Carbon rich hard mask underlayer (UL) material deposition has become inevitable process in all advanced lithography applications. UL processes which include chemical vapor deposition (CVD) and spin-on UL play a very important role for pattern transfer from patterned thin photoresist to the substrate. UL materials must satisfy several requirements, which have become more demanding with device shrinkage and increasing device complexity (FinFET, 3D integration). The most important properties of next generation UL materials are superior wiggle resistance, etch controllability, thermal resistance, planarization, and gap filling performance. In particular, planarization and gap fill properties of UL material for application on topo-patterned substrate are receiving much attention recently. CVD processes generally give better wiggle performance and thermal resistance, but poorer planarization and gap filling performance than spin-on UL processes. In addition, Cost of Ownership (CoO) of CVD process is higher than that of a spin-on UL process. Therefore spin-on organic hard mask (OHM) process has been investigated as an attractive alternative to CVD processing. In this paper, we focus on an investigation of key properties of spin-on UL materials for achieving good planarity and gap filling performance on topo-patterned substrate. Various material properties such as solution viscosity, glass transition temperature (Tg), and film shrinkage ratio were evaluated and correlations between these properties and planarization were discussed.

Semiconductor manufacturing technology is currently undergoing a transformation from immersion photolithography to double patterning or EUV technology. The resultant resist dimensional size and height shrinks will require improved pattern transfer techniques and materials.Underlayer (UL) processes which include chemical vapor deposition (CVD) and spin-on application play a very important role in various chip manufacturing integration schemes. A pattern wiggling problem during substrate etch has arisen as a critical issue when pattern dimensions shrink. CVD processes have shown better pattern transfer performance than spin-on processes but at higher cost and process complexity along with difficulty in obtaining planarization and good gap fill. Thus spin-on process development has received increased attention recently as an attractive alternative to CVD processing.In this work we focus on elucidating the mechanism of UL wiggling and have synthesized materials that address several hypothesized mechanisms of failure: hydrogen content, modulus, film density, charge control unit type and thermal resistance. UL materials with high thermal resistance additionally provide the ability to expand the applicability of spin-on approaches. Material properties and wiggle failure test results will be discussed.

Chemical vapor deposition of gold from dimethyl-2,4-pentandionato gold (III) and two fluorinated derivatives is reported. At substrate temperatures of 200–300°C, high purity gold films were obtained on SiO2 and Si substrates. Films with resistivities down to 1.3 times that of bulk gold were obtained at deposition temperatures of ~200°C, where films with very small grain size could be deposited.

The laser-induced deposition of high-purity aluminum metal has been achieved by pyrolytic decomposition of trimethylamine aluminum hydride. The chemical structure of the precursor affords a high ambient vapor pressure which results in rapid rates of aluminum film formation. In addition, the precursor is nonpyrophoric, in contrast to other trialkylaluminum precursors. These combined chemical and physical properties make trimethylamine aluminum hydride an ideal precursor for laser-induced chemical vapor deposition of aluminun films.

The interaction of water with the photoresist film stack is proving to be a key factor in the current generation of 193-nm immersion lithography. Photoresist performance, CD control, optics lifetime, defectivity, overlay and possibly even tool throughput can all be affected by this interaction. Defect control has been an area of increasing concern as the source of the defects can be quite different than that found in conventional dry lithography [1]. Defects can originate from the UPW (Ultra Pure Water) either as particulates or as dissolved solids that precipitate from residual droplets left behind after scanning. Another source of defects can be particulates generated by the immersion fluid as it flows through the exposure tool or as a consequence of water contact with the resist film or resist/topcoat film stack. Recently there have been reports of printable defects due to stains or "watermarks" on the surface of the photoresist [2]. In this report we describe techniques for the visualization of watermarking and particulate formation on a variety of film surfaces. We also describe experiments testing the staining of a variety of water contaminants and additives and their effect on imaging performance. We will also describe the effect of different topcoats on imaging and defectivity in terms of their surface properties.

The ability to extend 193 nm lithography resolution depends on increasing the numerical aperture (NA) of the exposure system, resulting in smaller depth of focus, which subsequently requires use of thinner photoresists. Bottom antireflective coatings (BARCs) are a necessity, but the organic composition of current 193 nm BARCs offers poor etch selectivity to the photoresist. As a result, image transfer with thin resists is becoming increasingly difficult. It is also more challenging to control reflectivity at high numerical apertures with a thin, single layer BARC.To address these issues, IBM has developed a new class of silicon containing BARCs. These materials exhibit high etch selectivity that will significantly improve the performance of high NA 193 nm lithography. The incorporation of silicon in the backbone of the polymers comprising these BARCS affords a high etch selectivity to conventional organic resists and therefore these polymers can be used as thick planarizing BARCs. The optical constants of these BARCs have been tuned to provide good reflectivity control at NA > 1.2 These materials can also be used as part of a dual layer BARC scheme composed of the thin organosilicon based BARC coated over a planarizing organic underlayer. This scheme has also been optically tuned to provide reflectivity suppression at high incident angles. By utilizing a thick BARC, a novel contact hole shrink process is enabled that allows tapering of the sidewall angle and controlling the post-etch critical dimension (CD) bias. Structures of the silicon containing polymer, formulation chemistry, optical tunability, lithography at high NA and RIE pattern transfer are reported.

