Search

A polymorphism in the gene encoding the serotonin (5-HT) transporter (5-HTT) has been shown to moderate the response to CO

Neurodegenerative diseases are complex, progressive disorders and affect millions of people worldwide, contributing significantly to the global burden of disease. In recent years, research has begun to investigate epigenetic mechanisms for a potential role in disease etiology. In this chapter, we describe the current state of play for epigenetic research into neurodegenerative disorders including Alzheimer's disease, Parkinson's disease and Huntington's disease. We focus on the recent evidence for a potential role of DNA modifications, histone modifications and non-coding RNA in the etiology of these disorders. Finally, we discuss how new technological and bioinformatics advances in the field of epigenetics could further progress our understanding about the underlying mechanisms of neurodegenerative diseases.

The Line-Edge-Roughness (LER) of resist pattern on fine feature has been characterised by means of top/down line width measurements by SEM in Top Surface Imaging (TSI) technology. The resist surface investigation using AFM has provided a correlation between resist Surface Roughness (SR) and the formation mechanism of LER. LER has been improved to 7 nm at 0.18 @mm dense patterns by the optimisation of dry development conditions based on these resist surface investigation.

Three-color flow cytometry immunophenotyping revealed significant increases of CD57+ and CD28- cells among both circulating CD4+ and CD8+ T lymphocytes of untreated hemato-oncological patients (n = 54) as compared to healthy donors (n = 55), with CD57 and CD28 expression on the patients' T cells being largely reciprocal. Marked expansion of CD57+ cells among circulating CD4+ T lymphocytes was frequently detected in patients with chronic leukemia of B cell origin (B-CLL, hairy cell leukemia) but not in patients with chronic myeloid leukemia, suggesting a causal relation with the tumor's major histocompatibility complex class II expression. Using immunomagnetic separation techniques, we further demonstrate that the patients' CD57+/CD28- T cells display a typical Th1-type cytokine secretion profile upon anti-CD3 stimulation, with a markedly higher secretion of the Th1-type cytokines IL-2, IFN-gamma, and TNF-alpha than their CD57-/CD28+ counterparts. Cytotoxic activity of circulating CD8+ T lymphocytes, measured ex vivo in an anti-CD3-redirected assay, was almost exclusively exerted by the CD57+/CD28- subset. Moreover, a marked cytotoxic activity was detected within CD4+CD57+ T cells from some B-CLL patients. Finally, the patients' CD57+/CD28- T cells displayed an increased tendency to apoptosis in culture. Collectively, our results indicate that the expanded CD57+/CD28- T cells in hemato-oncological patients represent differentiated effector cells, similar to their (quantitatively minor) counterpart in healthy donors. The reason for their expansion and their pathophysiologic significance, however, remains unclear.

Excellent resolution and latitudes can be obtained with the commercially available advanced i-line resists. However, printing close to the resolution limit results in severe optical proximity effects (OPE). These effects limit the intra-die CD control severely. In order to manufacture devices with sub-0.35μm design rules, optical proximity correction (OPC) is needed. The goal of this paper is to illustrate the ultimate CD-control that can be obtained using an advanced i-line resist process and stepper in combination with OPC. Applying OPC results in clear improvements for 350, 300 and 275 nm features. CD-error over the full space/line ratio range is better controlled, resulting in larger overlapping exposure windows. For all the features, end-of-line shortening is significantly reduced applying OPC.

The requirements imposed by optical proximity correction (OPC) on mask making, i.e. pattern generation, are addressed. Dry etching of MoSi is proposed to deal with the need for improved resolution. E-beam proximity correction (EPC) is used for more accurate linewidth control. An appropriate pre-correction fracturing technique is presented. A MoSi based embedded phase shift mask (5X) has been produced with a critical dimension of 0.3 μm using OPC corrected data. Highly accurate linewidth control was achieved by the reduction of the e-beam proximity effect and linearity errors to less than 3%. The effect of OPC and EPC on mask making throughput are discussed. An alternative OPC strategy is suggested that is more compatible with exposure grid requirements.

