GaAs Surface Preparation for Thin Films Deposition Using Sodium Hypochlorite

Preparation of Gallium Arsenide (GaAs) surfaces, prior to a thin-film deposition, serves the prevailing functions of lowering defectivity, contamination removal and promoting adhesion. When native oxides, metallics or particulates are a concern, the common GaAs clean includes a dilute acid (HCl, HF) or base (NH4OH). When adhesion is the primary concern, the GaAs surface may need to be roughened in order to increase the exposed surface area. This etching is generally achieved using an oxidizing chemistry (H2O2) in conjunction with an acid or base (HCl, NH4OH). Unfortunately, a number of oxidizing chemistries are often too strong or unstable to be useful in particular processes, such as metal adhesion. Sodium Hypochlorite Bleach (NaOCl) is an oxidizing chemistry that attacks GaAs and is also fairly stable as a diluted solution. The condition of the resulting surface is an indication of process control and effectiveness and is measured using reflectivity. The purpose of this paper is to discuss the use of active scientific investigation to systematically define and implement changes to a critical NaOCl surface preparation process. Opportunities for improvement of the process were discovered through process stability and manufacturability issues. A hypothesis was formulated and focused on the theoretical breakdown of the bleach during the process. The NaOCl solution degraded at a very fast rate and allowed for only 3 hours of processing per bath, with low statistical process capability (SPC). The investigation began with a literature search into the breakdown mechanisms associated with NaOCl. Following the theoretical stage, wet chemical experiments were conducted in a laboratory. The results of the lab work proved that the breakdown mechanism discussed in research literature was occurring in the existing process. Additionally, this data was verified through experiments in the factory. Process improvement work was completed using experiments in the factory. This work included recipe alteration, bath preparation changes, chemical storage improvements, and in-tool chemical containment improvements. The results of this scientific investigation are improvements involving: a 40% decrease in qualification frequency, 83% decrease in cycle time, 300% SPC improvement, 50% cost decrease, 100% elimination of reworks, and 1200% improvement in bath-life. This paper concludes with a discussion on future process improvements with NaOCl as well as applications of the lessons learned to other processes that use unstable chemistries. The work discussed in this paper is an example of the successful application of the scientific method and active research used to determine a root cause and solution for a critical process issue. Introduction The surface preparation process that this paper discusses is a multiple iteration process that uses HCL and NaOCl. During this process, the wafers are exposed to HCL to remove native oxides, while the NaOCl is used to oxidize and smooth the exposed GaAs. The final iteration of HCL is to ensure all oxides have been removed. In order to determine if the wafers were processed correctly, samples out of each lot are measured for correct reflectivity. The reflectivity percentage determines the roughness of the GaAs surface. The process that prompted this change was rather cumbersome. The old process started with the preparation of an unstable and delicate NaOCl bath; including the qualification test, this first step could take up to approximately 80 minutes. Prior to the improvements, this process took 30 minutes per run and required a new bath to be mixed every four hours. A simple calculation based on the above numbers shows that each bath was only applicable for about four process runs. Due to all these limitations, research was conducted to make this process more robust. Background Bleach has a storied chemical past. In 1787, the French Chemist Berthollet discovered bleach’s ability to whiten textiles and in the late nineteenth century Louis Pasteur discovered the disinfectant properties of sodium hypochlorite. Bleach can be defined chemically as a strong oxidizing agent. Hypochlorites or peroxides often supply the oxidizing power. Commonly known bleaching agents include calcium hypochlorite, hydrogen peroxide and sodium hypochlorite. Sodium hypochlorite is regularly used as a germicide in water treatment, a medical disinfectant and a common household cleaner and disinfectant [1]. Sodium hypochlorite is a strong electrolyte that completely dissociates in solution. It then works by decomposing to liberate nascent oxygen, which is much more active than ordinary oxygen gas [2]: NaOCl Na + ClO NaCl + (O) Sodium hypochlorite is considered concentrated in a 5-7% solution and is commercially sold as such [3]. At a pH of 11 or above it is highly stable. However, there are several factors that lead to the chemical’s degradation; including changes in temperature, foreign metals, and exposure to air. Temperature and air exposure, specifically carbon dioxide in the air, are the two primary concerns in the GaAs preparation process [1]. Since the process was ran at ambient temperature, a hypothesis was developed stating that the primary factor in the breakdown of the bleach was air exposure due to turbulent agitation during the process. This hypothesis was reached through a literature search into the breakdown mechanisms associated with NaOCl. This degradation was inherent in the process because the recipe called for over 4 minutes of reclaimed bleach as well as a constant recirculation of the chemistry in the equipment tank. Based on this hypothesis, a project was developed with specific goals to improve bath life and reduce run-to-run variability while keeping the current reflectivity controls. The first step in the project was to recreate the process conditions in the laboratory. Both pH and concentration of the NaOCl were analyzed under varying concentrations and agitation conditions. Concentration was determined using an iodometric titration [4] and pH was determined using a meter. The agitation was created using a magnetic stir bar and those techniques were the determining factor in confirming the hypothesis. When the solution was stirred evenly (Figure1), the concentration and pH stayed relatively constant. However, when the solution was mixed vigorously (Figure2), the pH and concentration dropped drastically. This turbulent agitation infused air into the mixture and better replicated the process conditions in the fab. The degradation data can be viewed in Figures 3 and 4. pH vs. T ime 8 8.5 9 9.5 10 10.5 11