessors based on halogen bulbs, these headlamps generated more luminous flux on the road, reduced power consumption and trimmed system dimensions. These headlamps even featured dynamic functions that improved driver visibility by adaptively illuminating selected parts on the road.
Recently, thanks to their compactness, many headlamp manufacturers have started to incorporate white, phosphor-converted LEDs into their designs. The small size and the particular emission profile of these high-power chips promise to concentrate light in certain angular areas, where enhanced visibility is needed. Another virtue of this solid-state source is that it unlocks greater freedom in styling, but these benefits must be weighed against two major weaknesses: Compared to the xenon source, the luminous flux from a chip is an order of magnitude lower, and the luminance is more than 50 percent less.
The luminous intensity of the asymmetric hot spot produced by the headlamp is proportional to the product of: the luminance of high-power white emitters; the aperture of the lighting module; and the efficiency associated with the collection of the flux produced by the source, to that illuminating the road. One way to increase the luminous intensity is to turn to a multi-module package. However, this increases the emitting area or leads to a reduction in the collection efficiency of the light flux. To realize the regulatory requirements of the luminous intensity on the distribution patterns (background and hotspot), multiple packages are needed, but this is detrimental to the geometry of the headlamp, which is relatively heavy and cumbersome.
Figure 2. (a) Bird's-eye-view of several variable illumination patterns corresponding to the dynamic assisting functions (b) Marking lighting as a driver assistance lighting function: using a night-vision system, pedestrians and animals are detected, locate, targeted and automatically illuminated.
LED-matrix systems address a higher variability on the distribution pattern. They feature arrays made from 80-100 individually addressable Lambertian white emitters per light-engine (see Figure 3). These systems take advantage of the instant switch on/off and dimming properties of the LEDs, and they provide a given light distribution pattern by partially overlapping several small light spots in the far-field. Even without the incorporation of additional mechanical actuators, these light engines can combine high-beam and low-beam capabilities with bending and swivelling light, and are capable of producing a glare-free high-beam. However, they do have an Achilles heel: The range of visibility is extremely limited by the limited luminous intensity of the LEDs.
At first glance, it would appear that increasing the number of LEDs in the matrix would address this shortcoming, while enabling the generation of free-patterned new dynamic light distributions, similar to those resulting from a video projector. But the requirements for such a LED-matrix light-engine are rather challenging. In addition to a higher luminous flux of the LED chips they include the need for a far-field resolution of 0.1°, a specification that corresponds to projecting a 5 cm diameter spot on a wall located at 25 m in the far field. To meet these requirements, if aberration-free secondary optics with a 60 mm focal length is used, the dimensions of the LED must be no more than 100 µm by 100 µm. But a typical phosphor-converted white LED is far bigger than this, having sides of 1 mm².
Figure 3. A matrix LED light-engine: up to 100 pixels; 130 lumen per LED (Source: Osram). Note that when this technology is used for making headlamps, each sub-set of chips is equipped with its secondary optics.
Another issue is the vast number of LEDs that are needed "“ for a field of view that is ±30° in the horizontal direction, and ±10° in the vertical, a matrix of 600 by 200 chips is required. This results in the requirement for a high power, active emitting display, which is 5 orders of intensity levels above the available technical solutions, realized by OLEDs. Despite these fundamental limitations, some semiconductor companies, car manufacturers and set-maker suppliers have been promoting this approach. However, they are compromising with the number of LEDs, using a matrix of just 1024 chips.
We question this approach, preferring to evaluate the requirements of vehicle lighting systems for both today and tomorrow, and based on this assessment, consider the options for the lighting source. Modern vehicle lighting systems are set to feature more lighting functions, such as dynamic driving assistance. However, incorporating this may make them bigger and heavier, while they should be heading in the opposite direction, getting smaller and lighter, to aid styling and trim energy consumption. This implies that to make better headlamps, there is a need to move on from the LED.
One of the lessons that can be learned from geometrical optics is that it is possible to reduce the size of the optical system, so long as the light source is scaled down accordingly. The implication is that to make a better headlamp, a far, far brighter light source is needed. We have pursued this goal, directing efforts at creating a high-luminance white-light source that approaches a point source. This is a disruptive approach, because it moves away from trying to increase luminous flux, with the new target being an increase in luminance, or in other words, brightness.
