Directed Self Assembly, double patterning and crying in beer

DSA will almost certainly play a role in future double patterning.


In the creative, or desperate, rush to find ways to pattern 10 nm node using double patterning immersion 193nm lithography, a designer from ARM is left “crying in his beer” at the consequent design difficulties. This heart rending admission was one of many insights that came at the sessions on Directed Self Assembly (DSA) at Advanced Lithography.

Double Patterning has become the -we have no choice- path forward to 10 nm node lithography, and Directed Self Assembly ( DSA) has become the path of choice to reduce the impact of double patterning. The idea of double patterning is to divide and conquer the problem. Force the designers to layout the device patterns as a set of oriented lines. First pattern the lines, and then separately pattern the breaks in the lines – hence double patterning. Unfortunately, at 10 nm node, even the simple set of oriented lines is beyond the resolution of immersion 193 nm steppers ….so double pattern the lines – double double patterning?
DSA simplifies the problem because only every other line is patterned by lithography. Every other line directs the self assembly of the block copolymer that has a natural pitch equal to the final pattern pitch. Patterning the breaks in the lines remains challenge, with EUV or multiple e-beam being proposed as alternatives to multiple 193 nm patterns.

At advanced lithography there were all sorts of papers on DSA; defect density, strategies for directing patterns to get closer to real device patterns, pattern transfer, smaller pitch materials, and design implications.

The defect density papers all showed great progress. Kurt Ronse from Imec summarized the results to me “as indicating that there did not appear to be anything unique about DSA defects, and I do not see any fundamental barriers to implementing DSA.” DSA relies on thermodynamic equilibrium to create features, a fundamentally different mechanism from conventional photoresist. Many of us thought that kinetic barriers to reaching the equilibrium might cause permanent pattern defects in DSA. It now looks like those worries are unfounded and device quality DSA is going to happen.

The directing papers, showed how to create real device like patterns with DSA. One key trick is to make sure that the thickness is a multiple of pitch so if the surface is un-patterned, the phases lay flat. Now if the surface is patterned the phases are directed to form vertical patterns, if the surface is unpatterned the phases are horizontal and unpatterned. Other groups focused on “grapheo-epitaxy” using the walls of large geometries to direct the positioning of smaller phases. This has the potential to place multiple small features in one large geometry as shown below. The challenge is the design and layout rules required to direct the right pattern to the right place. The conference had multiple examples of DSA with device like patterns.

Guided multiple contact patterns form Reluca et. al. Leti.

In all applications, one of the blocks must be removed to create a usable resist pattern. Dry etch has been the solution of choice. Tiron Reluca and a group from Leti in France proposed unzipping one of the blocks using deep UV so that it could be removed by a liquid developer just like a photoresist. They showed lower levels of line edge roughness.

Finally, there were a bunch of papers on different material that would phase separate at less than 10 nm half pitch. The phase width scales as the square root of the molecular weight, so at 10 nm these materials with 10 -15 monomer repeats – are only just polymers. These small phases were first reported in 1978 …unapologetic self promotion … in a non patterning application.

All of the double patterning approaches and many of the directing strategies involve adding new layout constraints (design rules) … otherwise known as passing the buck. I though the most provocative paper at the conference was a joint Lars Liebmann/David Pietromonaco, IBM/ARM, contribution that focused on the device design implications of design rules. The IBM processing guy talked about processing ideas, and the ARM design guy explained why he hated them . The designer (Pietromonanco) showed why he needed to design complete functional blocks to understand how a particular design rule affected layouts. He focused on the fact that oriented lines with breaks design rules interacted with other device rules to force him to use additional layers on interconnect to wire the transistors together. For one popular example, he found he needed 6 more metal layers. Other rules left him “ too busy crying in his beer to care”. His point was that the additional metal layers further eroded any $ per transistor gains from shrinking, and so undercuts the whole commercial rationale for shrinking. He also pointed out that the ASIC foundry guys would find these constraints worse than the memory or large microprocessor guys.

I though this paper was an example of one of the things these conferences can do best, bring specialists from different disciplines together to illuminate real world challenges.

As far as double patterning is concerned, DSA is almost certainly going to play a central role. If you are working on directing strategies, it is essential to have a designer to talk to.

About the Author

Mike Watts has been patterning since 1 um was the critical barrier, in other words for a longtime. I am a tall limey who is failing to develop a Texas accent here in Austin. I have a consulting shingle at

My blog “ImPattering” will focus on the latest developments in the business and technology of patterning. I am particularly interested in trying to identify how the latest commercial applications evolve.


[…] into triple for 10 and likely quadruple for 7. And if double patterning can make a designer cry in his beer, then quadruple patterning will make him switch to scotch. But it’s not just that taping out a […]

Leave a Reply

(Note: This name will be displayed publicly)