E-beam Proximity Effect Correction Research Articles

Temperature dependent effective process blur and its impact on exposure latitude and lithographic targets using e-beam simulation and proximity effect correction

It is well known that cold development yields higher contrast and improved exposure latitude particularly for ZEP520 from Zeon Chemicals. In this paper, the authors quantify the effective process blur as a function of temperature. The effective process blur for our development process conditions were found to be 10, 42, and 71 nm for developer temperatures at −12, 21, and 30 °C, respectively. Knowledge of how to tune the process blur can be used in a unique application. Instead of using the best possible process blur, exposure latitude is traded for improved exposure time. Optimizing the e-beam exposure time is always desired while maintaining a target critical dimension and desired shape at the wafer. In particular, the exposure time can be dominated by shape overhead delays stemming from the over digitization of curved shapes within a pattern. As such, it is better to expose a pattern with the least number of shapes as possible while obtaining the desired shape at the wafer. The authors demonstrate how e-beam simulation can be used to determine the optimal effective process blur to obtain a target desired shape while minimizing the fractured shape count to ultimately reduce overall exposure time.

Easy to adapt electron beam proximity effect correction parameter calibration based on visual inspection of a “Best Dose Sensor”

The quality of e-beam proximity effect correction depends on the quality of the proximity effect model parameters, which are not accessible to direct measurement. Monte Carlo simulation is capable of determining the electron scattering coefficients, but does not include the process induced effects such as resist blur. Therefore, various experimental methods have been suggested (Stevensat et al. (1986) [1]; Rishton and Kern (1987) [2]; Hudek (2006) [3]). Most are either highly labor intensive due to a large required number of critical dimension measurements, or simple in the experimental evaluation, but limited in accuracy (Babin and Svintsov (1992) [4]), since resist effects interfere with the evaluation criterion.This paper presents an easy to adapt experimental calibration method based on visual inspection of a “Best Dose Sensor”, and its application to calibrate long- and mid-range effects.

Implementation of E-Beam Proximity Effect Correction using linear programming techniques for the fabrication of asymmetric bow-tie antennas

Asymmetric nano-bow-tie antennas offer the possibility of direct light-to-electrical energy conversion. These nano-antennas are easily integrated with Metal-Insulator-Metal (MIM) tunnel junctions in between the antenna segments for the purpose of coupled signal rectification. The architecture of the tunnel junction together with the antenna size precision require nano-scale patterning accuracy. Electron Beam Lithography (EBL) is used for patterning purposes. In this paper Proximity Effect (PE), a very common resolution problem in EBL, is reduced by a dose modulation technique employing linear programming (LP) algorithms. Production of tightly controlled antenna segment dimensions is achieved in conjunction with a small area tip and a small tunnel junction gap. It is expected that precise control of the gap geometry will enhance detection speed, enabling the utilization of the device for the visible range energy harvest purposes.

Automatic determination of spatial dose distribution for improved accuracy in e-beam proximity effect correction

Proximity effect correction in e-beam lithography is expected to be an essential step in the fabrication of high-density fine-feature circuits in the future. In the past, successful proximity correction by a shape modification approach with a single dose for the entire circuit pattern has been demonstrated. This approach has a few advantages over dose modification, including the fact that it is compatible with future multiple-beam or projection-based systems. As we continue to reduce the minimum feature size, accuracy of correction becomes more critical. In order to improve correction accuracy of the shape-only modification, a hybrid approach allowing region-wise dose control has been proposed. In this paper, a practical fast scheme for determining spatial dose distribution for the hybrid correction scheme is presented.

Application of neural network to enhancing accuracy of E-beam proximity effect correction

In most cases, proximity effect correction in E-beam lithography is inherently a sequential process of correcting one circuit primitive at a time. Consequently, the phenomenon of “recursive” effect may lead to a degradation in correction accuracy. A well-trained neural network can reduce this problem of the recursive effect by correcting circuit elements in a group. We have designed a neural network based proximity corection scheme in order to overcome the recursive nature of sequential correction and enhance correction accuracy, especially in circuits with high density and fine feature sizes.

A New Approach of E-beam Proximity Effect Correction for High-Resolution Applications

The e-beam proximity effect is well known as one of the limiting factors in e-beam lithography. As features get smaller the need for e-beam proximity effect correction (EPC) increases. There exist different approaches to cover these effects by varying dose or shape of the pattern layout during the exposure step. Both of these basic approaches have drawbacks limiting their application. For example, the EPC correction by shape variation underlies constraints such as neighbouring features, the exposure grid of the e-beam tool and writing time. EPC by dose variation fails in cases the feature size is very close to the process dependent forward scattering parameter alpha. To guarantee CD uniformity in these cases a negative dose assignment would be necessary, which is practically impossible. The paper presents a new correction scheme based on dose assignment and geometry variation at the same time. After a short introduction the theoretical description of the method is given. Dose and geometry modifications are calculated using a Fast Fourier Algorithm. Due to the ability of changing the geometry depending on feature size and feature neighbourhood even very small features can be printed correctly. This flexibility is the main advantage compared to already existing methods. Pattern fidelity and linearity can be improved drastically. Resolution limits can be pushed further down. The application of the automatic correction to high resolution features with a feature size of 100 nm are demonstrated. The CD linearity is investigated and demonstrated using a real device patterns.

Electron beam lithography process for advanced optical masks

The requirements for advanced optical masks have been tightened according to the accelerated lithography roadmap for semiconductor manufacturing. The accuracy of the optical mask may be affected by the pattern exposure tools, the materials, and the processes. Higher acceleration voltage e-beam pattern exposure was evaluated with a chemically amplified resist. With the higher acceleration voltage e-beam system, proximity effect correction is needed to control the critical dimension (CD). CD linearity is maintained down to 400 nm for different kinds of patterns by adopting the 50 kV e-beam system and proximity effect correction. The results show that the method is an attractive candidate for fabrication of next generation optical masks. Dry etching is effective in achieving small feature sizes, such as optical proximity correction (OPC), with high pattern fidelity. Loading effects were evaluated, and inductively coupled plasma dry etching was found to be stable against those effects. Fabrication of the OPC pattern designed for a 0.15 μm rule device was demonstrated with these methods.

Advanced application of hierarchy reorganisation in E-beam lithography

The E-beam proximity effect correction (PEC) of very large layouts is still regarded a very difficult step during data preparation. The CPU-time for today's integrated circuits extends into weeks, file sizes into Gigabytes. Recently, Sigma-C achieved a breakthrough in handling memory chips by developing a general approach for hierarchical data preparation. A very important achievement is the hierarchy reorganisation, which enables the correction of chips like the 256 Mb. In this paper the further development, needed to extend the theory to logic devices, to innovate the PEC of random layouts, will be presented. First results of the correction of logics will be demonstrated in comparison to flat processing.

Diffusion in AZ-5214 image reversal process and its application to e-beam proximity effect correction

Here we report our discovery of the diffusion phenomenon of AZ5214 resist during a post-exposure bake process step and its application to improve the electron beam lithographic performance of the resist. The diffusion phenomena are that the exposed features gain an extra amount of material from their immediate neighborhoods and become thick at the pattern edge and wider in feature size. In general, variations in pattern dimensions are unfavorable for process control. However, in this case, the changes in the exposed geometry are just opposite to those caused by the electron beam proximity effect. Therefore, by carefully controlling the parameters of the post-exposure bake, excellent proximity effect corrections have been obtained. The advantage of this method is its simplicity. There is no extra process step and complicated computation involved as in other proximity effect correction methods. It is achieved by using a rather simple step, a post-exposure bake, in the image reversal process.