Our approach to creating a very bright point source is to pump a remote, yellow-emitting phosphor with a high-power, multi-mode, blue edge-emitting laser diode. This class of laser is increasing in power, with a 3 W version recently available. The emission surface is just 30 µm by 4 µm, compared to 800 µm by 800 µm for a high-power LED, which equates to a factor of more than 5000. If one also accounts for the emitting cone and the optical power of the light-sources, the luminance of the laser diode is 500,000 times higher than that of the LED.
Figure 4. Droop causes the efficiency of the LED to rapidly decrease at high power densities. Such limitation is not observed in laser diodes. Taken from J. J. Wierer et. al. Laser Photonics Rev. 7 963 (2013)
Another strength of the blue laser is its superior efficiency. As the current through the LED is cranked up its efficiency declines due to a mysterious malady known as droop. Its origin is controversial, but the principal cause is probably Auger recombination. Droop does not plague lasers, however, with increases in drive current causing increases in efficiency to a stable level (see Figure 4).
Encouragingly, the blue laser is getting ever more efficient. Operating at 25 °C, 1.7 W chips produce a wall-plug efficiency of 25 percent, while the more recently launched 3 W successors have an efficiency of 35 percent, and may soon hit 40 percent. Further improvement is expected, as more manufacturers release products that will primarily target the video projector market. As they compete with each other to deliver better-performing products, better sources should become available for illumination purposes.
This level of performance is promising, but if the laser is to be used in the headlamp, it must operate under rather harsh environmental conditions. Headlamps are near the combustion engine and the batteries, so the local temperature can hit 80 °C. And inside a headlamp it can be even higher "“ in the case of the LED, in here it can reach 110 °C, while inside the chip the junction temperature can be 150 °C (see Figure 5).
Figure 5. (a) Temperature distribution on the hood of the front of a BMW vehicle as the combustion engine is turned on. (b) Temperature distribution around a headlamp. (c) Temperatures inside an LED headlamp culminates at 110 °C.
Fact is that high-power, multi-mode, blue laser diodes operate at lower junction temperatures than LEDs, with temperatures ranging from 90 °C to 110 °C. But even at these elevated temperatures, the performance is compromised, with a roll-over in performance at a case temperature of 70°C, occurring for a drive current of 2.3 A. So, if the laser is to deliver a performance that is good enough for use in a headlamp, it must be cooled.
Unfortunately, the commercial high-power blue laser diodes that are currently available are not designed to fulfil the automotive requirements, because they do not have an operating range spanning -40°C to 85°C. However, thanks to the remote-phosphor configuration, these chips can be prevented from overheating using the air flow around the headlamps to cool them as the vehicle moves. Using this approach the laser, which is driven in continuous-wave mode below the rollover current, is maintained below 60 °C (see Figure 6). This chip pumps a phosphor via a robust optical fibre (see Figure 7).
Armed with our new form of headlamp, we set our sights on extending the visibility range to the maximum tolerated by the European ECE regulations. This led to an illuminance of 1 lux at a distance of more than 600 m in front of the vehicle, realised by using the combination of an LED that generated a broad illumination pattern of 50 lux and quasi-collimated white light from a laser that generated a far-field spot of 290 lux (see Figure 8).
We estimate that the typical requirements for a light source that can deliver this level of performance are a peak brightness of more than 1000 cd/mm², an average brightness of more than 500 cd/mm² and a minimum luminous flux of 500 lumens.
There are also requirements for the phosphor, which has a conversion efficiency influenced by composition and temperature. If the phosphor gets too hot "“ this could result from heating by a laser spot size below 100 µm "“ it can quench luminescence.
To prevent this from happening, we house the phosphor in a reflective configuration that supports local cooling. The phosphor is about 100 µm-thick, has a radius of about 1 mm, and heat is efficiently dissipated through the thickness (not the radius) by carefully selecting the substrate so that it takes into account the coefficient of thermal expansion of the assembly. The emitting surface has a diameter of about 350 µm (full width at half maximum). In addition to aiding heat dissipation, this configuration allows recycling of back-scattered light from the phosphor, thanks to reflection at the phosphor-substrate interface.
Figure 6. Experimental setup showing white light produced by two cerium-doped yttrium aluminum garnet (Ce:YAG) phosphor-chips in a remote position that are excited by high-power multimode 450 nm laser diodes (over 1W).