In printing random logic circuits down to 0.3 μm using i-line lithography, optical proximity correction is required to maintain across-the-chip linewidth uniformity. Using a rule-based approach with parametric anchoring, process characterization time is kept to a minimum. Corrections are more effective if post-OPC design grid sizes are kept small (∼5 nm at 1X).

Lithography has enabled the continuous shrinking of device geometry over many years. Whereas it has been predicted many times that optical lithography would run out of steam, it still is the technology of choice, and this will certainly remain for a considerable time. Traditionally, this scaling has been made possible by using lower wavelength light sources and higher numerical aperture projection lenses. It is generally believed that i-line will be the mainstream technology for 0.35 @mm lithography, while DUV (248 nm) will be needed for 0.25 @mm. During the last years, various resolution enhancement techniques, such as phase shifting masks, off-axis illumination and surface imaging have been proposed. Very often, these techniques have been optimized to reduce the depth-of-focus loss, which is unavoidable using the lower wavelengths and higher numerical apertures. In this paper the current challenges for printing 0.35-0.25 @mm dimensions will be reviewed. Although the mainstream technologies for 0.35 and 0.25 @mm are more or less settled, the main challenge will most likely be critical dimension (CD) control. Optical proximity effects and substrate reflections start to play a major role in the CD budget. Methods to overcome these problems are obviously optical proximity correction techniques and the use of anti-reflective coatings and surface imaging. Moreover, a lot of gain can be obtained from reoptimizing current processes for maximum exposure latitude, combined with the minimally required focus budget. It will be shown that optimization of the lithography process (stepper parameters, mask, illumination, resist process) towards CD control results in different conclusions than optimization towards maximum depth-of-focus (DOF). Furthermore, a practical example of 0.25 @mm lithography, using the combination of attenuated PSM and off-axis illumination, will be given. In an attempt to indicate the future of optical lithography, a simulation study is presented which starts from this example. Reasonable processing windows are obtained down to 0.12 @mm by using 'known' techniques.

We propose the combination of an electron beam system in mix-and-match with a DUV stepper on the same resist layer (intra-level mix-and-match) as an extension of available DUV equipment for the investigation and process development for newer CMOS generations. Practical results are presented for the gate exposure of a random logic 0.25@mm CMOS design. Feasibility of a similar approach for the contact hole definition is shown.

A positive-tone wet silylation process for I-line lithography is developed, using safe solvents and the commercial photoresist AZ 5214E (by Hoechst). The process is capable of resolving 0.3 μm lines and spaces on photoresist. Various silylating conditions, solution concentrations and two silylating agents namely HMCTS and B(DMA)DS have been tried. Silicon concentration, focus and exposure latitudes have been defined.

The combination of off-axis illumination (OAI) schemes with ‘weak’ phase shifting techniques is promising to improve the process latitudes of both dense and isolated features. In this paper, the performance of this combination to print 0.25 μm poly levels for logic circuits is investigated experimentally and by simulations, using DUV stepper ( λ = 0.248 μm , 0.42 NA ). Various resist types have been addressed, mainly with respect to proximity issues.

The lithographic performance of phase shifting and transmission masks are compared. This paper describes the ultimate resolution and process latitudes, that can be obtained by combining a phase shift mask with the DESIRE process at 248 nm. Special emphasis is put on the influence of coherence. Both simulation and experimental results are given.

Multiple focus exposures[1], optimization of the partial coherence[2–4] of the stepper and the use of surface imaging[5] have been studied as possible ways to improve depth of focus and exposure latitude for deep-UV lithography at sub-0.5 μm dimensions. Simulation and experimental results are compared. The multiple focus technique gives a marginal improvement in depth of focus, but at the expense of resolution and exposure latitude. Surface imaging combined with higher coherence factors gives the best resolution with the widest latitude.

