SHAPED E-BEAM NANOPATTERNING WITH PROXIMITY EFFECT CORRECTION

Size: px

Start display at page:

Download "SHAPED E-BEAM NANOPATTERNING WITH PROXIMITY EFFECT CORRECTION"

Rosalyn Hall
5 years ago
Views:

1 SHAPED E-BEAM NANOPATTERNING WITH PROXIMITY EFFECT CORRECTION Michal URBANEK a, Vladimir KOLARIK a, Milan MATEJKA a, Frantisek MATEJKA a, Jan BOK a, Petr MIKSIK b, Jan VASINA b a) ISI ASCR,v.v.i., Kralovopolska 147, Brno, Czech Republic, urbanek@isibrno.cz b) ELTEK spol. s.r.o., Tomeckova 4298, Kromeriz, Czech Republic Abstract: Electron beam writer is a tool for writing patterns into a sensitive material (resist) in a high resolution. During the patterning, areas adjacent to the beam incidence point are exposed due to electron scattering effects in solid state (resist and the substrate). Consequently, this phenomenon, also called proximity effect, causes that the exposed pattern can be broader in comparison to the designed. In this contribution we present a software for proximity effect simulation and a software for proximity effect correction (PEC). The software is based on the model using the density of absorbed energy in resist layer and the model of resist development process. A simulation of proximity effect was carried out on binary lithography patterns, and consequently testing patterns were exposed with a corrected dose. As pattern generation, we used the e-beam writer TESLA BS 600 working with fixed energy 15keV and variable size rectangular shaped beam. The simulations of binary testing patterns and exposed patterns without PEC were compared. Finally, we compared the testing structures with PEC and without PEC, and we showed that the PEC tool works reliably for the e-beam writer BS Keywords: e-beam writer, shaped beam, proximity effect correction 1. INTRODUCTION Electron beam lithography is a technique for creating extremely fine patterns in resolution below 10 nm nowadays. This technique derived from scanning electron microscopy is required by the modern electronics industry for integrated circuit preparation [1]. The pattern writing in a high resolution is carried out into layer of sensitive material called electron resist (almost polymer), which is deposited on substrate surface. During the writing, several effects occur, which can significantly change the desired resolution of designed patterns [2]. The final pattern is most affected by two scattering effects in resist layer and substrate: forward scattering and backscattering. Due to this electron scattering effects, the exposed patterns can be significantly broader than the designed. This phenomenon, called proximity effect, causes that areas adjacent to the exposed receive nonzero electron dose and thus pattern can vary from the intended size. 1.1 Model of the absorbed energy Proximity effect correction requires knowledge of the energy density absorbed in electron resist layer during the exposure by electron beam. The proximity effect can be described by the two Gaussian functions model see (1)[3]. This model represents the distribution of energy absorbed in the material in distance r from the incidence point of primary electrons, where is the forward scattering range, is the backscattering parameter and label the ratio between the energy absorbed from the backscattered electrons and primary electrons.

f ( r) 2 2 1 1 r r exp exp 2 2 2 1 2 (1) We use for pattern generation a rectangular shaped beam, then we can calculate distribution function as numeric integration of individual pixels in stamp.

We calculate the distribution function as sum or difference of pre-calculated stamps saved in library and it prevents from calculating a big number of Gaussian function, this leads to a higher speed

2 f ( r) r r exp exp (1) We use for pattern generation a rectangular shaped beam, then we can calculate distribution function as numeric integration of individual pixels in stamp. It means that stamp consists of number a*b beams with cross section of one pixel and the distribution function can be generated for individual stamps of different sizes as following. We calculate the distribution function as sum or difference of pre-calculated stamps saved in library and it prevents from calculating a big number of Gaussian function, this leads to a higher speed of calculation. 1.2 Sensitivity curve Electron beam resist as transfer and recording medium is characterized by its sensitivity curve. It means if we expose resist to a range of doses and then develop pattern, plotting the relative thickness of resist versus dose we obtain the sensitivity curve of resist. The sensitivity of resist is defined as the value of dose, when all resist is removed (see fig 1). Fig.1: Sensitivity curve of PMMA 950k resist, thickness of 200 nm. 2. SIMULATION AND CORRECTION ALGORITHMS 2.1 Proximity Effect Simulator (PES) The previous words obviously show that the proximity effect correction is for creating patterns in high resolution at the most desired. The first step for its correction is to understand its influence on the designed patterns and visualize the latent pattern in electron resist layer, it means visualization the density of absorbed energy. Consequently, the PES for dose distribution was created. Input parameters for PES are variables from equation (1). Values of those parameters are known from the previous papers [3,4] and follows: α = 0.20, β = 1.20 and η = The PES allows simulation of dose distribution of designed pattern (Fig. 2 right)

Fig.2: Test pattern layout with designed widths of central lines and exposure of adjacent areas (left), proximity effect influence on

2 Development simulator (DeS) If we know the resist sensitivity curve (Fig.1) we can simulate the resist development process as well (Fig.3).

