( ELSE 06 Item 24 ) -------------------------------------------- [ 06/23/06 ]
Subject: Mentor Calibre MDP, Xyalis, Brion Technologies
MASK DATA PREP -- What follows below is an interview with a litho guy with
his thoughts about the Mentor Calibre MDP tool. I wish to apologize for the
"intro" aspect to this; it took me and this engineer a bit of work to sync
up on what he was saying, I figure that maybe many of my readers might need
some of this litho language explained, too. The Calibre MDP review is
buried in the explainations. Enjoy, if you can understand it! :)
I'm responding to your request for input on the Calibre MDP tool. I'm a
mask lithographer. I'm about as backend as you can get in the design
flow, John, and as such, I talk a different language than many of your
readers, so I hope they'll bear with me. I don't understand most of
what your readers say either!
"Fracture" is the process of translating polygonal stream data (GDS2 or
OASIS) into mask lithography tool formats. It's becoming increasingly
important to optimize both mask write time and mask Crital Dimension
(CD) control. Advanced masks are exposed using Variable Shaped Beam
(VSB) electron-beam lithography systems. These tools compose the mask
pattern from rectangular and triangular primitives ("shots") that are
decomposed within the write tool from the fracture format ("figures").
For complex patterns, there is almost a 1:1 correspondence between the
shots and fracture figures. To first order, write time is directly
related to the number of fracture figures. Write time is a major
component of mask cost and cycle time, and to a lesser extent mask
yield. It is desirable for the fracture software to create the minimum
number of figures necessary to accurately represent the pattern to
minimize the write time.
However, there is an element of fracture quality to consider. Features
that are composed of two or more shots in the critical dimension are
statistically likely to have greater size and placement variability than
features that are composed of only one shot. This is due to the addition
of size and placement errors of the individual shots. Thus, while the
total number of shots is important, the arrangement of these shots is
also important to have the best CD control.
Basic examples of a good fracture strategy include not splitting gates
lengthwise and attempting to orient "slivers", or extremely thin shots,
perpendicular to the critical dimension. In our experience Calibre MDP
does this as good or better than anyone, and the SW's user-adjustable
parameters that are set to optimize the fracture strategy for each
individual pattern respond predictably. It has become commonplace to
have to perform fracture Design of Experiments (DOE) matrices with a
small piece of pattern data to empirically derive the best combination
of fracture parameters for complex patterns -- which isn't necessary
if you're using Calibre MDP.
Mask Data Prep is starting to look a lot like litho optimization for
wafer lithography. Besides the "normal" translation requirements for
mask manufacturing (rotation, scaling, mirroring, reverse toning,
sizing, etc.) there is an increasing need for pattern analysis and
context-based sizing. To meet the CD uniformity requirements for
advanced nodes at low k1 values (sorry for the litho-speak) it is
desirable to characterize and compensate for local CD variations that
are a function of pitch, line width, or pattern density. One huge
advantage Calibre MDP has, in my opinion, is it's ability to draw on
the enormous potential inherent in Calibre DRC and RET. Having this
functionality on the same platform as MDP allows the mask manufacturer
to easily take advantage of the OPC knowledge already in use to improve
wafer lithography.
As a relative new-comer to MDP, Calibre has had to climb the learning
curve quickly. Perhaps their biggest problem has been integrating the
Calibre MDP flow into the established Synopsys (Numerical) CATS flow
which flatten hierarchy early and act on fracture formats rather than
on hierarchical stream data. Initially Calibre MDP was inefficient
dealing with flat or mostly flat data especially when not originally
created by other Calibre processes. More robust read-in processes that
adapt to flat data, the introduction of Calibre's MTflex distributed
processing strategy, and section-based processing seems to have resolved
these early problems. We still have to use the incumbent Synopsys CATS
tools to translate its proprietary formats to GDS or OASIS, but the
trend towards using OASIS instead of fracture formats to transfer data
to the mask manufacturer seems to be slowly resolving this issue.
- [ An Anon Engineer ]
Mask Related Tools
Xyalis sells a metal fill tool that is rule based (can use any shape)
resulting in a much smaller database than their competitors according
to their data.
Shearwater sells Lavis, a Japanese tool that is a viewer for GDSII,
LEF/DEF, MEBES, etc. but also allows editing and Boolean operations.
They claim is high performance and high capacity and is used by Aprio
for their viewer. They sell another tool from Saratoga Data that
optimizes and compresses GDSII data. They say compression is 75%-98%
and averages about 96%, while still retaining hierarchy.
Artwork Conversion Software sells a variety of utilities such as fast
GDSII and MEBES viewers and plotters, and tools that can do Boolean
operations on GDSII or extract pad information automatically.
Laflin sells Hotscope, a high speed GDSII and MEBES viewer.
Brion Technologies sells a mask inspection tool that they say is faster
than KLA because they have a hardware accelerator.
- John Weiland of Intrinsix Corp.
Index
Next->Item
|
|