As the semiconductor industry moves to sub 193 run imaging wavelengths the switch to thin film resists with marginal transparency can result in a significant degradation of resist profiles. Tremendous progress has been made recently in lowering the absorbance of 157 run polymer systems, however etch resistant single layer resists are likely to have absorbance values of 1-2/mum. Current generation EUV resists have even higher absorbance values of 2-3/mum. While the use of thin films mitigates (but does not eliminate) the effect of high resist absorbance the impact of resist footing, notching, and standing waves are exacerbated. Surface effects such as top rounding or surface contamination are also likely to be a significant concern.One way to minimize the effect of non vertical profiles due to high absorbance and/or resist footing is through the use of optimized underlayer/photoacid generator (PAG) systems. The PAG's are selected so as to have higher diffusiveness but lower reactivities than the PAGs used in the resist itself in order to have a greater effect on the resist profile. Characterization data will be presented using a high absorbance resist (> 3.0/mm). Lithographic data on this system will be described and imaging data presented showing features down to 90 nm in a resist 120 rim thick.

As resist feature sizes have decreased and the performance demands on chemically amplified photoresists have increased the role of the photoacid generator (PAG) in determining overall resist performance has become increasingly apparent. Over the past 20 years a variety of different types of PAG's have been introduced as researchers have sought to optimize properties such as acid strength, acid volatility, diffusion length, wavelength response, solubility etc. PAGs that produce very strong organic acids are widely used, in part because of requirements for high photospeed resists. Most of these acid generators are based on perfluoroalkyl sulfonic acid based onium salts. In an effort to identify and characterize alternative PAGs we have investigated the performance of a variety of photoacid generators that are not based on sulfonic acids. In this report we will describe the relative reactivities of these PAGs under a variety of exposure wavelengths and processing conditions including acid diffusion proprieties and photospeed measurements.

Unexpectedly good UV transmittance at 157nm of poly(norbornene sulfone) bearing a pendant hexafluoroisopropanol functionality has prompted us to employ this fluoroalcohol as an acid group for the design of chemical amplification resists for use in 157nm lithography. The backbone structures to which the hexafluoroalcohol group is attached are polynorbornene and polystyrene. Furthermore, our discovery that poly(methyl α-trifluoromethylacrylate) is adequately transparent at 157nm has led us to incorporate the α-trifluoromethylacrylic unit in the polymer backbone by radical copolymerization with styrenes and norbornenes. Thus, four platforms are currently available to us in preparation of 157nm resist polymers; 1) all-acrylic, 2) all-norbornene, 3) acrylic-norbornene, and 4) acrylic-styrenic systems.

The impact of develop time and developer concentration was investigated for 193nm resists based on "alternating" polymers of malefic anhydride and norbomene monomers (COMA), prepared by free radical copolymerization. The COMA materials show significant and unique performance dependence on the development process. The development process for COMA materials was found to be a powerful process variable. This paper suggests an explanation for these findings.

The impact of airborne basic molecular contamination (MB) on the performance of chemically amplified (CA) resist systems has been a long standing problem. Low ppb levels of MB may be sufficient for robust 0.25 micrometer lithography with today's advanced CA resist systems combined with adequate chemical air filtration. However, with minimum CD targets heading below 150 nm, the introduction of new resist chemistries for Next Generation Lithography, and the trend towards thinner resists, the impact of MB at low and sub-ppb levels again becomes a critical manufacturing issue. Maximizing process control at aggressive feature sizes requires that the level of MB be maintained below a certain limit, which depends on such parameters as the sensitivity of the CA resist, the type of production tools, product mix, and process characteristics. Three approaches have been identified to reduce the susceptibility of CA resists to MB: effective chemical air filtration, modifications to resist chemistry/processing and cleanroom protocols involving MB monitoring and removal of MB sources from the fab. The final MB concentration depends on the effectiveness of filtration resources and on the total pollution originating from different sources in and out of the cleanroom. There are many well-documented sources of MB. Among these are: ambient air; polluted exhaust from other manufacturing areas re-entering the cleanroom through make-up air handlers; manufacturing process chemicals containing volatile molecular bases; certain cleanroom construction materials, such as paint and ceiling tiles; and volatile, humidifier system boiler additives (corrosion inhibitors), such as morpholine, cyclohexylamine, and dimethylaminoethanol. However, there is also an indeterminate number of other 'hidden' pollution sources, which are neither obvious nor well-documented. None of these sources are new, but they had little impact on earlier semiconductor manufacturing processes because the contamination levels are low enough that they were tolerable. The purpose of this article is to investigate some of these frequently overlooked sources of basic molecular contamination and to thereby increase the reader's awareness of their potential risks.