BY CHANGMIN LEE, JAMES S. SPECK, SHUJI NAKAMURA AND STEVEN P. DENBAARS FROM UCSB, CHAO SHEN AND BOON S. OOI FROM KAUST, AND AHMED Y. ALYAMANI AND MUNIR M. EL-DESOUKI FROM KACST
A radio-frequency famine looms. Usage of mobile wireless is rocketing, and compromises must be made to meet this demand within the allocated spectrum. Introducing more complex systems are on the agenda, alongside new coding to support a surge in mobile data, but whichever path is taken will be challenging and inefficient.
One way to mitigate the impending RF crisis is to shift to the visible spectrum. This is not a new idea, as visible light communication has been studied since the emergence of blue LEDs with incandescent-level output powers. But recently this technology has been gathering pace, spurred on by the first demonstration of light-fidelity – commonly known as LiFi – in 2011, by Harald Haas’ group at the University of Edinburgh, UK.
Although the data rate of LED-based visible light communication is increasing, thanks to advances in both device design and the technology of the communication system, the modulation speed is limited by the relatively slow response of the LED. A typical commercial device requires a chip size of about 1 mm2 to produce sufficient light for illumination, and this leads to a high capacitance. And that’s bad news for LiFi, because a high degree of charge trapping occurs, limiting the bandwidth and ultimately applying the brakes to the switching speed of data transmission.
Trimming the size of the LED increases data transmission speeds. MicroLEDs have a bandwidth that is predominantly limited by the lifetime of carriers, which is of the order of nanoseconds, and their modulation bandwidth can approach 1 GHz.
However, while the data transmission rate may be suitable for visible light communication, the output power falls woefully short. Cutting device size is actually something of a double-whammy, with light output falling through a reduction in chip size, and taking a further hit through a more severe efficiency droop.
A panacea for both these ailments – that is, the low modulation bandwidth and the significant efficiency droop – is a switch to a laser-diode-based lighting system. Efficiency droop is eliminated, because the carrier density clamps above the lasing threshold, while emission is then governed predominantly by photon density – and that means that faster data transmission is possible, because laser diodes can be modulated at far higher speeds than LEDs (see Figure 1). Take this approach, and bandwidth is no longer shackled by carrier lifetime, but instead limited by the photon lifetime of the cavity, which is of the order of picoseconds. What’s more, impacts of capacitance and resistance on switching speed are also diminished, because a comparable light output power is possible with a smaller chip.
Figure 1. Modulation bandwidths for laser diodes in the blue-violet spectrum are far higher than those for comparable LEDs.
Due to these strengths, the laser diode lies at the heart of optical telecommunication systems. Devices based on GaAs and InP have been deployed in vast numbers over many decades, and are now delivering bandwidths of up to around 100 GHz.
One of the biggest opportunities for laser-based LiFi is in homes and offices, where it can simultaneously provide communication links and lighting. But its promise extends beyond this. The multi Gbit/s data rates equip the technology for large-capability broadcasting, vehicle autopilot for smart traffic, and real-time telesurgery. In addition, it could be used outdoors, as it can operate at high powers, thanks to the capability of laser-based white-light sources to generate a luminous flux above 1000 lumens.
Another feature of visible light communication is the creation of a line-of-sight optical network. Light cannot penetrate through walls, so there is a high level of privacy and security associated with this form of free-space data communication. Conventional RF based communication techniques, including Wi-Fi, Bluetooth, and cell phone networks, can’t offer this. These networks have a high risk of being hacked, making them undesirable for commercial and military use. With laser-diode based communication, a further attribute is the higher emission power per unit area. The small footprint reduces packaging costs, and there is the promise of incorporating LiFi in mobile devices.
Working with the phosphor
If violet or blue LEDs are used to make high power white-light communication systems, they can incorporate phosphor materials to span the longer wavelengths (this is a common approach in LED lighting, where blue LEDs are coated with yellow phosphors). There is a significant downside with this combination, however: the modulation bandwidth is significantly limited by the slow phosphor relaxation lifetime, which is typically less than 100 ns, and it cannot exceed a few megahertz. Rejecting slow phosphor-converted colour components with optical filters overcomes this limitation, but at the expense of increased complexity of the system.
