Nanoporous organosilicate films have been recently prepared using tetraalkylammonium cations in acid and basic media, outperforming other materials. Resulting films using basic medium were called zeolite-inspired low-k dielectrics. Here we study the dependence of the properties of these films on the used silica sources: methyltrimethoxy silane (MTMS) and tetraethyl orthosilicate (TEOS). A set of experiments varying the MTMS:TEOS ratio were prepared in acid medium and characterized. A textural, physico-chemical, mechanical, and electrical characterization of this series of experiments is presented.

The results of copper Chemical Mechanical Planarization (CMP) experiments with a model slurry chemistry based on the combination of Glycine-water-Benzotriazole (Gly-H2O2-BTA), and different types of composite A (silane coupling agents between the polymer core and the silica shell) and B (electrostatic attraction between the polymer core and the silica shell) abrasives, are presented. While the presence of BTA allows a ten-fold reduction in the static etch rate from 95 nm/min. to 10 nm/min., combining oxidizer and complexing agent leads to removal rates higher than 400 nm/min. Different surface morphology and RMS roughness are observed after polishing with composite abrasives and different peroxide concentrations. Oxidizer concentrations as low as 0.1 vol.% lead to high non-uniformity and defectivity values. In particular, composite B performs better than pure colloidal/fumed silica during copper CMP using the IC-1000 pad, giving comparable Material Removal Rate (MRR), but a better surface finish due to the contribution of the elasticity of the polymer in gently transferring the applied load to the wafer surface. Cu CMP with pure polymer particles is a promising alternative to the hard inorganic material especially if combined with suitable surfactants that act from both particle stabilization and friction reduction / lubrication improvement perspectives.

The use of the medium/high-hardness pad IC-1000 is compared to the use of a soft Politex pad. In the former case, differences in terms of MRR, MRS roughness, and total defects are observed between the composite abrasives A and B; in the latter case, the behaviour of the two composites is similar. In the case of a soft pad in combination with composite abrasives, there is a remarkable improvement in the defectivity without any loss in MRR.

As revealed by SEM inspection of the composite particles collected in the slurry drain after CMP, for all the composites, the silica shell coverage is not disrupted by the shear forces and chemistry during the 1 min. polishing. Consequently, the stability and agglomeration properties of the particles in the complex Cu CMP chemistry can be helpful in explaining the experimental results in terms of MRR and surface finishing.

Although dielectric polishing is primarily mechanical in nature, surface chemical effects can be tailored to enhance material removal, selectivity and planarity. The use of ceria abrasives in oxide polishing is particularly attractive because of favorable polishing characteristics that are generally not obtainable using conventional fumed or colloidal silica abrasives. Unfortunately, a characteristic of ceria abrasive is an enhanced tendency toward defectivity in comparison with conventional silica. Entirely novel composite structures comprising 300 nm polymer particles coated by ceria (specific surface area of the powder 61 m2/g), are achieved by either adding silane coupling agents (Composite A) or tuning the pH in order to form electrostatic attractive interactions between the core and the shell (Composite B). The polymer core shows mechanical properties that are highly tunable by variation of its synthesis parameters, while the major advantage of the ceria coating is an enhanced chemical action of the abrasive particles, commonly referred to as the chemical tooth model. In this study we report the evolution of RR, haze, and defects such as particles and scratches in CMP experiments on high-density plasma (HDP) silicon oxide using four abrasive types (14 nm primary size ceria particles, 300 nm polymer particles, composites A and B) at pH 3 and 10. Interestingly, the two types of composite exhibit different RR. This is attributed to differences in morphology and surface composition. Composite B presents more similarities with the ceria, as confirmed by particle / silica surface adhesion forces in different pH solutions, measured by a colloidal AFM technique.

