Selective decomposition of nitride semiconductors to enhance led light extraction
AssigneePalo Alto Research Center Incorporated
InventorsDavid P. Bour, Clifford F. Knollenberg, Christopher L. Chua
Life of this patent
Estimating the term…
Abstract
A method of texturing a surface within or immediately adjacent to a template layer of a LED is described. The method uses a texturing laser directed through a substrate to decompose and pit a semiconductor material at the surface to be textured. By texturing the surface, light trapping within the template layer is reduced. Furthermore, by patterning the arrangement of pits, metal coating each pit can be arranged to spread current through the template layer and thus through the n-doped region of a LED.
Claims25
A method of texturing a first surface to be textured in a LED comprising;
- forming a LED structure over a substrate,
- using a focused texturing laser with a spot size less than 0.01 mm2 to produce a series of pits in the first surface to be textured, the first surface to be textured positioned between the substrate and the active region of the LED.
The method of claim 11 wherein the texturing laser is a frequency quadrupled YAG laser to achieve a wavelength near 266 nm.
A method of roughening a LED surface to improve light extraction efficiency comprising the operations of:
- forming a n-doped layer over a substrate,
- forming an active region on the n-doped layer, the active region to emit light;
- and,
- using a focused texturing laser delivering a power per unit area in excess of 0.4 J/cm2 to pit a surface between a substrate and an active layer of the LED.
A nitride based LED structure comprising:
- a substrate;
- a template layer coupled to the substrate;
- a n-doped layer coupled to the template layer;
- an active layer coupled to the n-doped layer;
- and,
- a textured surface between the substrate and the active layer, the textured surface including a plurality of pits having a depth between 50 nm and 5 microns, the textured surface to redirect light output by the active layer.
Dependencies are inferred from the claim text.