SPIE Digital Library Get updates from SPIE Newsroom
  • Newsroom Home
  • Astronomy
  • Biomedical Optics & Medical Imaging
  • Defense & Security
  • Electronic Imaging & Signal Processing
  • Illumination & Displays
  • Lasers & Sources
  • Micro/Nano Lithography
  • Nanotechnology
  • Optical Design & Engineering
  • Optoelectronics & Communications
  • Remote Sensing
  • Sensing & Measurement
  • Solar & Alternative Energy
  • Sign up for Newsroom E-Alerts
  • Information for:
SPIE Defense + Commercial Sensing 2017 | Register Today

OPIE 2017

OPIC 2017




Print PageEmail PageView PDF

Optoelectronics & Communications

An efficient light source for fast, low-cost optical communications

Using surface plasmon polaritons to accelerate the recombination of charge carriers increases the data-generation rate of a light-emitting diode.
30 December 2008, SPIE Newsroom. DOI: 10.1117/2.1200812.1421

LEDs are reliable and inexpensive light sources that are ideal for short communication links for computing applications, typically between or within circuit boards.1 Their only drawback is their modulation bandwidth—i.e., the rate at which they can be switched on and off—which is usually limited to less than 1Gbps.2 Using slow LEDs for high-speed communication requires several devices and several frequency channels, which demands more space on the chip as well as multiplexing and demultiplexing units, all of which severely impacts the cost of the link. A fast LED can provide optical data at a high rate in a single frequency channel and is by far the preferred solution. There are several speed limitations in an LED, however, the most severe of which is the finite recombination time of carriers (electrons and positive ‘holes’) in the light-generating process. The faster the recombination, the faster new carriers can be introduced in the active region to produce the next bit of signal.

The carrier radiative recombination rate in a semiconductor depends on the electron and hole concentrations in the active layer and the radiative recombination rate of an isolated electron-hole pair. One way to increase the modulation speed of an LED is to introduce a large carrier concentration in the active region. A higher modulation bandwidth of ~1.7GHz has been demonstrated2 using carrier concentrations on the order of 1019cm−3. But it is not possible to increase the concentration further without heavily treating the semiconductor with dopants, which unfortunately tend to act as nonradiative recombination centers for electrons and holes. This degrades the efficiency. To achieve ultrafast modulation without losing efficiency, we harnessed the so-called Purcell effect.3

Figure 1. Radiation pattern expected from the plasmonic LED without grating (left) and with (right). Colors represent the amount of power that is emitted by the device at various angles, as measured by the projection of the k-vector of the light component onto the horizontal plane. Red is maximum power, blue is zero power.

As shown by Purcell half a century ago, the electromagnetic environment can drastically alter the spontaneous emission rate of a light emitter,4 such as electron-hole pairs trapped in a quantum well (QW). For instance, a gallium nitride (GaN) QW placed a few nanometers from a silver film can see a radiative enhancement—or Purcell factor—as high as 1000 in the 400–500nm wavelength range.5,6 This effect finds its roots in the excitation of the metal free electrons at its surface, which in turn creates intense local electromagnetic fields known as surface plasmon polaritons (SPPs).6 Although SPPs in that regime can potentially reduce the carrier radiative lifetime to below a picosecond, they are also strongly absorbed by the metal without producing any useful radiation, which limits the prospects of fabricating a reasonably efficient device. We used weaker SPP modes at higher wavelength, giving smaller Purcell factors but able to be efficiently converted into useful radiation, via either a grating or the natural roughness of the metal surface.

Figure 2. The LED structure used in this work. The device is made of a p-i-n junction double heterostructure with a single tensile strained GaAs0.88P0.12quantum well (QW, red layer), located a few tens of nanometers away from a silver (Ag) surface.
LED design

To design the plasmonic LED device, we numerically simulated the light emission from a transverse magnetic (TM) dipole (representing the QW exciton) above a flat silver surface. This tells us which wave components ‘draw’ the most energy from the dipole to turn into plasmons. In a second simulation, we designed a grating structure optimized to re-radiate as much energy as possible from the SPP modes into free space. We found that for a TM dipole located 40nm above the silver surface and radiating at 800nm, a Purcell factor of 10 was attainable. The radiation efficiency could be boosted from 18% without grating to 30% using a square grating with square holes 5nm deep and filling 25% of the surface area. Figure 1 shows the radiation pattern from the device without grating (left) and with (right).

Optical pumping experiment

We performed an initial optical pumping experiment to confirm the faster rate of the LED device. Instead of a grating, we relied on the natural surface roughness of the silver film to re-radiate the SPP modes. The generic structure we tested is shown in Figure 2. A tensile-strained GaAsP/AlGaAs (gallium arsenide phosphide/aluminum gallium arsenide) QW emitting light at 800nm wavelength was positioned short distances (e.g., 40nm) from a silver-semiconductor interface. The tensile strain ensures that the QW has a large TM component and interacts fully with the TM polarized plasmon mode. Time-resolved fluorescence data was taken on LED structures with various QW-silver distances, after excitation by 2ps light pulses from a laser diode operating at 780nm. A reference sample with no silver was also tested.

Typical results are shown in Figure 3. The decay time for the sample with a 40nm gap between the QW and the silver is ~44ps (corresponding to a 3dB modulation speed of 3.6GHz), while the as-grown sample has a decay time of ~500ps (corresponding to a maximum modulation speed of 300 MHz), in other words, a tenfold increase in speed as expected. Moreover, the plasmon device was 1.3 times brighter than the reference sample. A systematic study of carrier lifetime versus QW-silver gap is shown in the inset of Figure 3. As anticipated, the decay time increases with the QW-silver gap, as the interaction with the SPP modes becomes weaker. We extrapolate that for free-carrier densities of the order 1019cm−3 in the active region, the corresponding plasmon-enhanced device would have a bandwidth well over 10GHz.

Figure 3. Time-resolved photoluminescence for two structures: as-built epitaxial stack (red solid line) and epitaxial stack with 40nm gap between the QW and the silver layer (dashed black line). The 40nm sample has a double-exponential profile. The slow response time (≃300ps decay time) is due to the measuring apparatus, which we have characterized separately. The inset shows a plot of the decay time versus the distance between the QW and the silver surface.

We have designed an efficient plasmon-enhanced LED with a carrier recombination time of just 44ps. Optical communications based on a more efficient, high-modulation-bandwidth LED could provide an ideal solution for low-cost, short optical links between and within circuit boards. The next step is to measure the efficiency of the improved LED operating under electrical pumping.