The formation of CoSi 2 and TiSi 2 on poly-Si lines, with nominal dimensions ranging from 0.5 μm to 1.0 μm, has been studied. Side-wall spacers were formed to allow a self-aligned silicide formation (SALICIDE) process. Silicidation was obtained, in an RTP system, via a solid-solid reaction between the metal and the silicon. Electrical measurements revealed a larger apparent line-width loss, representing a higher sheet resistance, for CoSi 2 -polycide lines compared to TiSi 2 lines. Cross-sectional TEM and SEM investigations showed that the silicide/poly-Si interface was severely bowed for both CoSi 2 and TiSi 2 . A parallel bowing of the silicide surface occurred on the TiSi 2 lines, while the surface of the CoSi 2 lines remained quite flat. Moreover, similar bowing of the silicide interface is also observed after silicidation of mono-Si in narrow oxide windows. Plausible explanations for the silicide/Si interface bowing and the development of different cross-sectional geometries for CoSi 2 and TiSi 2 -polycide lines, are presented. It is suggested, that the observed difference in the cross-sectional geometry of the two silicides is responsible for the larger line-width loss seen on CoSi 2 lines.

The refractory metal disilicides TiSi2 and TaSi2 were investigated for their usefulness as dopant diffusion sources. During furnace annealing and rapid thermal processing, strong decomposition reactions occur between the dopants D (B or As) and the respective silicide (MSi2) to form MxDy compounds. With the help of special sample preparation methods and various analytical techniques, the compound phases TiB2, TiAs, TaB2, and TaAs were unambiguously detected. The fraction of freely diffusing B in TaSi2 is determined to be below 5% of the total dose; by far, the major part of the dopant is bound within the TaB2 phase detected. Careful sample preparation and analysis of secondary‐ion‐mass spectrometry profiles is necessary to avoid artifacts caused by these compound particles. The MxDy‐compound formation has detrimental consequences: The solubility of arsenic and even more of boron in TiSi2 and TaSi2 is limited to rather low‐concentration levels (e.g., B in TaSi2: 4 × 1018 B/cm3 < CB(900 °C) < 1.6 × 1019 B/c...

The redistribution of B and As ions implanted into thin layers of WSi2 and CoSi2 on poly‐ or monocrystalline Si and the outdiffusion into the Si substrate during furnace annealing (FA) and rapid thermal processing (RTP) were investigated by several analytical techniques. Shallow junctions (depth xj < 100 nm) with interface concentrations Cint close to the solid solubility of the respective dopant in Si (Cint≳3×1020 cm−3 for As; (Cint ≳ 8 × 1019 cm−3 for B) were obtained with RTP. For FA above 800 °C, the diffusion of B from CoSi2 into Si results in a drop of Cint < 2 × 1019 cm−3 because of strong B segregation and probably reactive loss at the SiO2/CoSi2 interface. No evidence on metal‐dopant‐compound formation could be found. The dopant redistribution is demonstrated to be a superposition of lattice and grain‐boundary diffusion, solubility limits, layer inhomogeneities, dopant segregation at the interface and grain boundaries, and probably phase transformation of the dopants segregated at the SiO2/silici...

The final lithographic performance of the Desire process is mainly determined by the selectivity of Si incorporation during the silylation step. The influence of the silylation conditions on both the Si contrast and the silylation kinetics have been studied. The silylation process has been optimized with a response surface analysis. The application of these optimum conditions has resulted in a wide process window.

Improvements in the contrast of e-beam exposed resists have been reported via the use of multiple interrupted development techniques. Additional baking is claimed to effect greater improvements in optical lithography. This so called intermediate development bake or (IDB) process, has recently been further simplified through adoption of a warm water treatment for the baking step. In the w arm w ater IDB process, the need to spin dry wafers is eliminated, thereby improving development throughput. We report on the application of all of these treatments to resists exposed using e-beam. Contrast curves obtained with multiple ww IDB processes at different temperatures and times are presented and compared with those obtained utilizing conventional development with varying PEB conditions. The effect on sizing both the top and base of the resist image as a function of exposure is also examined. Olin Hunt HEBR 214 has been at the focus of our work, although we additionally report on HEBR 242 and HPR 204, as well as Aspect System 912.

Dry developable resist systems based on silylation have been presented as an alternative solution for submicron and deep submicron lithography. The DESIRE process (Diffusion Enhanced Silylated Resist) is the most well-known example of such a process. In this paper the dry development step using reactive ion etching has been investigated using a TEGAL MCR (magnetically confined reactor). The MCR system is a versatile tool which provides two major improvements over classical RIE: the magnetical confinement of the plasma and the triode RF coupling system. In this paper the influence of the pressure, the bias power and the oxygen flow on the etch rate and the shape of the photoresist pattern was investigated. The etch rate of the resist and associated uniformity depend on the plasma etching parameters. The exposure dose is however the major parameter determining the critical dimension (CD).

