Uncovering The Secrets Of High-performance Green Lasers
Photoluminescence exposes improvements within the quantum well
Photoluminescence intensity uniformity is far higher in the active regions of state-of-the-art green lasers, than the green-emitting structures of yesteryear.
Green-emittingc-plane lasers have rapidly advanced from nowhere in recent years. Prior to 2009 such a device didn’t exist, and since then performance has leapt from a 515 nm laser emitting 5 mW to a 525 nm, 1.01 W source. Although the driving force behind these improvements has been something of a mystery, secrets are now unravelling, thanks to research by a team from Kyoto University, Japan, and Nichia Corporation.
This partnership has shown that the strong performance of today’s best lasers – such as the 1.01 W, 525 nm laser reported by Nichia in late 2012, which has a threshold current density of 1.68 kA cm-2 and a wall-plug efficiency of 14.1 percent – could stem from a very low level of potential distributions in the active region.
“Potential fluctuations generally broaden gain spectra and as a consequence reduce maximum gain, which makes lasing difficult because gain cannot overcome loss," explains Mitsuru Funato from Kyoto University. It is likely that these fluctuations result from variations in the InGaN composition within the quantum well.
To study the level of fluctuations in state-of-the-art green lasers, the team have used optical techniques to scrutinise a laser structure grown by engineers at Nichia. This MOCVD-grown heterostructure was not intentionally doped with the likes of silicon and magnesium to make it easier to assess its fundamental optical properties. However, insights into its potential lasing characteristics were uncovered through the fabrication of an identical structure with p-type and n-type doping. This lased at 512 nm, and had a threshold current density of just 2.75 kA cm-2.
Several characteristics of the undoped structure were compared with that of a typical green-emitting InGaN/GaN quantum well of yesteryear – a ‘conventional’ structure that produced a photoluminescence (PL) spectra, but was of insufficient material quality for making a laser. Based on PL spectra at 7K and room temperature, the undoped structure has an internal quantum efficiency of 27.5 percent at 295 K.
By curve fitting these spectra, the researchers revealed that, at low temperatures, the full width half-maximum for the PL is 70 meV and the Stokes shift is 60 meV. They believe that these very low values indicate that the inhomogeneous broadening in the state-of-the-art structure is remarkably suppressed compared with that in the wells of the conventional structure.
Microscopic PL measurements were performed on both types of sample to determine spatial inhomogeneity. Many bright spots were present in the intensity map produced for the conventional sample. Dislocation density in this structure is 2 x 108 cm-2, and the density of bright spots is far higher than this (see Figure), so the team concluded that potential fluctuations are responsible for the inhomogeneity. In contrast, the microscopic PL measurements on the undoped structure show uniform fluorescence.
To probe this sample more rigorously, the researchers turned to confocal microscopy, uncovering islands with dimensions of typically less than a micron. They found variations in intensity from 53 percent to 100 percent, which is far smaller than that seen in the conventional structure, where intensity fluctuations can vary by more than 90 percent.
Emission wavelengths were also far more uniform in the undoped structure. The peak wavelength in this sample varies from 536.9 nm to 539.5 nm, while that produced by a typical green-emitting InGaN/GaN well from several years ago had a emission wavelength variation exceeding 15 nm. The team has also carried out time-resolved photoluminescence measurements. They revealed a recombination lifetime for the undoped structure of 68 ns.
Armed with this broad range of measurements on the undoped structure, the researchers went on to determine that the composition of the well is 28 ± 2 percent and its thickness is 2.5 ± 0.27 nm (9 ± 1 monolayers). This indicates that the undoped well is thinner than its conventional counterpart, and its indium composition is higher. A thinner well has its advantages, because it increases the radiative recombination efficiency by reducing the impact of electric fields resulting from piezoelectric polarisation. However, it is possible that a thinner well reduces optical confinement, due to its smaller volume.
Fabricating a high-quality, indium-rich well will not have been easy. “To achieve high indium composition, the growth temperature must be lowered, which degrades crystal quality," explains Funato. This can mean that growth pits appear in the InGaN films, which can incorporate unintentional impurities or defects that act as deep carrier trap centres.
The team will now measure gain in laser diodes featuring intentional doping.
M. Funato et. al.
Appl. Phys. Express 6 111002 (2013)