Comprehensive methods for contamination control in UHP fluids.

摘要

The demand for high performance electronic devices is ever increasing in today's world with advent of digital technology in every field. In order to support this fast paced growth and incursion of digital technology in society, smarter, smaller integrated circuits are required at a lower cost. This primary requirement drives semiconductor industries towards the integration of larger number of smaller transistors on a given circuit area. The past decades have seen a rapid evolution of material processing and fabrication techniques, as focus shifts from submicron to sub-nanometer length scales in device configuration. As the functional feature size of an integrated circuit decreases, the threshold of defect causing impurities rises drastically. Huge amount of resources are spent in downstream and upstream processing in order to restore system from contamination upsets and in the upkeep of Ultra-High-Purity (UHP) process streams to meet these stringent requirements. Contamination once introduced into the system also drastically reduces process yield and throughput resulting in huge losses in revenue. Regular UHP fluid distribution system maintenance as well as restorative operations involve a purging operation typically known as Steady State Purge (SSP). This purge operation involves large amount of expensive UHP gas and time. Depending on the scale of the system and type of process involved this results in significant tool, process downtimes and can have a wide range of environment, health and safety (ESH) ramifications.;A novel purge process, referred to as Pressure Cyclic Purge (PCP) was studied for establishing gas phase contamination control in UHP applications. In understanding the basic mechanism of this technique and to analyze its extent of application in aiding purging operations, a coupled approach involving experimental investigation and computational process modelling was used. Representative and generic distribution sections such as main supply lines and sections with laterals were contaminated with a known amount of moisture as impurity. The dynamics of the impurity transport through the system from purging with SSP as well as PCP was captured by a highly sensitive analyzer. The surface interactions between the moisture and EPSS were characterized in terms of adsorption and desorption rate constants and surface site density. A computational process model trained using experimental data was then validated and used to study the steady and cyclic purge mechanisms and predict complex purge scenarios. Industrially relevant and applicable boundary conditions and system definitions were used to increases the utility of the computational tool. Although SSP compared closely with PCP on simple systems without laterals, a drastic difference in dry-down efficiency was noticed in systems with dead volumes in the form of capped laterals. Studies on system design parameters revealed that the disparity in performance was observed to increase with larger number and surface area of dead volumes, opening a path to critical understanding of the differences in process mechanisms. Beneficial transient pressure gradient induced convective flow in the dead volumes during cyclic purge was identified to be the main factor driving the enhanced dry down rate. Similar trends were observed on using surface concentration as the purge metric. Hybrid purge schemes involving a combination of SSP and PCP were found to yield higher benefit in terms of efficient use of purge gas. Removal of strongly interacting contaminant species showed a higher benefit from use of controlled PCP scheme. Although, parametric analysis carried out on the operating factors of cyclic purge suggested that the enhancement in dry down increased with higher pressure range, it was highly conditional towards configurational factors in design and operation such as system dimensions, holding time, cycling pattern, valve loss coefficients and the complex inter coupling between them. The robustness of the process simulator allows the development of optimal purge scenarios for a given set of system parameters in order to perform a controlled purge. The benefit of using a hybrid PCP scheme was evaluated in terms of UHP purge gas and process time as a function of purity baseline required.;Apart from UHP gas distribution systems, process vessels, chambers and components along the process stream are also prone to molecular contamination and pose a threat to product integrity. The dead volumes acting as areas of contaminant accumulation represent cavities or dead spaces in flow control elements such as mass flow controllers (MFCs), gauges, valves or dead spaces in process chambers. Steady purge has very little effect in cleanup of such areas and more efficient methods are necessitated to raise purge efficiency. The analysis of application of PCP is extended to such components through the development of a robust and comprehensive process simulator. The computational model applies a three dimensional physical model to analyze purge scenarios with steady and cyclic purge. The results presented pertain to any generic gas phase contaminant and electronic grade steel surfaces. Close investigation of the purge process helped elaborate the cleaning mechanism. Critical steps driving the purge process were identified as - dilution of