Atomic Force Microscopy (AFM) was employed to probe the mechanical properties of surface-charged polymethylmethacrylate (PMMA)-based terpolymer and a composite terpolymer core-silica shell nanosphere in air and water media. Since these materials exhibit enhanced mechanical properties, such as toughness and elasticity, and enhanced chemical stability, they are particularly interesting for potential applications in reducing defectivity during the process of Chemical Mechanical Planarization. The polymer particles were subjected to a thermal treatment aimed at improving their mechanical properties in terms of hardness (H) and elastic modulus (E). By analysis of force-displacement curves and on the basis of Hertz's theory of contact mechanics, Young's moduli were measured for the terpolymer compared with the composite that has expected mechanical property enhancement due to its silica shell. In air, E increases from 4.3 GPa to 6.6 GPa for the treated terpolymer compared with the respective value of 10.3 GPa measured for the composite. In water, E increases from 1.6 GPa to 4.5 GPa for the thermally treated terpolymer that is comparable with the respective value of 3.6 GPa measured for the composite. This observation suggests that as an alternative to the creation of polymer-silica composite nanoparticles for CMP, comparable mechanical properties can be achieved by a simple heat treatment step.

29Si magic angle spinning nuclear magnetic resonance (MAS NMR) spectroscopy was employed to characterize short-range atomic structure modifications to low k dielectric films that were subjected to post-deposition plasma exposure or UV curing. Comparison of spectra from single thick depositions of a CVD low k film (k∼3.0) with sequential thin depositions of the same film revealed ∼3% increase in Si-O crosslinking that was attributed to interfacial plasma damage. Comparison of a second CVD low k film (k∼3.0) before and after UV curing revealed ∼11% increase in Si-O crosslinking with commensurate losses of Si-OH and Si-CH3 groups. UV curing was believed to result in bulk modification. This crosslinking was found to increase the Young's modulus of the film from 11 to 16 GPa as measured on 700 nm films by nanoindentation. NMR analysis was found to provide significant information beyond that provided by FTIR but required special sample preparation and extensive data collection.

In this paper, we investigate the effect of copper diffusion barrier integrity on the leakage behavior and dielectric reliability of copper/micro porous organo-silica-glass (OSG) interconnects. Significant differences in the field dependence of TDDB median-time-to-failure are observed when comparing sub-critical and sealing barriers. Also for the temperature acceleration of TDDB, a significant difference is found which is reflected in the thermal activation energies. With fast voltage ramp measurements, I-V curves of samples with subcritical and sealing barriers are compared before and after constant current stresses. Above 1.4 MV/cm, the dominant leakage mechanism is found to be Frenkel-Poole emission regardless of barrier treatments and stress times. Below 1.4 MV/cm, however, the I-V characteristic is modulated by the barrier integrity, which can be attributed to copper diffusion into the intermetal dielectric.

Effects of alloying Ni with Pt and Ta on silicide properties for CMOS technology have been studied. It was found that Pt is soluble in NiSi, which is in line with literature, whereas Ta segregates towards the surface during thermal treatment. Additionally, Ta retards NiSi formation at low temperature. Thermal stability of NiSi on Si is improved more efficiently by alloying Ni with Pt compared to Ta. Silicide/diffusion contact resistance is extracted using the Transmission Line Structure. In our experiments, contact resistivity appeared to be virtually unaffected with respect to the alloying element. Thermal stability on narrow poly Si structures was also improved when Ni was alloyed with Pt. Similar leakage currents for Ni and Ni(Pt) silicides on N+ and P+ junctions were obtained. The results presented in this work suggest that Pt is a better candidate as alloying element to improve NiSi thermal stability for CMOS processes than Ta.

Silicidation of small features of various geometries and sizes using Ni-silicide was studied. Effects of dopants, surface preparation and silicidation parameters on silicide morphology were investigated. It was found that Ni silicide thickness and quality of the silicide/silicon interface (presence of NiSi2 pyramids) are dependent on the area dimensions and geometry. NiSi formed on narrow lines is thicker compared to wide areas. We found that pyramids of epitaxial NiSi2 protruding into the Si substrate are formed during the silicidation of undoped, boron-doped and BF2-doped active areas. The presence of pyramids was not observed on As-doped silicon. Rough interface due to the pyramids could be the reason for a more pronounced linewidth dependence of leakage current in the case of p+/n diodes. A higher density of pyramids was found on narrow lines (below 0.2μm) and particularly islands structures. No pyramids were found when a Ge preamorphisation implant was used.

