Smaller design pattern feature sizes continue to increase mask data file sizes, which increases mask data processing
(MDP) times. To satisfy the need for faster turn-around-time, MDP has progressively migrated from single-computer
computation, to multi-threading, and then to distributed processing on multiple computers. The availability of low cost
multi-core processors can be used advantageously to reduce Mask Data Preparation runtime. Compared to single core
processors, multi-core processor have higher performance, however, total available memory and I/O bandwidth need to
be increased proportionally with the additional cores. Memory per core and available I/O bandwidth limit the maximum
number of cores that can be effective with distributed processing. When a single job is broken down to 2 or more tasks,
the granularity of the tasks influences the efficiency of the processing. Smaller tasks allow for smaller memory footprint,
better distribution of tasks and increased scalability, but increase input file access time and reduce output data
compaction. By choosing a combination of multi-threading and distributed processing, faster run-time and better
scalability can be achieved, as compared to either technique alone. The optimal configuration depends on the number of
cores per processor, number of processors and memory per core.
Projection Mask-Less Lithography (PML2) is a potentially cost-effective multi electron-beam
solution for the 22 nm half-pitch node and beyond. PML2 is targeted on using hundreds of
thousands of individually addressable electron-beams working in parallel, thereby pushing
the potential throughput into the wafers per hour regime. With resolution potential of < 10
nm, PML2 is designed to meet the requirements of several upcoming tool generations.
Smaller feature sizes and aggressive Reticle Enhancement Techniques have led to greatly increased
mask data file sizes, longer processing times, and shrinking error budgets. Improvements to Mask
Data Preparation software can mitigate these trends. Processing time can be reduced by using
algorithms which are compatible with scalable multi-core Distributed Processing. Increased pattern
uniformity in the fractured output can reduce Critical Dimension variation on the finished mask
plate. Procedures for estimating pattern uniformity and CD variation are described.
Current and future Mask Data Preparation continues to see larger file sizes and longer processing times. Distributed
processing using multiple processors provides more compute power, but file Input/Output time remains a significant
portion of MDP processing. Data compression, fast disk storage, and fast network hardware are shown to provide some
benefit, but are not sufficient for unlimited scalability. Most MDP file formats store pattern data in a single disk file,
which creates a performance obstacle in the process flow. Dividing data into multiple files is shown to improve writing
speed, and to facilitate pipelined execution of multistage process flows. The advantages, disadvantages, and system
management of distributed files in the terabyte era are described.
As the industry moves to 45nm and beyond, rapidly increasing file sizes are an obstacle to achieving
fast turnaround time for mask manufacturing. Conventional Mask Data Preparation (MDP) requires
the production of large files, in a format specific to each make and model of E-beam tool. An
alternative approach extracts the data from a data file already present in the MDP flow, and provides
it directly to the E-beam tool. This extraction is called a "Data Exploder", because the output data
volume can be larger than the input data. Exploding the data in real-time saves the time required to
write and then read large disk files. The Data Exploder is compatible with multithread and
multiprocess parallel reading. The practical application and limitations of the Data Exploder are
described, including throughput performance, requirements for disk storage, network interconnect, and CPU configurations.
Optical Proximity Correction improves wafer image fidelity by combining small correction shapes with the original pattern data. Although these small shapes improve the exposure of the wafer image, the increase in total figure count results in longer fracture processing and E-beam writing time to create the mask. In this paper we describe alternative OPC treatment for jogs on non-Manhattan features, which reduce the additional figures produced, and make the data friendlier to the fracture and mask fabrication phases. Illustrations of example pattern data and improvement results in terms of figure counts are described.
Optical Proximity Correction (OPC) improves image fidelity by adding and subtracting small enhancement shapes from the original pattern data. Although the presence of these small shapes improves the final wafer image quality, it causes an increase in total figure count, longer fracture processing time, and the introduction of sliver figures. These undesirable artifacts can have a negative impact on the mask write time and mask image quality. In this paper we outline alternative OPC treatments which reduce the additional figures produced, and make the layout configurations friendlier to the subsequent mask fabrication phase. These include the alignment of neighboring small shapes during the OPC operation, and the preservation of jog alignment during the biasing phase. Illustrations of example pattern data, and improvement results in terms of figure counts are described.