Christiaan Vinckier

A Detailed Study on the Growth of Thin Oxide Layers on Silicon Using Ozonated Solutions

The oxidation of silicon using ozonated, deionized water solutions was investigated as a function of several parameters: reaction time, pH, ozone concentration, temperature, and influence of anions. The oxidation of silicon was dependent on ozone concentration especially near neutral pH. This concentration dependence disappears at concentrations greater than 15 mg/L ozone. No temperature effect was found between 20 and 50°C. Lowering the pH leads to a less pronounced concentration dependence with no specific anion effect between HCl or . The oxidation of silicon by ozonated solutions does not lead to extensive roughening of the silicon surface as shown by atomic force microscopy measurements. Various thermal oxidation models were evaluated and the Fehnler expression represents the experimental data fairly well. The overall oxidation thus follows logarithmic growth kinetics. It is proposed that ozone dissociates at the interface in a one‐step reaction forming the oxidizing species, namely, . This radical diffuses through the layer under the influence of an electric field which develops over the oxide layer. The field‐imposed drift is the limiting factor in the oxidation process. The bulk chemistry of the ozonated solutions is of no importance to the oxidation of silicon. The initial oxidation rate, defined at an oxidation time of 6 s, was dependent on the ozone concentration below 15 mg/L and leveled off above this concentration as it was limited by the field‐imposed drift of the oxidation precursor.

Fundamental study of the removal mechanisms of nano-sized particles using brush scrubber cleaning

To ensure high device yields, wafer surface contamination and defects must be monitored and controlled during the entire process of semiconductor manufacturing. Particle surface concentrations on the wafers, mostly related to chemical mechanical polishing (CMP) processes, must be kept at the lowest possible levels. Brush scrubber cleaning has the potential to achieve this goal. However, the particle removal mechanisms are still under discussion especially the removal of nano-sized particles. This paper investigates the interactions between the particle, the brush and the wafer surface and explores the potential and limitations of the brush scrubbing technique. Furthermore the effect of the various brush/wafer parameters on the particle removal efficiency (PRE) is studied. From a mechanistic viewpoint it is shown that brush scrubbing acts in a mixed lubrication regime. From an extensive analysis of the relevant forces and moments it can be concluded that in the hydrodynamic lubrication regime, particles ar...

The application of ozone in semiconductor cleaning processes : The solubility issue

The solubility of ozone in aqueous solutions expressed as a pseudo-Henrys law coefficient for ozone, * H T O3 , is measured over a pH range of 1 to 8, and temperature range of 15 to 45°C in the presence of several additives: HCl, HNO 3 , HAc (Ac is acetate), NaAc, and NaOH. The value of * H T O3 was found to he a function of temperature, pH, and nature of the anion. The pseudo-Henrys law coefficient is maximal at low pH and decreases with rising pH of the aqueous solution. The addition of chloride enhances the ozone decay rate and thus decreases the solubility helow pH 2 while the addition of acetic acid/acetate results in an increase of * H T O3 for pH > 3. The solubility enthalpy ΔH 0 sol for ozone is found to be equal to (-22.38 ± 0.79) kJ/mol. For applications of ozone in semiconductor cleaning processes, knowledge of the dependence of * H T O 1 on pH, temperature, gas phase concentration, and additive is important in order to achieve optimized cleaning conditions.

Kinetics of the silicon dioxide growth-process in afterglows of microwave-induced plasmas

A fast flow reactor technique, by which thin silicon dioxide layers can be grown, is described in detail. Wafers 3 in. in diameter are treated in the afterglow of a microwave‐induced plasma in oxygen/argon mixtures. This method allowed us to produce SiO2 layers of a uniform thickness up to 300 A. It is shown that the oxide growth rate initially follows a parabolic dependence on the oxidation time while at thicknesses from about 170 A on, a linear relationship is observed. Various physicochemical parameters affecting the oxidation rate are investigated, such as the flow velocity, the wafer position, the microwave power, and the temperature. It is also shown by chemical titration techniques that oxygen atoms in their electronic ground state are the major oxidizing species under the experimental conditions used.

Nucleation and Growth Behavior of Atomic Layer Deposited HfO2 Films on Silicon Oxide Starting Surfaces

Growing nanometer-thin HfO 2 films by atomic layer deposition (ALD) for implementation in advanced transistor structures is controlled by the density of reactive OH sites on the surface. The impact of thin SiO 2 starting surfaces, grown by wet chemical processes and by wetting a thermal oxide, on the nucleation and growth of ALD HfO 2 has therefore been evaluated. Our results demonstrate that both surface pretreatments display the same dependence of the initial HfO 2 growth on the interfacial layer thickness. This correlation is first characterized by a linear increase, which can be interpreted in terms of increasing OH surface concentration. Once an ellipsometric oxide thickness of approximately 0.8 nm is reached, saturation of the HfO 2 deposition occurs. Maximal OH coverage of the surface or steric hindrance of the adsorbed precursor molecules could explain this observation. However, the increased growth-per-cycle at lower deposition temperatures can be attributed to an improved hydroxylation of the surface, excluding steric hindrance as the primary factor causing saturation. Furthermore, electrical characterization revealed that both interfacial oxides show identical leakage scaling behavior down to an equivalent oxide thickness of 0.8 nm.

