We have investigated new aromatic polymers for nanoimprint and subsequent dry etching, namely thermoset and thermoplastic compounds. They were tested in a SiO2 patterning process under low pressure and high plasma density conditions and feature a selectivity about twice as high as poly(methylmethacrylate) (PMMA). The imprint behavior is comparable to PMMA and, in particular, the thermoplastic polymers show excellent imprint quality. This was demonstrated by replication of large arrays of lines down to 50 nm width. The thermoset polymers showed excellent dry etch stability and Teflon-like antisticking layers helped to imprint them without sticking.
The submicron contact hole etching process using CHF3–CF4–Ar plasma was performed in a commercial parallel plate reactive ion etching reactor (LAM 4520). Response surface methodology was employed to correlate input parameters with etching results. For the aspect ratio investigations, a special test pattern consisting of line/space and contact/space arrays of various dimensions (1 and 0.6 μm) was used. Aspect ratio dependent etching effects were observed. The bottom surface chemistry of SiO2/Si patterned structures after plasma etching have been investigated by using quantitative x-ray photoelectron spectroscopy (XPS). After the SiO2 etching treatments, the silicon surface of the contact (or trench) bottom presents modifications similar to that observed on unpatterned SiO2/Si samples. These modifications are described by a two-layer model involving a fluorocarbon overlayer and an interfacial oxide layer. The fluorocarbon layer thickness strongly increases when the surface area of silicon substrate exposed to the plasma decreases whereas, as unexpected, the fluorine density of the fluorocarbon film decreases. A refluorination mechanism depending on the diffusion of the fluorine atoms from the plasma through the fluorocarbon overlayer is proposed. This work points out the interest of XPS surface analysis in improving the contact hole etching process.
Summary form only given. Plasma processes using high-density sources have been extensively developed to meet more and more stringent constraints required by integrated circuits fabrication. Among the various steps, dielectric etching is the more challenging as the process relies on polymerizing hydrofluorocarbon gases that produce simultaneously deposition and etching. It is thus difficult to achieve adequate SiO/sub 2//mask etch selectivity and to continue etching in high aspect ratio features at the same time. More recently, the development of integrated optical components leads to new technological challenges in particular concerning deep etching (= 10 gm). Hence, fabrication of MEMS (micro electro mechanical systems)and O-MEMS (Optical MEMS) requires several conditions: (i) a higher etch rate to reduce the process time (400 nm/min), (ii) an extreme selectivity (> 15), (iii) a much longer etching process. In order to improve SiO/sub 2//Si selectivity, there have been several studies of high density plasma etching processes. Usually, selective etching of silicon oxide with a silicon mask is obtained in RIE plasmas using fluorocarbon gases. The formation of a fluorocarbon film by CF/sub n/ radical deposition on the silicon surface reduces the silicon etch rate whereas oxide etching is weakly affected leading to a considerable improvement of the selectivity. However, the low ion flux generated by the RIE system leads to very low oxide etch rates less than 50 nm/min. In contrast, in high density plasma sources, the high degree of dissociation and the high ionic density enable to reach higher oxide etch rates (> 200 nm/min). Nevertheless, the SiO/sub 2//Si selectivity is strongly affected. The present work is based, in a first step, on the comparison between three fluorocarbon gases (CF/sub 4/, C/sub 2/H/sub 6/, CHF/sub 3/) and their mixture with hydrogen or methane for a better understanding of the selective etching of SiO/sub 2/ using a silicon mask. We show that adding methane instead of hydrogen to any fluorocarbon gas allows to increase considerably the SiO/sub 2//Si selectivity without decreasing too much the oxide etch rate. However, we demonstrate that is not possible to increase both the selectivity and the etch rate by changing the hydrofluorocarbon mixture, and that the only way to obtain this combined increase, is to reduce the gas residence time. Finally, pattern transfer into silicon dioxide with high SiO/sub 2//Si selectivity was performed with success.
Summary form only given. There has been much interest in inductively coupled plasmas for semiconductor processing and for integrated optic applications. We are concerned with several processes using Inductively Coupled Plasma, such as deep silicon oxide etching using fluorocarbon gases (CF/sub 4/, C/sub 2/F/sub 6/, CHF/sub 3/) and their mixtures with methane (CH/sub 4/) or hydrogen (H/sub 2/), low-k etching using fluorocarbon gases mixed with oxygen (O/sub 2/) or nitrogen (N/sub 2/) and resist etching in oxygen, fluorocarbon or SF/sub 6/ plasmas. Our main aims are to develop, understand and optimize these processes. When considering etching, a fundamental process parameter is the ion flux impinging the substrate, as well as the Electron Energy Distribution Function (EEDF). We use Langmuir probe diagnostic to measure these two parameters. Most of the gases used are polymerizing gases and deposition on the probe tip is a major problem to be addressed to obtain reliable probe measurements. Furthermore, all the gases studied (except nitrogen) have a common point, their strong electronegativity, that leads to another problem to be faced. Actually, we have observed strong plasma oscillations when using electronegative gases. This kind of plasma instabilities were attributed to a periodic capacitive to inductive discharge jump because of a periodic negative ion creation. As a consequence of the plasma oscillations, probe acquisitions are disturbed and do not allow to determine plasma electrical characteristics. Hence, time resolved probe measurements must be employed in order to follow electron and ion densities variation during instability. In this work we present our first studies concerning time resolved probe measurements during instabilities. First, we have identified for the different gases used, the instability window (source power versus pressure). We show for example that C/sub 2/F/sub 6/ plasmas present oscillations over a very large range of power and pressure. Second, for non polymerizing gases, we have measured time resolved electron and ion densities during plasma oscillations.