A method of improving the upper part of a developed resist profile, over that obtained with an optimised post exposure bake process is demonstrated. This is achieved by administering an additional low temperature bake, midway through the development cycle; the so called intermediate development bake (IDB). Furthermore, the demonstrated effectiveness of a short infra red IDB, holds promise for future integration with standard track development processes. The effect has been shown to work for both a dedicated i-line resist (Shipley 3413), and a multiwavelength g/i line compatible material (Fuji-Hunt 6100). Comparisons of S 3413 resist profiles obtained with various IDB processes, as well as alternative methods for improving profiles, are presented.

The application of DESIRE, a dry developable optical resist system, in a 0.5 μm NMOS process is described. The investigations are aimed at understanding the major lithography related aspects such as resist resolution and profile, pattern distortion due to swelling of the resist, pattern transfer and registration. Working 0.5 μm devices have been fabricated, illustrating the compatibility of DESIRE with submicron MOS processing.

To test the hypothesis that in pregnancies with a clinically suspected growth restricted foetus at term, induction of labour is as safe as expectant management, and does not lead to increased obstetrical interventions or perinatal morbidity.In one obstetric centre, 33 women with a clinically suspected growth restricted foetus at term were randomly allocated after stratification for parity to either induction or to expectant management. Obstetric and neonatal outcome variables were compared.There was a lower gestational age at labour (median 38(0) weeks versus 40(1) weeks) with a corresponding tendency to lower birth weight (mean 2428 g versus 2651 g), and a reduced need for ante partum medical surveillance, in the induction group. No significant differences in obstetrical interventions (25% versus 24%) and neonatal morbidity rates (50% versus 35%) were found.A larger multicenter study with a sufficient power and long-term follow-up to decide the best policy for the term growth restricted foetus is feasible.

In order to cope with the progressive scaling of CMOS, new lithography techniques are introduced for sub-90nm applications. Among them are the introduction of the 157nm wavelength and extreme ultra-violet lithography (EUVL). However, the delay in availability of these tools requires the extension of 193nm lithography towards future technology nodes. This paper gives an overview of the latest breakthroughs and the remaining challenges to the lithography community.

The feasibility of several materials for use in local interconnection technology in combination with the self-aligned CoSi/sub 2/ process is evaluated. In the self-aligned CoSi/sub 2/ process, CoSi/sub 2/ straps from Co/a-Si (1:2) bilayer reactions are formed from both substrate Si and the a-Si layer, resulting in an a-Si surface layer. Thermal treatments of Si-CoSi/sub 2/-TiW (10:90) revealed a net Si transport in the TiW top layer at elevated temperatures (750 degrees C). Formation of (Ti/sub 0.24/W/sub 0.76/)Si/sub 2/ and WSi/sub 2/ phases, CoSi/sub 2/ phase integrity, and a net movement of the CoSi/sub 2/ layer into the Si substrate have been observed at temperatures above 750 degrees C. The most viable strap material for a CoSi/sub 2/ self-aligned silicide technology is a reactively sputtered TiW(N) layer because of high-temperature stability and compatibility with present IC processing. >

For the first time, it has been shown that 0.18 /spl mu/m and 0.13 /spl mu/m generations can be printed with acceptable CD control using optical projection lithography (248 nm and 193 nm wavelengths respectively). To prove this, a combination of stepper parameters and resolution enhancement techniques was optimised first. The Norman-Debora software allowed us to predict the actual CD control for printing 0.18 /spl mu/m grouped and isolated features using KrF and printing 0.13 /spl mu/m features using ArF, based on realistic process windows and focus/exposure budgets which have been calibrated to experiments for 0.25 /spl mu/m and 0.18 /spl mu/m.

