Analytical and experimental evaluation of a counting method for particles added during the mask handling process

Abstract

Extreme ultraviolet lithography has emerged as a highly promising next-generation lithography technique. However, some practical issues around mask technologies remain unresolved. Typically, the technique requires counting the particles released from or added to a mask blank (achieved using the knowledge of their position), a process referred to as an increase/decrease inspection. The counting process of this inspection has not been sufficiently studied because of the complexities of the counting error caused by counting losses and noise. However, the analysis of the counting process is essential to improve the counting accuracy. The author modeled the counting process by considering the difference between the actual occurrence of adhesion (or release) of the particles and the detection of their adhesion (or release) and verified the model experimentally. The author determined the necessary conditions for achieving the required accuracy by the multiple-inspection method, in which particles detected multiple times are identified as actual particles. The author found the counting errors in the increase/decrease inspection by the multiple-inspection method can be expressed as a function of the following: the numbers of unchanged, added, and released particles; the single-capture rate; and the number of inspections. Moreover, we found that the multiple-inspection method with four inspections is more accurate (pseudoparticles Np<10-4, counting error Nerr=0.065, standard deviation σ = 0.26) than the single-inspection approach (Np=10, Nerr=3.3, σ = 1.7). The author confirmed that the multiple-inspection method reliably determines whether particles were released or adhered using positional information.