Nanoscale imaging of reduced forward bias at V-defects in green-emitting nitride LEDs

TL;DR

This work tackles the green-gap in wall-plug efficiency ($WPE$) for long-wavelength III-nitride LEDs by testing a charge-injection mechanism where V-defects' semi-polar facets reduce internal barriers, enabling higher electrical efficiency ($EE$). The authors use a novel nanoscale approach with a scanning tunneling luminescence microscope (STLM) tip as a local hole injector to map local optoelectronic properties around V-defects, measuring both current paths and light emission with sub-10 nm resolution. They observe a ~1 V reduction in the local forward bias ($V_F$) at V-defect rims and a small ~10 meV blueshift in the emitted electroluminescence, directly supporting the proposed injection mechanism through the V-defect facets into near-surface quantum wells. This combination of nanoscale electrical and optical probing demonstrates that injected current primarily traverses the multi-quantum-well region via the $\{10\bar{1}1\}$ facets, with lateral transport to $(0001)$-plane QWs prior to radiative recombination, providing a direct validation of the mechanism behind WPE improvements in green-emitting nitride LEDs.

Abstract

Record wall-plug efficiencies in long-wavelength, III-nitride light-emitting diodes (LEDs) have recently been achieved through improvements in electrical efficiency in devices containing V-defects. Numerical modeling suggests this may be due to reduced barrier heights for charge injection in thinned, low-Indium quantum wells parallel to semi-polar V-defect facets. To test this proposition, a novel approach in which the tip of a scanning tunneling luminescence microscope as a local hole injector, is used to map the optoelectronic properties of commercial, green-emitting LED heterostructures around V-defects with nanoscale spatial resolution. A 1 V reduction in the forward bias necessary for current injection at V-defect rims is observed. This, combined with the observation of small (~10 meV) blue shifts in the locally emitted electroluminescence, unambiguously confirms the charge injection mechanism.

Figures (11)

• Figure 1: Macroscopic characteristics of the functional LED. (a) The current-voltage characteristic at 300 K with a sketch of the LED heterostructure shown inset. (b) The relative EQE, along with the corresponding electroluminescence spectra measured at a forward current of 20 mA obtained when $\textrm{V}_{\textrm{LED}} \approx 3$ V as indicated by the dotted lines in (a).
• Figure 2: (a) A schematic drawing showing the addition of the STLM scanning tip used to locally inject holes ($\textrm{V}_{\textrm{TIP}} > 0$) into the ALE-treated heterostructure where the removal of the p-GaN exposes the V-defects at the sample surface as seen in the topography, (b). The topography is obtained using V$_{\textrm{TIP}} = 5.5$ V and V$_{\textrm{LED}} = 0$ V with a set point current of $\textrm{I}_{\textrm{TIP}} = 2$ nA. (c) The I$_{\textrm{TIP}}$ map corresponding to the topography in (b) shows large excess current, up to the µA range, particularly near V-defect rims. (d) The $\Delta\textrm{I} = \textrm{I}_{\textrm{n}}-\textrm{I}_{\textrm{p}}$ map is identical to the I$_{\textrm{TIP}}$ map to less than 0.1 % at the V-defect rims, demonstrating that tip-injected current passes through the heterostructure to the n-type ohmic contact.
• Figure 3: (a) Example current-voltage characteristics of the local LED (see inset). The black and red curves correspond to hole injection from the tip into a V-defect rim and a $\left(0001\right)$-plane at the points of the same color indicated in the topographic map, (b). (b) The topographic map obtained using $\textrm{V}_{\textrm{TIP}} = 7$ V and a current set point, $\textrm{I}_{\textrm{TIP}} = 0.5$ nA. (c) The simultaneous $\textrm{V}_{\textrm{F}}$ map (obtained for $\textrm{I}_{\textrm{F}}$ = 10 $\mu$A) showing a systematic reduction in $\textrm{V}_{\textrm{F}}$ at V-defect rims.
• Figure 4: (a) Masks defined to select image pixels corresponding to V-defect rims (top left) and the $\left(0001\right)$-plane (bottom left). The middle and right frames show these masks superimposed on the topography and on the ($\textrm{I}_\textrm{TIP} = 10 \mu$A) $\textrm{V}_\textrm{F}$ map respectively. (b) Average $\textrm{V}_\textrm{F}$ calculated for a range of arbitrarily chosen values of $\textrm{I}_\textrm{TIP}$ on pixels selected by the masks shown in (a). The difference ($\Delta\textrm{V}_{\textrm{F}}$), (black circles) is the measured difference in forward bias obtained when injecting holes directly into the V-defect rims rather than the heterostructure's $\left(0001\right)$-plane.
• Figure 5: (a) STM topography of the ALE heterostructure ($\textrm{V}_{\textrm{TIP}} = 6.5$ V, $\textrm{V}_{\textrm{LED}} = 0$ V, and $\textrm{I}_{\textrm{TIP}} = 0.5$ nA). (b) A typical local electroluminescence spectrum exhibiting a phonon replica (light green curve) and a centroid around 2.3 eV (orange curve).(c) The integrated electroluminesence intensity showing clear maxima around V-defect rims where excess current occurs (see also Fig. \ref{['experiment']}(c)). (d) Centroid energy map showing electroluminescence blueshifts on the scale of 10 meV near V-defect rims.