It is generally accepted that optical lithography will be the technology of choice for the fabrication of devices for several generations to come. In this paper some of the current challenges to print 0.25 /spl mu/m dimensions are addressed. Although the mainstream technologies for 0.35 /spl mu/m and 0.25 /spl mu/m processes are more or less settled, the main challenge will most likely be CD control. Optical proximity effects and CD variations resulting from substrate reflections start to play a major role in the CD budget. Several methods to reduce these contributions (such as the use of optical proximity correction techniques and the use of anti-reflection coatings and surface imaging processes) are presented. Moreover, it will be indicated that considerable improvements in CD control can be obtained by optimization of stepper parameters. Finally, in an attempt to indicate the future of optical lithography, a simulation study is presented, indicating the available process latitudes which can be obtained by combining several of the above listed techniques.

Freshly collected chronic lymphocytic leukemia B cells (B-CLL cells) are known to be inefficient at stimulating allogeneic T cells, and to lack significant expression of B7 (CD80 and CD86) costimulatory molecules. We investigated the potential of CD40 triggering to up-regulate the expression of adhesion and costimulatory molecules on B-CLL cells, and to enhance their immunogenicity towards allogeneic T cells. B-CLL cells cocultured with human CD40 ligand-expressing mouse fibroblasts rapidly up-regulated CD54 and CD58 adhesion molecules and B7-1 (CD80) and B7-2 (CD86) costimulatory molecules, and acquired a strong stimulatory capacity towards CD4+ as well as isolated CD8+ allogeneic T cells. Costimulation by both CD80 and CD86 proved critical for allogeneic T cell proliferation and CD25 and HLA-DR expression, since these were strongly inhibited by anti-CD80 or anti-CD86 monoclonal antibodies, and completely abrogated by CTLA4-Ig fusion protein, which blocks both CD80 and CD86. B7 costimulation also proved critical for restimulation of primed B-CLL-reactive T cells. Most importantly, priming of purified CD8+ T cells with CD40-triggered allogeneic B-CLL cells resulted in cytotoxic activity against the unstimulated B-CLL cells. These findings raise the possibility that CD40 triggering of B-CLL cells might be exploited in immunotherapeutic protocols.

SUMMARY We investigated the phenotype and functional capacities of tumour-infiltrating lymphocytes (TIL), freshly isolated from primary renal cell carcinoma (RCC) specimens (n = 20). Three-colour flow cytometry immunophenotyping revealed that RCC TIL consist mainly of CD3+ T cells, with a clear predominance of CD4−CD8+ over CD4+ CD8− T cells, and a marked population of CD4+ CD8+ T cells. Natural killer (NK) cells were also strongly represented (> 25% in 15 of 20 tumour samples), while B cells constituted a minor TIL subset (

Dry development using thin layer imaging (TLI) in either a bilayer approach or top surface imaging are currently investigated as viable alternatives to extend optical lithography to 0.13 μm and below. This article describes a systematic study of dry development in a LAM TCP9400SE inductively coupled plasma etcher for both a single layer TLI resist process and for a bilayer resist process using O2 and SO2/O2 chemistries. The effect of the important machine parameters such as TCP power, bias power, O2 and SO2 gas flows, on the process characteristics (etch rate, selectivity, uniformity, anisotropy) and on the lithographic performance (resolution, profile control, proximity) of a TSI process at 248 has been investigated by means of statistically designed experiments. As line edge roughness (LER) is a critical issue for TSI, the effect of the dry development conditions on LER have been quantified. The effect of temperature on profile control is also presented. In a second part of this article, these trends ha...

The ADEQUAT project [Advanced Developments for Quarter Micron complementary metal–oxide–semiconductor (CMOS) Technologies] is executed by a European consortium under the JESSI/ESPRIT programs. It aims at developing new device structures and process modules for the 0.35 and 0.25 μm CMOS logic technology generations, offering improved performance while meeting the reliability and manufacturability constraints. Several examples of choices illustrating this optimization process are presented together with the overall project organization and roadmap.

In this article, the imaging principles of projection steppers for optical lithography, using standard transmission masks, are reviewed, and compared with the image formation using various types of phase shifting masks (PSM), in order to get a better insight in the working principles of the various PSM techniques. The printability of periodic lines using a standard transmission mask is compared with the use of an alternating shifter PSM and the dependence of imaging on spatial coherence is discussed extensively. Furthermore the imaging of isolated lines is discussed, using a bright field standard transmission mask, a chromeless phase edge PSM, and a dark field rim PSM. The coherence dependence of an alternated shifter is verified experimentally. Experiments also confirm the superior performance of a chromeless phase edge PSM to print isolated lines.

