Author:

Keywords:

EUV photoresist, DRM, dissolution, photopolymer

Abstract:

EUV lithography is proposed as the next technology to be used in microelectronics production for 10nm device technology node and below. In order to make this technology viable for the industry, research is required to achieve required patterning performances of EUV photoresists, improving simultaneously resolution, sensitivity and roughness [1]. The working principle of photoresists is that light exposure (92.5eV photon for EUV) induces a switch in solubility of the polymer matrix. For traditional resist, a photocatalized reaction results in the removal of hydrophobic functional groups, rendering the exposed polymer soluble in a water-based developer. A lot of focus has been put in understanding the exposure step [2] and the deprotection reactions [3], but the development step has retained less attention, even though it impacts significantly roughness and resolution. Various models, from simplistic (Ferguson model, Mack model) [4,5] to more complex (Notch, Houle)[6,7], describe the phenomenon through a relation between the concentration of functional group and a corresponding dissolution rate. For our work, we have used a dissolution rate monitor (DRM) tool. It is based on interferometry to measure the thickness of the resist with a high sampling frequency. A continuous measurement of the thickness is performed while the resist is put in contact with a constant flow of developer. As such, the thickness loss vs time (dissolution curve) of a resist exposed at various EUV doses can be compared. In this work, we compared the dissolution curves of various EUV resists : three positive tone chemically amplified resist (water-based solvent), a negative tone chemically amplified resist (organic solvent) and a non-chemically amplified resist based on a metal-oxide thin film. Contrary to our expectations, we observed that most of the EUV resists show a non-linear dissolution curve. This non-linearity implies that a specific concentration of functional group does not correspond to a fixed dissolution rate, but that this rate is time dependent. This experimental result suggests that the current models (Ferguson, Mack, Notch) used for predicting resist profiles after development should be modified to take into account this nonlinearity, and that we have the experimental data available to tackle this challenge. Our ultimate goal is to correlate the dissolution characteristics with the patterning performances of the resists (resolution, LWR), as we observed that the dissolution rates show important differences among resist types (minimum and maximum value, contrast). [1] Erdmann, A. , Fühner, T. , Evanschitzky, P. , Agudelo, V. , Freund, C. , Michalak, P. , and Xu, D. Microelectron. Eng. 132 (2015), 21–34. [2] Biafore, J. J., Smith, M. D., Setten, E. van, Wallow, T., Naulleau, P., Blankenship, D., Robertson, S. A., and Deng, Y., Proc. of SPIE 7636 (2010), 76360R [3] Kozawa, T., Santillan, J. J. and Itani, T., Jap. J. of Appl. Phys. 54-3 (2015), 036507 [4] Ferguson, R.A., Memorandum No. UCB/ERL M91/78 (1991) Chap 2. 18-20 [5] Arthur, G., Mack, C.A., Eilbeck, N. and Martin, B., Microelectron. Eng. 41 (1998) 311-314 [6] Mack, C. A., and Arthur, G., Electrochem. Solid-st. 1(2) (1998), 86–87. [7] Houle, F. A., Hinsberg, W. D. and Sanchez, M. I., Macromol. 35(22), (2002) 8591–8600