ALD WNC nucleation and growth was observed strongly affected by differentsubstrate materials and surface chemistries. Nucleation was inhibited onmost pristine low k surfaces, which is attributed to low concentrations ofchemisorption sites (Si-OH). Plasma treatments were used to alter thesurface chemistry to improve nucleation. Surface closure and surfaceroughness of the WNC layer were found to strongly correlate with startingsurface condition. Resistivities of the resulting films were also founddependent on starting surface treatment. But the relationship between Wcontent of the films, surface treatment and resistivity was not fullycomprehended.

This paper presents an overview of Ni-alloy (Ni, Ni-Pt and Ni-Ta) silicide development for the 45 nm node and beyond, including applications to self-aligned silicide (SALICIDE) processes, reaction with SiGe and strained Si on SiGe, and applications to fully silicided (FUSI) gates. Key SALICIDE issues addressed include the use of spike or low temperature rapid thermal processes (RTP) to control silicidation and junction leakage on small features, factors affecting the formation of epitaxial pyramidal NiSi2 grains, and NiSi thermal stability and agglomeration kinetics. Alloying with Pt or Ta is shown to improve thermal stability of NiSi films, although with quite different behaviors. While Pt is incorporated predominantly in solution in NiSi, Ta segregates to the surface of the films. Ni-Pt alloy silicides were also found to achieve low sheet resistance on narrow gates, low contact resistivity and low junction leakage, making them attractive for CMOS applications. For the Ni/SiGe reaction, a narrower RTP process window for low sheet resistance and a lower activation energy for agglomeration were observed when compared to the Ni/Si reaction. The lower thermal stability was correlated to Ge segregation from the Ni(SiGe) films. The Ni/doped poly-Si reaction was studied for FUSI gate applications, showing a retardation of the silicidation kinetics for high B doses and a large pile- up of dopants (for As, B or P) at the NiSi/SiO2 interface due to dopant snowplow during silicidation. The work function (WF) of NiSi was observed to shift with the addition of dopants, effect attributed to modifications of the interface dipole by the pile-up of dopants. No significant degradation was observed when comparing gate oxide breakdown statistics for Ni FUSI to conventional poly-Si gates. The process window for a FUSI gate-last process (performed after S/D Ni silicidation) was evaluated showing a potential integration problem due to possible degradation of the S/D silicide during the FUSI gate reaction.

Making use of SPER (Solid Phase Epitaxial Regrowth) As and B deep source/drain junctions with high activation can be obtained at temperatures below 700°C. However, higher thermal budget is required to regrow and activate the dopants in the poly gates. Low junction leakage and low contact resistance can be obtained for Ni-silicided As and B SPER junctions making use of deep As and B implants. Because of the low thermal budget source/drain junctions obtained by SPER are an attractive alternative to conventional spike annealed junctions for technologies making use of metal gates.

Our understanding of the role of the initial surface on atomic layerdeposition (ALD) of Cu diffusion barrier materials is limited by thecomplexity of the sequential reactions and the heterogeneous nature oftypical dielectric substrates. The atomically controlled surface chemistryof self-assembled monolayers (SAMs) provides a means of creating modelsubstrates for ALD. Here we report on ALD of WCxNyfilms on SAMs derived from bromoundecyltrichlorosilane adsorbed on silicondioxide. The as-prepared SAM is macroscopically ordered with the expectedBr-termination and has a well-defined chemical composition as determined bycontact angle measurements and X-ray photoelectron spectroscopy,respectively. Temperature programmed desorption spectroscopy confirms thatthe SAM is stable to 550°C. It survives multiple cycles of ALD at 300°C asevidenced by the detection of mass fragments characteristic of the alkylchain and supported by the persistence of a Br 2p peak at 71 eV. X-rayfluorescence, ellipsometry and atomic force microscopy reveal that theunderlying SAM influences WCxNy film coverage, thickness, andmorphology.

