Pre-exposure Bake Research Articles

Controlled-width track in through silicon via using 3D holographic photolithography with modified electrodepositable photoresist

We present a novel lithographic process for patterning controlled-width tracks onto anisotropically micromachined silicon. The technique is based on the use of computer-generated holographic masks with a custom alignment and exposure tool. Experimental and simulation results are presented. 3D holographic photolithography significantly reduces the problem normally present with photolithography on non-planar surfaces—namely diffractive line broadening. A negative-acting electrodepositable photoresist (InterVia 3D-N) is used in the study. Its deposition onto the 3D substrate is optimized by modification of coating temperature and thickness and of pre-exposure bake conditions. We show the successful patterning of a constant-width 8 µm line down the sloping sidewall of a 500 µm thick silicon wafer. This is beyond the conventional resolution limit and indicates the potential of the technique for realizing high-density vertical routing in electronic packages and MEMS.

An optimal high contrast e-beam lithography process for the patterning of dense fin networks

There are many difficulties to overcome towards the integration of 10 nm CMOS technology. One such major challenge is to keep a tight control of the leakage current of devices while increasing the current drive at a reduced supply voltage. In this context, multi-gated structures, which are used to control the transport in ultra-thin channel (e.g. FinFET), are a promising solution. A critical step during the fabrication process of a FinFET is the patterning of dense, high aspect ratio fins. High demand is therefore placed on e-beam lithography techniques to obtain narrow, sharp, densely packed resist lines. This paper presents a detailed study on the optimum e-beam exposure process using a negative tone e-beam resist, namely Hydrogen Silsesquioxane (HSQ). The impact of the pre-exposure bake temperature, of the Tetramethyl Ammonium Hydroxide (TMAH) concentration in development solution and of development time has been investigated. The standard process uses 2.38% TMAH as a developer, samples being pre-baked on a hotplate at a temperature between 150 and 220 °C for 2 min. By using a lower pre-bake temperature of 90 °C and a more concentrated TMAH solution dosed at 25%, a seven-fold improvement of contrast can be obtained in terms of contrast values. Cross sectional SEM views show fin networks with a pitch ranging from 40 nm to 200 nm. The line profiles are steep and an excellent uniformity is obtained across the whole network, even for lines located at the edge. Dense patterns are presented with lines as narrow as 15 nm and with a 25 nm space.

Surface roughness of hydrogen silsesquioxane as a negative tone electron beam resist

The surface roughness of hydrogen silsesquioxane (HSQ) as a negative tone electron beam (EB) resist has been studied as a function of exposure dose, developer concentration, and pre-exposure bake temperature. In the range of low exposure doses, the surface roughness of HSQ drops rapidly with an increase in the dose. It levels off at about 1 nm around the dose at which the respective contrast curve saturates. The TMAH developer concentration, varied in the range of 2.5–25%, has only a slight influence on the HSQ roughness. The increase of the baking temperature from 90 to 220 °C, however, causes a significant rise in the roughness of more than 100%. These results are discussed in terms of structural changes and network formation in thin HSQ films. They prove that a process combining high developer concentration and low baking temperature is preferable for fabrication of high resolution patterns using HSQ as a negative tone EB resist.

Ultrasonic sensor for photoresist process monitoring

An ultrasonic sensor has been developed to monitor photoresist processing in situ, during semiconductor manufacturing. Photoresist development, pre-exposure bake, and post-exposure bake were monitored for the Shipley 1800 series I-line resists, and the pre-exposure bake of Shipley APEX-E deep-uv (DUV) resist was monitored as well. Development monitoring was achieved by measuring thickness changes in the resist as it was removed. Data regarding dependence of development rate on exposure dose was obtained for the I-line resist with exposure doses varying from 20 to 68 mJ/cm/sup 2/. Measurements showed an increase in average development rate from 0.04 to 0.155 /spl mu/m/s, with the rate leveling off at around 55 mJ/cm/sup 2/. Pre-exposure bake monitoring results demonstrated the ability of the sensor to measure the glass transition temperature of the resist film during prebake as well as the ability to invert out the elastic constants of the film using reflection theory. The glass transition temperature (T/sub g/) is an important parameter in both the pre- and post-exposure bakes and therefore could be useful in monitoring these processes. Results of pre-exposure bake T/sub g/ measurements are presented for both-I-line and DUV resists. The glass transition temperature during prebake was found to be higher for the DUV resist than for the I-line series. The I-line resist post-exposure bake measurement of glass transition temperature confirmed the reported T/sub g/ of 118/spl deg/C for the I-line novolac resin. The multiple uses of this sensor make it suitable for integration into a manufacturing setting.