HfO2 Atomic Layer Deposition Using HfCl4 ∕ H2O : The First Reaction Cycle

The growth behavior and film quality of HfO 2 deposited by atomic layer deposition (ALD) using HfCl 4 /H 2 O depends on the hydroxylation of the exposed surface. In this work, we investigate the dependence of the first HfCl 4 chemisorption reaction at 300°C on the OH density of the silicon surface. We observe that the hydroxyl density, and hence the Hf deposition, on O 3 /H 2 O wet oxides depends on the initial preparation as well as on the stabilization time in the ALD reactor. A good understanding of the initial nucleation behavior is necessary because wet oxides are used in complementary metal-oxide-semiconductor transistors as surface pretreatment for aggressively scaled HfO 2 gate dielectrics. Moreover, the HfCl 4 chemisorption reaction is used to estimate the hydroxylated fraction of these surfaces by means of theoretical models. Finally, the temperature dependence of the OH density, as available in the literature, is applied to gain insight into the stoichiometry of the HfCl 4 chemisorption reaction.

The increasing importance of the use of ozone in the microelectronics industry

Abstract Recently the microelectronics industry is investigating the application of ozonated solutions in the cleaning of semiconductor devices as an alternative for the frequently used H2SO4-mixtures. The use of ozone would result in more environmentally friendly and cost-saving cleaning concepts. To optimize this new wet chemical cleaning processes, fundamental understanding of the behavior of ozone in ultrapure water is required. The decomposition and the solubility of ozone in ultrapure water were investigated as a function of pH, temperature and various additives. Some applications will also be discussed, namely the oxidation of silicon and the mechanistic aspects of the removal of organic contamination.

Production of chemi-ions and formation of CH and CH2 radicals in methane-oxygen and ethylene-oxygen flames*

The mole fractions of CH, CH2, CH3, O, H, OH, O2, and some other species were measured throughout the reaction zones of a series of low-pressure flames burning methane or ethylene in oxygen, diluted by argon. In some flames, the C atom was detected; its ionization potential was found to be 11.1±0.2 eV. For each flame, the total amount of ions produced in unit time was also determined, using the saturation current method. The values for all flames were directly proportional to the corresponding volume integrals ∝[CH][O]dz over the whole reaction zone. It is concluded therefore that the reaction CH+O→CHO++e− is indeed the source of chemi-ions in hydrocarbon flames. The rate constant was found to be 1.7×1011 mole−1 cm3 sec−1 at T=2000–2400°K. The ions are formed in a fairly wide region, extending from about the middle of the visible luminous zone to its outer edge. It is established that CH is not formed directly from CH3; instead, CH is derived from CH2 via CH2+H(OH)→CH+H2(H2O). The rate constants of these reactions were found to be about ten times smaller than the kinetic coefficient of the important CH-removal process CH+H→C+H2, which in turn is some twenty times larger than the rate constant of CH+O2→(products). Evidence has been obtained that the predominant source of CH2 in ethylene flames is the reaction C2H4+O→CH2+CH2O, which is shown to be only a few times slower at T=2000°K than the simultaneous process C2H4+O→CH3+CHO. In methane flames, CH2 is produced from CH3 via the reaction CH3+OH→CH2+H2O; its rate constant is nearly three times less than that of the reaction CH2+O2→(products), which in fuel-lean flames is the major CH2-removal path. The rate constant of the latter reaction was found to be about 1.2×1013 at T≅2000°K.

Efficient combination of surface and bulk passivation schemes of high‐efficiency multicrystalline silicon solar cells

Conventional and electromagnetically casted multicrystalline silicon solar cells are fabricated following different passivation schemes. Thin layers (∼100 A) of thermal dry and plasma‐enhanced chemical‐vapor‐deposition (PECVD) SiO2 are implemented for surface oxide passivation of multicrystalline silicon solar cells and compared. It is found that growing thin layers of thermal dry oxide results in efficient surface passivation. However, for thin PECVD SiO2 layers it is necessary to perform low‐temperature forming gas anneal, postdeposition, in order to observe the surface passivation effect. In addition, hydrogen plasma passivation has been optimized for achieving deep penetration of atomic hydrogen in the material (≳30 μm) and as a consequence very effective bulk passivation of multicrystalline silicon solar cells. By combining front and back thermal dry SiO2 passivation with hydrogen remote plasma treatment, a cell efficiency of 17% (independently confirmed) on 4 cm2 area and 180 μm thickness is realize...

Study of the Surface Reactions in ALD Hafnium Aluminates

Hafnium aluminates have been investigated as high-κ dielectrics for implementation in sub-45 nm nonvolatile memory technologies. The growth behavior and quality of these dielectrics strongly depend on the applied deposition technique. We examine the surface reactions that occur during the atomic layer deposition (ALD) of hafnium aluminates from HfCl 4 , Al(CH 3 ) 3 , and H 2 O. When grown in the ALD sequence (HfCl 4 /H 2 O) a [Al(CH 3 ) 3 )/H 2 O] b , HfCl 4 chemisorption is enhanced while that of Al(CH 3 ) 3 is inhibited compared to the reaction during ALD of the respective binary oxides. Density functional theory simulations suggest that this observation is related to the more efficient hydrolysis of the Al-C bond compared to Hf―Cl. However, earlier works on ALD of HfCl 4 /H 2 O suggest that the low growth per cycle of this process is not caused by a limited hydrolysis of the Hf―Cl bond but by dehydroxylation of the generated Hf-OH surface sites into less reactive Hf―O―Hf surface sites. Hence, we propose that the enhanced HfCl 4 and inhibited Al(CH 3 ) 3 chemisorption during ALD of hafnium aluminates results from a difference in dehydroxylation behavior between both binary oxides.