Porous materials are being investigated as low dielectric constant (low-k) materials. While porosity decreases the k-value of a material by decreasing its density, it simultaneously allows unwanted adsorption and diffusion of chemicals inside the porous matrix. To investigate this, different porous low-k materials, specifically silicon oxycarbide (SiOCH), methylsilsesquioxane (MSQ), and a polymer, were exposed to polar (ethanol) and non-polar (toluene) solvents. A difference in diffusion of polar and non-polar solvents would be an indication of the density of polar centers which attract polar molecules (such as water) and increase the dielectric constant of a film. The diffusion coefficient for toluene at room temperature was found to be approximately 2×10-5 cm2/sec for MSQ (40% porosity), 10-7 cm2/sec for SiOCH (7% porosity), 2×10-8 cm2/sec for the polymer. The observed diffusion can be described by a model of a viscous flow in a porous medium. The toluene/ethanol diffusion coefficient ratios were 4.4, 1.3, 1 for MSQ, SiOCH, and the polymer, respectively. The difference in toluene/ethanol diffusion can potentially be used to screen a material's affinity for water adsorption.

An overview of silicide development for the 65 nm node and beyond is presented. The scaling behavior of Co based and Ni based silicides to sub-100 nm junctions and sub-40 nm gate lengths was investigated. Co and Co-Ni silicides required a high thermal budget to achieve low diode leakage. Even for lower thermal budgets, the sheet resistance of Co and Co-Ni silicides increased at gate lengths below 40 nm. NiSi had low sheet resistance down to 30 nm gate lengths exhibiting a reverse linewidth effect (sheet resistance decreased with decreasing linewidth), achieved lower contact resistivity than CoSi2 and lower diode leakage for similar sheet resistance values. Bridging issues cannot be ignored for NiSi, in particular for thicker Ni films, higher RTP temperatures and in the presence of Ti. Material issues for the application of NiSi were also investigated. Ni2Si was found to grow with diffusion limited kinetics in the 225-300°C range, with an activation energy of 1.5 eV. Results of the kinetic studies were used to design a two-step RTP process that limited the silicide thickness on small features by a low thermal budget first RTP step, reducing the reverse linewidth effect and avoiding excessive silicidation. In the presence of an interfacial oxide, undesired epitaxial NiSi2 pyramidal grains grew directly at temperatures as low as 310°C on p+ Si. Thermal stability of NiSi was also investigated. We found that the initial mechanism of degradation for thin NiSi films was agglomeration, with activation energies of 2.5-3 eV. The surface after agglomeration remained quite flat with alternating NiSi and exposed Si areas, while the interface roughened significantly. Thick films also degraded initially by agglomeration at low temperatures, but by transformation to NiSi2 at higher temperatures. The addition of Pt improved thermal stability of NiSi films against agglomeration. The Ni/Si-Ge reaction was also studied, finding that the addition of Ge reduced the thermal process window and resulted in a slightly higher resistivity.

For the 70nm CMOS node, it is anticipated that conventional implantation and spike annealing approaches, even with pre-amorphisation and co-implantation, are unlikely to provide pMOS junctions consistent with the ITRS requirements. Here the junction performance is limited by equilibrium solid solubility.

As laser annealing and in-situ doping techniques currently have unsolved integration problems, there is a renewed interest in using solid phase epitaxial regrowth (SPER) to form ultra-shallow metastable junctions. Such junctions have the potential to have an active dopant profile similar to the as-implanted profile. This offers above equilibrium solid solubility and abrupt profiles compatible with 70nm and even 45nm nodes. However there are concerns about residual defects, deactivation, diffusion and uniformity.

