Characterization of compositional heterogeneity with infrared spectroscopy and Evaluation of its effect on line-edge-roughness in chemically amplified photoresist polymer thin films


Shuhui Kang

Polymers Division, MSEL

100 Bureau Dr., Stop 8541

Bldg 224, Rm. A327

(301)-975 4602 (Phone)

(301) 975-3928 (Fax)


Mentor: Vivek M. Prabhu

Sigma Xi Member:  No

Category:  Materials




            Lithographic imaging for the fabrication of microelectronic devices is enabled by chemically amplified polymeric photoresists. In this process, an optical image is converted into a chemical latent image through a thermally activated reaction-diffusion process involving a photoinitiated acid catalyst.  The drive to fabricate ever smaller feature dimensions has led to increasingly stringent requirements on the photoresist.  Line edge roughness (LER), a metric of the feature quality is a major limiting factor for sub-50 nm photolithography.  The optical image quality of the exposure, distribution of the subsequently generated photoacid catalysts, and compositional heterogeneities are key factors to LER.

            In this study, we demonstrate a general approach to characterize compositional heterogeneity in polymer thin films using Fourier transform infrared (FTIR) spectroscopy.   Polymer films with varying degrees of heterogeneity were prepared using a model chemically amplified photoresist where a photoacid catalyzed reaction-diffusion process results in the formation of methacrylic acid rich domains.  Within these domains, the carboxylic acid groups dimerize through hydrogen bonding.  FTIR measurements of the relative fraction of hydrogen-bonded versus free carboxylic groups are used to quantify the degree of compositional heterogeneity. A solid sphere model was developed to describe the size of the heterogeneity as a function of photoacid concentration, reaction conditions, and initial copolymer composition.

            In addition, computer simulation was developed to understand the three-dimensional statistical fluctuations that occur in the vicinity of the feature line edge. These simulation data provide an explanation for the large statistical variation in LER observed at low deprotection gradients by photolithography methods.