Phase shifting masks have proven their potential to enhance resolution and depth‐of‐focus in optical lithography. Nevertheless, many questions still have to be solved, before the phase shifting concept can be introduced in production lines. Controllable mask fabrication is a major concern. This article addresses the influence of physical material properties and process steps in the attempt to fabricate phase shifting masks with acceptable phase shift and transmittance. Spectrophotometer measurements and computer simulations showed the need of matched refractive indices for all transparent materials in the reticle. Second, the sensitivity of the lithographic performance to these shifter deviations was investigated by aerial image simulations. These predicted a high sensitivity to phase errors, especially for negative resists. Using a negative tone top surface imaging resist process and a positive tone wet developed resist, experiments confirmed these predictions. Special attention was paid in finding an accurate and efficient method to measure the transmittance and phase shift deviations of a reticle experimentally.

Surface imaging in combination with dry development has been suggested as a means of overcoming the inherent limitations present in conventional wet develop lithography. The diffusion‐enhanced silylated resist process has been demonstrated as one such solution. Silylation is one of the more critical steps in the process, with hexamethyldisilazane being typically used as silylating agent. Alternative gas‐phase silylating agents, such as dimethylsilyldimethyl‐amine, 1,1,3,3‐tetramethyl disilazane and N,N‐dimethylamino trimethylsilane, have additionally been studied [Ki‐Ho Baik et al., J. Vac. Sci. Technol. B 8, 1481 (1990)]. Liquid‐phase silylation holds promise for simplifying the hardware requirements as well as improving silylation properties. Several mono‐functional and polyfunctional silylating agents have been investigated. Bis(dimethylamino)dimethylsilane has been found to be promising. N‐methyl‐2‐pyrrolidone was used as diffusion promoter and xylene as the solvent. The influence of solvent compositi...

The diffusion enhanced silylated resist (DESIRE) process has been presented as one of the most attractive surface imaging technologies for advanced optical lithography. It has been shown that the silylation step, usually carried out in a hexamethyldisilazane (HMDS) ambient, is one of the more critical steps in the process. In this paper an in‐depth study of the silylation is presented. Several alternative mono‐ and polyfunctional silylating agents are evaluated for application in the DESIRE process. DMSDMA, TMSDMA, and TMDS are promising candidates, showing several advantages over HMDS. For the case of HMDS and TMDS, the kinetics of the silylation are additionally studied using Rutherford backscattering spectroscopy, Auger electron spectroscopy, and infrared measurements for various process conditions. In an earlier study we have reported lateral swelling for silylation with HMDS resulting from volume expansion of the silylated resist. Because of the reduced volume expansion and/or the lower silylation temperature, these alternative silylating agents reduce the swelling and surface roughness. It has been observed that the silylation process window can be greatly increased by optimizing the presilylation bake.

The ability and advantages of a process in which junctions are formed by diffusion from an implanted silicide were shown before. However, part of the B-dose penetrates through a CoSi2-layer of 100 nm and therefore the heavier BF2 is investigated in this work. It will be shown that excellent profiles in the Si can be obtained with an RTP-diffusion step. In the case of TiSi2, both BF2 and Ga are considered as alternatives. Our results indicate that Ga in combination with an RTP-diffusion step yields the best result. On the other hand, the classical furnace treatments resulted in a pronounced TiGa compound formation which limits the available dose for diffusion into the Si. BF2 is not likely to be used at all with TiSi2 because of metal-dopant and metal-fluorine compound formation.

The ion-beam-induced reaction of metal films on a silicon substrates is studied with emphasis on applications in very-large-scale integration processing. Using the Ti/Si system as an example it is shown that, in combination with subsequent rapid thermal annealing, silicides with excellent properties are formed in a uniform and reproducible way. When using dopant elements for the mixing, shallow junctions with a depth of the order of 0.1 μm are demonstrated, which show excellent electrical properties. The advantages of the ion-beam-induced metal-silicon reaction in a self-aligned silicide process are discussed.