A more elegant approach is a laser-based white-light communication system, because this is not restrained by the phosphor response at high frequency. This would be possible by combining several lasers at different wavelengths to generate a source with a high colour-rendering index. An alternative is to use a blue laser in phosphor-based white-light system and modulate at gigahertz frequencies – this is a vast improvement on the megahertz domain for phosphor-converted signals that would then only constitute the background noise.
At University of California, Santa Barbara (UCSB), our team is developing high-speed III-nitride laser diodes and photonic devices for visible light communication, in collaborating with researchers at the Photonics Laboratory at King Abdullah University of Science and Technology (KAUST) and King Abdulaziz City for Science and Technology (KACST).
To determine what is possible with commercial technology, we have assessed the capabilities of a commercial laser diode. It produces a 2.6 GHz modulation bandwidth with a Gbit/s on-off keying data rate. That’s a 100-fold improvement in bandwidth over a commercial LED – and even higher rates are possible by turning to a spectral-efficient modulation scheme, such as orthogonal frequency division multiplexing.
Identifying the best laser
We have compared these modulation rates with those of a class of laser that we have pioneered in our department – an edge-emitter on either a semi-polar or non-polar plane of GaN. These devices differ from conventional LEDs and lasers diodes, which are grown on the c-plane of the Wurtzite crystal structure of GaN and plagued by polarization-induced electric fields in the quantum wells (see Figure 2). These fields cut the overlap of the electron and hole wave-functions, driving down the internal quantum efficiency of conventional devices.
Figure 2. Phosphors are detrimental to the modulation bandwidths of white lighting systems based on LEDs and laser diodes.
Our lasers have far weaker polarization-induced electric fields, and thus a much higher wave-function overlap. Advantages stemming from this include less droop, a higher output power, a higher optical gain, and a higher differential gain. The latter is hugely beneficial for LiFi, because it should lead to a higher modulation bandwidth.
Putting this theory to the test, we have measured the modulation bandwidth of one of our violet semi-polar laser diodes. It is capable of operating at 5 GHz, setting a new benchmark for all blue-violet lasers and LEDs (see Figures 3 and 4). Using on-off keying, a data rate of 5 Gbit/s is possible, which is the highest value in this spectral range for this transmission technology. Note that the bandwidth of the photodetector in our system limits our measurement, and our devices could reach even higher bandwidths through optimization of the epitaxial structure and device design.
Figure 3. Switching from the polar c-plane to a semi-polar plane improves the wave-function overlap for InGaN/GaN quantum well structures. Results shown for a structure under bias.
Figure 4. A 5 GHz bandwidth by small signal modulation (left) and 5 Gbit/s of on-off keying large signal modulation (right) for a 410 nm semi-polar laser diode.
To evaluate the capability of our laser-based visible light communication, we have considered the physical parameters of our semi-polar lasers. Calculations indicate an intrinsic maximum bandwidth of up to 27 GHz, limited by the photon lifetime in the laser cavity, rather than values for device capacitance and resistance. This theoretical limit exceeds that for the LED by more than one order of magnitude.
Selecting a violet laser for the source delivers several benefits, including those associated with data transmission, material growth, and lighting. As sunlight has a lower irradiance in the violet than at longer wavelengths, this laser allows data signals to have a lower background noise level under sunlight (see Figure 5) – and it’s a similar story under indoor lights. This means that visible light communication provides a robust form of communication, regardless of whether it is night or day.
Figure 5. Violet spectrum overlapping with solar spectrum (left) and the CIE plot of white emission by pumping a red-green-blue phosphor with a violet laser diode (right).
Additional strengths of the violet laser include a lower cost than its blue cousin and superior performance, thanks to a lower indium content in the quantum well (typically 10 percent, rather than 20 percent). What’s more, the use of a violet laser improves colour quality. If lighting systems are formed using phosphors and lasers, the combination of an ultra-violet source and red-green-blue emitting phosphors deliver a higher colour-rendering index and a warmer colour temperature than the pairing of a blue laser and yellow phosphor. The colour-rendering index can exceed 90, while the correlated colour temperature is less than 3000K.
Our efforts will now focus on developing a higher speed laser through improvement in the epitaxial structure and the device architecture. Also, we will develop InGaN-based photonic passive components, such as a modulator, optical amplifier, and photodetector, as the absence of high-speed photonic components for the blue-violet spectrum would be a bottleneck for visible light communication systems in the near future.