Processing control for 0.25 μm x-ray exposures of commercially available resists: The potential for adaptive control

The processing of chemically amplified resists to yield 0.25 μm structures requires control of a number of key parameters. Some of these may be at the limits of the control range or at the limits of achievable uniformity. Where it may not be possible to improve the control or the uniformity, it may be possible to use adaptive control to compensate. The focus of this article is to explore what is needed for adaptive control with two commercial x-ray resists, Shipley SAL 605 (negative tone) and Hoechst AZ PF-514 (positive tone) and to explore the feasibility of adaptive control with the SAL 605 resist. Statistical optimization has been used previously to provide robust process latitude by focusing on the preexposure bake temperature and time, the x-ray dose, and the postexposure bake temperature and time. To determine the required control and the slope of the linewidth change with the change in each variable, a statistical model of the response surface was constructed. The statistical results and new work with developer normality point out the major contributions from the wafer baking conditions using a track vacuum hot plate. The different slopes of the variations with linewidth suggest that whole wafer compensation may be possible for certain parameters, and that field-to-field variation arising from hot plate uniformity might be compensated for the SAL 605 resist with another. Preliminary results with Fourier transform infrared spectroscopy will be presented that support these approaches and suggest that further efforts to provide in-line measurements might be justified by correlating the infrared cross-linking signal with the processing parameter that yields the correct linewidth.

Role of residual casting solvent in determining the lithographic and dissolution behavior of poly(methylmethacrylate)

The effect of the pre-exposure bake and the choice of casting solvent on the sensitivity and contrast of poly (methylmethacrylate) (PMMA) has been documented to an extent not previously reported in the literature. Films baked at temperatures below Tg (the temperature which marks the onset of long-range, coordinated molecular motion), exhibited improved sensitivity and poorer contrast when compared to those baked above Tg. Dissolution rates and relative concentrations of casting solvents remaining in the films were also measured. Cross-correlation of these results with energy deposition calculations indicate that the pre-bake temperature more strongly correlates with the observed improvement in sensitivity than the presence of residual casting solvent.

Application of real time infrared spectroscopy to monitoring the kinetics of chemically amplified resists

Currently, the major approach to improving the performance of chemically amplified resists is by means of statistical design methods. These methods do not make use of information about the underlying chemical processes, information which could be useful in expediting the optimization process. We have developed techniques and equipment to monitor changes in resist chemistry during the bake steps using in situ using real time Fourier transform infrared spectrometry (RT-FTIR). Using the Shipley SAL 605 negative chemically amplified resist, we monitored exposed resist as it was being baked at 110 °C; we were thus able to see several peaks change, including the growth of a peak at 982 cm−1 which we associate with the formation of an ether linkage during cross-linking. When the height of this peak is plotted over time from the start of the bake, it shows several things, among them (a) the reaction being monitored takes much longer to reach its final level of completion (≳300 s) than the statistically derived optimum (∼60 s), (b) as expected, the reaction rate is a function of dose, and (c) the results are reproducible between identically treated wafers. We speculate that the reason we see peak growth long after the resist is cross-linked enough for processing is due to the multiple available sites on the cross-linker molecule which allow continued bond formation after enough have formed to make the resist sufficiently insoluble. With chemically amplified resists there has been concern over residual solvent which can interfere with the catalytic mechanism of the photogenerated acid. The same RT-FTIR techniques were also shown to be effective in monitoring the loss of solvent from the resist during the preexposure bake, giving the time at which the solvent can be assumed to have been removed—in the case of Shipley SAL 605, approx 30 s. Finally, the RT-FTIR technique can be used to follow the individual contributions of the components of the resist formulation as their concentrations are changed to affect different resist properties. The monitoring is directly on the resist-coated wafer, and special handling techniques, to be described in the text, are necessary to make these measurements directly applicable to resist processing.