Self-aligned silicidation technologies, based upon TiSi2 (and CoSi2) have been widely investigated during recent years. Junction spiking and limited thermal stability of the dopant concentration at the silicide/Si interface are important limitations of the technique of conventional formation of silicided junctions. Therefore, the diffusion of dopants from implanted silicide is compared for CoSi2 and TiSi2. Excellent diode integrity and low contact and series resistances are obtained for diffusion from CoSi2, whereas the diffusion from TiSi2 is strongly restricted by metal-dopant compound (TiB2 and TiAs) formation.

The thermodynamic equilibrium of structures consisting of a thin film silicide (TiSi2 or CoSi2) on doped Si (with As or B) is investigated. Isothermal sections of the ternary phase diagrams for Ti–Si–B, Co–Si–B, Ti–Si–As, and Co–Si–As have been evaluated, indicating the stability of high B concentrations in Si underneath a CoSi2 layer, the instability of high As concentrations in Si underneath a CoSi2 layer, and of B and As concentrations underneath a TiSi2 layer. The obtained thermodynamic predictions agree very well with experimental results (i) on the redistribution of dopants during silicide formation, (ii) on the diffusion of dopants from an ion implanted silicide, and (iii) on the stability of highly doped regions underneath the silicide, both for the case of TiSi2 and CoSi2. It is shown that even though the inaccuracy of reported thermodynamic data is substantial, thermodynamic calculations provide a useful guidance and are consistent with the experimental results.

The effect of silicidation on the generation and removal of defects in silicon is studied in terms of dopant behaviour and shrinkage or growth of extended defects in silicon during silicide formation. A reduction in the number of interstitial-type defects is observed both for near-noble metal silicides and for refractory metal silicides, suggesting a generation of vacancies. In the case of TiSi 2 , the removal of end-of-range defects was observed from electrical measurements on p + junctions, preamorphized with germanium. Complete annihilation of these defects during CoSi 2 formation failed. This might be attributed to the lower stress level of CoSi 2 on silicon compared with TiSi 2 . The higher stress level of TiSi 2 on silicon can lead to generation of dislocations at film edges for silicide thicknesses exceeding 100 nm.

The interaction in the system poly-Si on TiSi 2 on Si and in the polycide system both with CoSi 2 and TiSi 2 is investigated when subjected to furnace heat treatments at high temperature. It will be shown that the first system suffers from instability due to SPE-regrowth which is partially avoided by doping the poly-Si. The polycide system shows severe roughening of the TiSi 2 . However, in the case of CoSi 2 , the degradation is more pronounced, viz. the Co is detected all over the poly-Si layer and this can hardly be suppressed by doping of the CoSi 2 -layer.

Cobalt silicide is investigated in view of possible application in a self-aligned technology. Extremely smooth, highly conductive CoSi 2 films are obtained using rapid thermal processing for silicide formation starting from deposited cobalt layers (on Si). The phase formation is studied by XRD and RBS. No lateral silicide formation is observed at contact edges. The influence of Si consumption and dopant behavior on diode performance is studied. Shallow arsenic (0.15 µm deep) and boron (0.3 µm deep) junctions are successfully silicided. Very low contact resistances are obtained between the silicide and n+ and p+ regions. MOS transistors were fabricated with CoSi 2 on the source, drain, and gate. An increase in current driving capability is noticed while no degradation of other electrical parameters due to the silicide processing steps is observed. At some critical points, comparison is made with the TiSi 2 process.

We have successfully realized very short MESFET's with gatelengths down to 75nm. Maximum transconductances as high as 335 mS/mm were achieved, even for the smallest gatelengths. During device fabrication two different electron beam systems have been used in a mix&match mode.

The interaction of Ti with Si and SiO2 is studied by means of Auger and X-ray photoelectron spectroscopy. The “peak-to-peak-minus-3-eV-height” method is used for the quantification with AES, while XPS is mainly used to characterize the oxide and nitride states. It will be shown that AES can successfully be applied for quantitative analysis of the resulting TiNx/TiOx/TiSix multilayers in the oxygen lean cases, while a small artifactal nitrogen signal occurs otherwise.