Joachim Heinitz

Optimization of MSB for future technology nodes

In the ITRS roadmap [1] increasingly long mask write and cycle time is explicitly addressed as a difficult challenge in mask fabrication for the 16nm technology node and beyond. Write time reduction demands have to be seen in relation to corresponding performance parameters like Line Width Roughness (LWR), resolution, placement as well as CD Uniformity. The previously presented Multi Shaped Beam (MSB) approach [2, 3] is considered a potential solution for high throughput mask write application. In order to fully adapt the MSB concept to future industrys requirements specific optimizations are planned. The key element for achieving write time reduction is a higher probe current at the target, which can be obtained by increasing the number of beamlets as well as applying a higher current density. In the present paper the approach of a 256 beamlet MSB design will be discussed. For a given image field size along with a beamlet number increase both beamlet pitch and size have to be optimized. Out of previous investigations, one finding was that by changing the demagnification after the beam forming section of the MSB column the overall performance can be optimized. Based on first electron-optical simulations for a new final lens a larger demagnification turned out to be advantageous. Stochastic beam blur simulation results for the MSB reduction optics will be presented. During the exposure of a pattern layout the number of used beams, their shape and their distribution within the image field varies, which can lead to space charge distortion effects. In regard to this MSB simulation results obtained for an image field of approximately 10x10ìm² will be presented. For the 256 beamlet MSB design and resist sensitivities of 20μC/cm2, 40μC/cm2 and 100μC/cm2 write time and LWR simulations have been performed. For MSB pattern data fracturing an optimized algorithm has been used, which increased the beamlet utilization factor (indicates the mean number of beamlets which are used per multi-shot). Finally an update with regard to the required changes of the data path architecture for the 256 beamlet MSB approach will be given. Data integrity as an important aspect of the production worthiness of such a systems will be discussed specifically.