Dicing System

Silicon wafers are grown as a spool conventionally leveraging the Czochralski growth method, upon which they are sliced into individual wafer then polished down. The process of crystal growth renders the resultant wafers as round disks of material, while a final wafer will contain up to thousands of square die. Post wafer fabrication it is necessary to singulate the wafer to separate the fabricated die for final packaging. Singulation commonly makes use of either dicing with a diamond tipped saw or a form of laser ablation.

From a home brew perspective either common dicing method isn’t really all the viable as saw dicing is often messy as it requires a constantly water flow, while laser ablation requires a high power laser coupled with a precision stage. Instead, other method would be preferred to achieve a lower cost and cleaner solution.

To achieve these aims I chose to leverage a technique found in the earlier days of the semiconductor industry, wafer scribing. The most common type of Silicon leveraged for devices is the <100> orientation due to its superior interface and electrical properties, and advantageously the Silicon will cleave along 90 degree angles. Then instead of having to precisely cut each time by hand or having to try to dig in initially from an edge before the cleave, tools were developed to better automate the process.

I obtained one of these early tools which contained a basic microscope to align the scribe lines, an actuated diamond tip scribe, and a motor to move the wafer along a single scribe line. The tool, however, needed a bit of TLC to get it back up and running, as well as adding a few missing parts. The system would not actuate well and I was able to replace the solenoid for the unit. There appeared to be a spot for a light, but the light was missing where I was able to add on a custom adjustable LED light source. Then finally I added a basic diaphragm pump with a power supply in order to hold down the samples and subsequently power all the components.

The system works well now and is great for dicing up Silicon pieces. The main drawback of the method is that other materials like GaN do not cleave this way, instead cleaving like a “pizza pie” (similar to <111> type Silicon). In these cases scribing isn’t as effect, however with a deep enough scribe, one can still get the material to generally cleave where you want.