Requirements for the phosphor are not limited to exceeding a particular efficiency, and also include colour requirements defined by regulations. Optimising the phosphor involves not only the determination of the material composition, but also the selection of the right grain size, grain distribution and density, and the thickness of the phosphor chip. And of top of this, the package assembly must be designed to maximise the useful luminous flux. We have done this, realising a conversion efficiency of more than 300 lm/Wradiant and a quenching temperature exceeding 220 °C. Operation in environments with temperature as high as 110 °C is possible with our phosphor assembly, which is located in the centre of the headlamp. The laser assembly is, however, positioned on the periphery of the headlamp, where a cooling mechanism can benefit from the air flow.
For our new form of headlight to be practical, its performance must be maintained over the lifetime of the vehicle − and this dictates a lifetime for the laser. Depending on the lighting functions under consideration, this chip will have to operate for between 2000 hours and 6000 hours. This is well below the lifetime of a commercial blue laser, which is typically 30,000 hours.
Figure 7. (a) First generation of laser diode-based light-engines composed of three 1 W laser diodes (b) BMW new generation of light-engines featuring compactness, cost-effectiveness and efficiency based on a single 3 W laser diode. Figure 8. Bird's-eye-view of a light distribution, illustrating the extension of the visibility range to the maximum tolerated by the ECE regulations.
Safety issues?
Using lasers for headlamps may raise a few eyebrows, given that the light emanating from this high-luminance source can cause eye damage. To prevent this from happening, we took stringent measures to protect drivers, pedestrians and wildlife. The high-beam laser-light booster can only be used above a certain cruising speed − such as 40 km/hr − and following the activation of the main LED-generated high-beam. On-coming traffic is detected automatically a long way away with the high-beam laser-light booster, so it can be turned off well before the vehicle gets close.
The high-power laser diodes used as this point source in this headlight are classified as Class 4 high-power blue laser diodes, so the package must be constructed in a manner that prevents light from leakage out. To increase safety, this unit features monitoring photodiodes that detect failure modes of the laser diodes, phosphor and the optics in situations that include crashes and degradation-induced effects. If failure is detected, the booster is automatically switched off.
Following several years of development, last summer we released the first vehicles equipped with laser-based high-beams "“ these were incorporated into the BMW i8, a hybrid plug-in sports car that can go from stand still to 62 mph in just 4.4 s. A limited production run of a few thousand vehicles demonstrated the possibility of the technology, and gave us the chance to gauge the views of our customers.
Demand for the i8 has been very high, and has led us to plan for a release of a new BMW model that combines the laser-based, glare-free, high-beam and a marking light, which we refer to as a dynamic light spot. We will produce hundreds of thousands of this model of car.
Producing a far higher volume of cars with laser-based headlights has not been easy. Many suppliers lack experience with laser technology and the use of high-brightness, white-light sources. Such sources require optics with higher optical quality. However, we expect that this situation will change as our supplies embrace this new, superior technology for headlamps, and we start to incorporate it in all our upper-class premium vehicles and our compact-class premium ones.
This increase in the deployment of this technology must go hand-in-hand with its refinement, with new approaches needed that integrate the technology in a more compact housing, and enable more streamlined manufacturing.
Looking ahead, we can expect this new class of headlamp to evolve and incorporate new laser-based lighting functions. This includes driver-assistance dynamic lighting, which requires high brightness sources and high resolution in the far field, and would be possible with a laser scanning technology. Galvano scanners are bulky and too slow for illumination purposes in the automotive sector, but MEMS-based scanners are more promising: They are extremely compact, offer high scan frequencies, and are robust enough when integrated in the appropriate package. The primary challenge is to uncover an appropriate configuration and optimal driving method for combining the beam of a laser emitting more than 4 W with a MEMS-based scanning mirror, so the result is a high-quality, free-patterned illumination.
Figure 8. Bird's-eye-view of a light distribution, illustrating the extension of the visibility range to the maximum tolerated by the ECE regulations.
This is not the only option "“ there is also the combination of optical phase modulation technologies and laser diodes. By accurately controlling the phase of a beam, free patterning of light distributions is possible. Coded patterns could be programmed to create dynamic light distributions without the need for high scanning frequencies, and using a synchronized sequential red-green-blue-system would reduce the number of modulators to one.
While it is not clear which of these technologies will win in the end, laser-based headlights clearly have a great future. And by lighting up the road ahead far better than their predecessors, they should reduce the dangers of going out for a drive after dark.