In this paper we show how the Ge, F and B implant and SPER anneal can be optimised for abrupt, uniform and highly activated B junctions. There is latitude for higher doses and energies than conventional implants, however results show that this may lead to clustering causing enhanced deactivation and reduced mobility. We give attention to the probing issues involved in characterising partially annealed junctions.

With this approach, p-type junctions having a sheet resistance of 265 ohms/sq and depth of 22nm are realised which are compatible with 70nm and potentially 45nm CMOS nodes.

In this work the reverse bias junction leakage was studied for Co-silicided 100 nm deep As source/drain junctions. The effect of pre-clean and silicidation temperature was investigated. The area component of the leakage current was found to be dominant for silicided source/drain areas wider than 1 mm. Increasing the thermal budget for silicidation was found to improve the area leakage. For diodes consisting of active area stripes narrower than 0.5 μm, the leakage current is no longer improved by increasing the silicidation temperature. As a result the leakage current is found to depend strongly on the active area linewidth. It was found that the linewidth dependence of the junction leakage cannot be attributed to silicide induced stress. It is argued that the higher leakage current observed for narrow lines can be attributed to the stress induced by the STI isolation and to increased silicide thickness in the narrow active lines

During spike annealing of ultra-shallow junctions, large fractions of the dopants form a partially active pile-up at the interface between silicon and the screening oxide layer. In this paper, we show results of sheet resistance, SIMS and high resolution Elastic Recoil Detection measurements to investigate the physical and electrical behaviour of B and As dopant atoms at the interface.

Our results show that the fraction of dopants segregated at the interface is as high as 30-50% for B, but is dependent on dose and the type of screening oxide. Concentrations of up to 3e20cm-3 and more of active dopants are found on the Si side of the interface. The presence of nitrogen in the oxide at the interface causes a higher and sharper pile-up. Results indicate that a similar peak is expected for As, with active concentrations above 6e20cm-3. In an HF dip, the pile-up is removed together with the oxide or deactivated during native oxide regrowth.

Further experiments show that immediately after removing the screening oxide in an HF dip the sheet resistance for B decreases sharply due to carrier accumulation, then raises to about 6-9% above the initial level depending on the oxide and dopant species. The sharp decrease in resistance is not observed for As.

A ULK material based on a siloxane resin has been developed that can be processed using spin-coating and thermal cure to yield porous low-k films. The chemical bonds between the resin and porogen groups prevent the phase separation of the porogen from the resin during curing and lead to extremely small pores. The highly hydrophobic thin films made from this material displayed dielectric constant of 1.8, breakdown voltage of 4 MV/cm, a cohesive strength > 60 MPa, excellent crack resistance, and an average pore size of 2.2 nm by Positron Annihilation Lifetime Spectroscopy (PALS) and 2.5-3.0 nm by Ellipsometric Porosimetry (EP). In this paper, our strategy for designing low-k materials, the material properties and initial integration results for this new material will be discussed.

This work deals with the analysis of the resistances of Co-silicided As and B junctions. The influence of junction formation and salicide process on contact resistance was investigated. The contact resistance between the silicide and source/drain region was studied by making use of dedicated transmission line structures. The transmission line structures (TLM) consist of alternating silicided and unsilicided diffusion segments, obtained by making use of a salicide blocking mask. Our TLM results show that the contact resistance has a strong dependence on the doping profile of the diffused region and increases with decreasing surface doping, with increasing Co-silicide thickness, annealing temperatures and time of silicidation.

The relationship between silicide thickness, sheet resistance and silicon consumption is experimentally checked for Co-disilicide and Ni-monosilicide. The reverse bias leakage current of shallow Ni-silicided and Co-silicided square diodes is compared for varying junction depth and varying silicide thickness. A lower reverse bias leakage current is obtained for a Ni-silicided shallow junction as compared to its Co-silicided counterpart. This can be attributed to the reduced silicon consumption. The Ti cap does not play an active role during the Ni-silicidation of narrow active area and poly lines. It is shown that a Ni-silicidation process is scalable without Ti cap.