It means that the exposed areas after development process are exposed to the substrate and unexposed areas are fully covered by original

3 Fig.2: Test pattern layout with designed widths of central lines and exposure of adjacent areas (left), proximity effect influence on absorbed dose (nonzero dose surrounding the testing pattern right). 2.2 Development simulator (DeS) If we know the resist sensitivity curve (Fig.1) we can simulate the resist development process as well (Fig.3). E-beam patterning almost works in binary regime of lithography during the production of electronic devices. It means that the exposed areas after development process are exposed to the substrate and unexposed areas are fully covered by original resist layer, and therefore we realized simulation and test patterning for binary mode. Fig. 3: Development simulation in binary mode (detail of pattern with nominal linewidth 200 nm): left- without PEC correction, right with PEC correction (dose correction).

2.3 Proximity Effect Corrector (PEC) algorithms After the simulation of dose distribution and development process, the second step leading to proximity effect correction.

4 2.3 Proximity Effect Corrector (PEC) algorithms After the simulation of dose distribution and development process, the second step leading to proximity effect correction. Our correction method is primarily based on implementation following algorithms (Fig. 4): 1) Dose Correction exposure time is converted to designed pattern is generated over whole resist thickness, adjacent areas receive minimized dose 2) Edge - additional stamps are exposed along the pattern edges, adjacent areas receive minimized dose. 3) Corner - additional stamps are exposed in pattern corners, adjacent areas receive minimized dose. Fig.4: Detail of testing pattern with nominal width of central line 200 nm after PEC a) dose correction, b) edge correction, c) corner correction. 2.4 PEC execute After PEC application on designed testing pattern and simulation of development process (DeS) we can compare supposed result of the testing pattern without and with PEC. The linewidths of nominal width 200 nm after development process simulation before and after PEC (for surrounding exposure 0%) are compared. As we expected, linewidth without PEC with growing surrounding exposure expands, whereas the linewidth with PEC is constant (Fig. 5). Differences between simulated and nominal value of linewidths versus nominal linewidth without and with PEC is shown in figure 6 and it is apparent that the widths of lines for nominal linewidth below 400 nm are lower and above 500 nm are broader than designed. The linewidths with PEC correspond with designed for nominal linewidth from 200 nm.

5 Fig. 5: Simulated linewidth of 200 nm line Fig 6: Differences between simulated and nominal value of linewidths versus nominal linewidth. 3. TESTING PATTERN EXPOSURE We used e-beam writer BS600 with a rectangular shaped beam and fixed accelerating voltage of 15 kv [5] for testing pattern exposure (Fig. 2 - left). The exposure dose of 100% was 30C/cm 2. Testing pattern was exposed into PMMA 950k resist, thickness of 200 nm, used developer n-aac, development time 180 s and temperature 25 C, contrast of resist 2,1. After the development, we measured the width of central lines for all values of exposed adjacent to the lines (0, 25, 50, 75, 100 %) by AFM (Fig.7). The measured width of 200 nm line without PEC for varying surrounding values is lower than value from the simulation, because the pattern is not fully developed, but in agreement with simulated estimation. The width of corrected line fully corresponds with designed pattern. Fig. 7: AFM image of testing pattern nominalwidth of central line 200 nm, left - without PEC correction, right - with PEC correction.

6 Fig. 8: AFM measurement - linewidth of 200nm line. Fig. 9: Differences between simulated and nominal value of linewidths versus nominal linewidth. 4. CONCLUSIONS The exposure correction of central line becomes significant, where noticeable difference is between corrected and uncorrected line width. Line widths without correction are very noisy, but widths of lines with correction correspond with designed width for lines from 200 nm very well. As it is evident from the AFM measurements, we managed to create a tool for simulation and offline compensation of electron scattering phenomena for binary patterns generation using e-beam writer with a shaped beam. ACKNOWLEDGMENT This work was partially supported by the MIT CR under the contract No. FR TI1/576, the EC and MEYS CR (project No. CZ.1.05/2.1.00/ ALISI), the TACR project No. TE and by the institutional support RVO: REFERENCES [1] CHOUDHURY, R. Handbook of Microlithography, Micromachining and Microfabrication. Volume 1: Microlithography, SPIE PRESS and IEE, 1997, ISBN [2] KOLARIK, V. et al. Proximity effect simulation for variable shape E-beam writer. In proceedings of13th Recent Trends in Charged Particle Optics and Surface Physics Instrumentation, Skalský Dvůr, Czech Republic, 2012, p.75-76, ISBN [3] KRAATS, A. van der, RAUGHANTAH, M. Proximity effect in E-beam lithography, available from p. 1-5., cited [4] URBANEK, M. et al Determination of proximity effect forward scattering range parameter in E-beam lithography. In proceedings of12th Recent Trends in Charged Particle Optics and Surface Physics Instrumentation, Skalský Dvůr, Czech Republic, 2010, p.67-68, ISBN [5] KOLARIK, V et al., Writing system with shaped electron beam, Jemná mechanika a optika, Vol. 53, No.1, 2008, p , ISSN [6] OWEN, G., Methods for proximity effect correction in electron lithography, Journal of Vacuum Science Technology B, 8(6), 1990, p

COMPARISON OF ULTIMATE RESOLUTION ACHIEVED BY E-BEAM WRITERS WITH SHAPED BEAM AND WITH GAUSSIAN BEAM

COMPARISON OF ULTIMATE RESOLUTION ACHIEVED BY E-BEAM WRITERS WITH SHAPED BEAM AND WITH GAUSSIAN BEAM Stanislav KRÁTKÝ a, Vladimír KOLAŘÍK a, Milan MATĚJKA a, Michal URBÁNEK a, Miroslav HORÁČEK a, Jana