chamber by introduction of fresh gas during re-pressurization and chamber venting during depressurization. Surface and gas phase purging of chambers with dead spaces using steady and cyclic purge were studied and compared. Cyclic purge exhibited a higher rate of dry down. The effect of system, design and purge operating parameters on surface cleaning were studied. Although higher frequency cycles and larger operating pressure ranges optimized for a given geometry are found to deliver better pressure cyclic purge (PCP) performances, the benefit is found to be contingent to a strong interplay between system parameters. PCP is found to be advantageous than steady state purge (SSP) in terms of purge gas usage and operation time in reaching a certain purity baseline.;Specialty process gases supplied to the fabrication facility are typically stored in the form of liquids in enormous tanks outside the fab. Ammonia is a widely used in UHP concentrations for a variety of process including epitaxial growth, MOCVD, etching and wet processes in the semiconductor industry. The recent development in LED research has risen the demand and supply for Ammonia based compounds. Stringent baselines are maintained for the impurities associated with the manufacturing of such gases (e.g. Moisture in Ammonia). Apart from the difference in the rates of evaporation of the individual species from the storage cylinder causing accumulation of slower evaporating species, external temperature fluctuations also generate unsteady flux of desired species. When concentrations rise above this threshold additional purification or in most cases discarding large volumes of unused gas is warranted, causing loss of resources and causing ESH issues. Bulk gases are usually delivered over long lengths of large diameter pipes which produce large density of adsorption sites for contaminants to accumulate and eventually release into the gas stream. In order to establish contamination control in the gas delivery system, the surface interactions of the multispecies system with the delivery line surface was characterized. Desired concentrations of moisture in ammonia and UHP nitrogen mixtures were produced in a gas mixing section capable of delivering controlled mass flow rates to an EPSS test bed. Transient moisture profiles during adsorption and desorption tests at various test bed temperatures, mass flow rates and moisture concentration were captured by a highly sensitive analyzer. A mathematical model for single and multi-species adsorption was used in conjunction with experimental data to determinate kinetics parameters for moisture, ammonia system in EPSS surface. The results indicate competitive site binding on EPSS between ammonia and water molecules. Also, the concentration distribution of each species between surface, gas phase is interdependent and in accordance to the kinetic parameters evaluated.;Back diffusion of impurity is a major source of contaminant introduction into UHP streams. Back diffusion refers to the transport of contaminants against the flow of bulk process stream. Molecular species can back diffuse from dead volumes, during mixing operations etc., simply when there is a gradient of concentration. A steady state approach was used to analyze the mechanism and effects of various geometrical and operational parameters on back diffusion. High sensitivity moisture detectors were used to capture the dynamics of contamination in a section of a generic distribution system. Results showed that back diffusion can occur through VCR fittings, joints and valves under constant purge. General trends on the effect of design parameters on back diffusion were derived from studies on various orifice sizes, system dimensions, flow rates and test moisture concentrations. Coupled parametric studies helped identify critical variable groups to perform dimensionless analysis on back diffusion of moisture. Crucial points where back diffusion can be minimized or completely eliminated are identified to help set up guidelines for cyclic and steady purge parameters without excessive use of expensive UHP gas or installation of unnecessarily large factors of safety.;Wet cleaning of micro/nano sized features is a highly frequent process step in the semiconductor industry. The operation is a huge consumer of ultra-pure water and one of the main areas where process time minimization is focused. Comprehensive process model is developed to simulate the mechanism and capture the dynamics of rinsing high aspect ratio Silicon features in the nanometer scale. Rinsing of model trench, post etch contaminated with ammonium residue is studied. Mass transport mechanisms such as convection, diffusion are coupled with surface processes like adsorption and desorption. The effect of charged species on the trench surface and in the bulk, the resultant induced electric field on the rinse dynamics and decay of surface species concentration is studied. General rinsing trends and critical points in change in mechanisms were identified with critical groups such as mass transfer coefficient and desorption coefficient. The model is useful in evaluating process efficiency in terms of rinse time and DI water consumption under varying process temperature, contaminant concentration, and rinse fluid flow rate. The generic build of the model allows extension of its functionality to other impurity-substrate material couples.