Summary form only given. International efforts are being conducted to develop the 157 nm lithography, which is expected to be the next generation optical lithography, allowing to meet the 70 nm node and possibly go beyond. We are involved in a European IST program, joining several national laboratories and industrials, to address resist challenges associated with the 157 nm lithography. Resist absorbance is a major problem at 157 nm, "usual" carbon containing resists having too high absorbance at this wavelength. A possible way to face this problem is to use silicon containing resists, which present a suitable absorbance at 157 nm. However, their too low resistance to chloro or fluorocarbon plasma etching does not allow to use them in a single layer scheme. They must be associated with a "usual" resist in a bilayer scheme. Our laboratory is concerned with this bilayer scheme. More precisely, we study the opening phase, following the silicon resist development and consisting in etching the under layer resist in O/sub 2/, SF/sub 6/ or CF/sub 4/ plasma. Our aims are to identify the etch mechanisms and to measure and understand the resist/resist selectivity, which is one of the most important parameter of the bilayer process. We have started our studies with the PDMS polymer (polydimethylsiloxane), which is a suitable resist for 157 nm lithography. Open field areas of PDMS over Novolac AZ5214 resist (carbon containing resist) have been etched in inductively coupled oxygen plasmas, for different bias power. Resist etch rates have been measured in real time by in-situ multi-wavelength ellipsometry. PDMS surface, after different plasma etching time, have been analysed by quasi in-situ X-ray photoelectron spectroscopy. We have shown that during the first seconds of oxygen plasma, a SiO/sub x/-like layer is formed onto PDMS and notably decreases its etch rate. Angular XPS analysis have shown that this effect is not a bulk effect, and only concerns first PDMS atomic layers. The SiO/sub x/-like layer protects PDMS and allows to reach high resist/resist selectivity.
In this paper, we analyse, by the use of different plasma diagnostics, appearance potential mass spectrometry (APMS), optical emission spectroscopy (OES) and Langmuir probe measurements, a commercialized ICP source devoted to the etching of SiO2 using a Si mask. First, the influence of the gas composition (C2F6 mixed with H2 or CH4) and the residence time (varying gas flow rate) on the etching rates and selectivity is studied to optimize the process. Second, in order to improve the understanding of the etching mechanisms, the plasma is characterized according to the previous discharge conditions. We point out the presence of plasma instability due to the electronegative character of the fluorocarbon gas used. To determine the ion flux (ϕi) which is an essential parameter for oxide etching, Langmuir probe measurements have been associated with a plot of the bias power versus bias voltage (Pbias(Ei)). Absolute concentrations of CFx (x = 1–3), CH3 and CHF2 radicals have been determined by APMS and the atomic fluorine concentration has been sampled by OES using argon actinometry. The techniques employed for concentration determinations are largely discussed. Finally, we compare the evolutions of the etch rates and the evolutions of the different plasma species with experimental conditions.
SiO 2 is a well suited material for integrated optic applications and is also attractive for microelectromechanical system and micro-optical electromechanical system fabrication. Such optical components require deep oxide etching (several microns) and subsequent high selectivity with respect to the mask. In this article, we describe the influence of various process parameters (gas mixture, pressure, plasma power, and residence time) on the selective etching of SiO2 with respect to Si in inductively coupled plasma (ICP) fluorocarbon with the aim of finding the best compromise between high selectivity and high oxide etch rate. Oxide etch rate is improved by decreasing pressure or increasing source power within the acceptable process windows, respectively, 3–20 mTorr and 1000–2000 W, but the gain in selectivity is low (×1.5). Adding methane rather than more commonly usual hydrogen resulted in higher selectivity without significant decrease in the oxide etch rate. A relatively good correlation is found between the selectivity and the (C+H)/F ratio of the precursor molecule. However, we show that varying the hydrofluorocarbon mixture does not allow us to improve both oxide etch rate and selectivity. In this regard, the residence time is the most significant parameter: choosing the appropriate amount of methane mixed with C2F6, and decreasing tR leads to an improvement in both the selectivity (×7) and the oxide etch rate (×1.5). Finally, the influence of these parameters on pattern transfer is investigated.
This work is focused on the plasma development of siloxanes investigated as model Si-containing photoresist components that show a promise for bilayer lithography at 157 nm and other Next Generation Lithography technologies. In such lithography, the image is developed in the top photosensitive polymer and transferred to the (usually thick) organic underlayer by means of O2-based plasma etching. In this work particularly, the issue of line edge roughness (LER) induced by transfer etching and its reduction by means of plasma processing optimization is addressed. The experimental results reveal that low values of line-edge roughness are obtained in a high-density plasma reactor, if an F- but not O-containing etching first step is used in appropriate plasma conditions. The effect of different etching chemistries and processing conditions on imaging layer roughness formation is demonstrated with the aid of scanning electron microscopy images and image analysis for quantifying LER, and atomic force microscopy (AFM) for measuring surface roughness. X-ray photoelectron spectroscopy analysis of etched PDMS is used to show the evolution of the chemical modification of the PDMS layer, to measure the top oxide thickness, and to correlate both to processing conditions. In situ interferometry and ellipsometry are used to determine the etch resistance of the imaging PDMS layer and the selectivity of the transfer etching process. It is demonstrated that optimum LER correlates well with plasma processing conditions that ensure a nonselective first etching step prior to a highly selective main etching.
