DFM

'Design for yield' is a familiar term, but the challenges in today's increasingly large projects make a refresher on what it offers particularly timely.

In the absence of EUV lithography, the primary option for manufacturing on a 10nm process is to extend double patterning. But the options each have issues.

The encryption chain for today's highly collaborative designs needs to be managed with care.

Lithography is only just beginning to play a role in cell IP selection but early analysis already matters.

It is not just a choice between EUV and multiple patterning for future nodes, but even between varieties of multi-mask technologies. How will you decide?

A number of effects have led to a dramatic increase in interconnect resistance in the sub-32nm process nodes that demands the use of smarter routing.

Increasingly complex design rules in 14nm and 16nm make it harder to connect local routing to the inputs and outputs (pins) of standard cells.

If EUV is further delayed until 8nm, the industry has to explore other options for patterning, and the effects they will have on the DFM flow.

Extracting finFET parasitics means a shift to 3D models, field solvers for greater accuracy, and MCMM techniques.

The costs of advanced lithography techniques at 1xnm, and the yield and reliability risks from the resultant process variation, will stop many companies getting the typical economic advantages of scaling.