(Invited) Wet Etching inside Advanced High Aspect Ratio Structures: Impact of Dissolved Oxygen

Abstract

Listen

Translate article

Semiconductor technology is currently facing the challenge of processing structures with nanometer scale openings and high aspect ratios (HAR) in order to enable the manufacture of 3D nano-devices like FINFETs, 3D-NAND cells and DRAM capacitors. Simultaneously, process control has been pushed towards sub-nanometer dimensions as 1nm variation starts to impact device performance. These requirements pose a growing challenge to wet processing inside nanometer scale openings in HAR structures of advanced 3D devices. Top to bottom etch uniformity with nanometer etch control has become a paramount necessity. High selectivity silicon etch is a critical process for 3D-NAND. Typical silicon wet etch processes have very high selectivity towards silicon dioxide. Even a monolayer of silicon dioxide can affect the silicon etch rate. Dissolved oxygen (DO) has become a key aspect during silicon etching. The dissolved oxygen can react with the exposed silicon creating silicon dioxide slowing the silicon etch. In our paper, we specifically looked at the etching of poly Si by tetramethylammonium hydroxide (TMAH) and how the etch rate is affected by DO in the solution. Figure 1a shows experimental results of etching a blanket poly Si wafer depending on the oxygen in the environment: normal air or pure nitrogen. Even though the TMAH is dispensed with a low amount of DO, as it flows across the wafer it can absorb or desorb oxygen to/from the atmosphere and accordingly change the center to edge etch rate considerably. In order to test if the hypothesis was correct, we modeled the DO concentration in the TMAH solution and predicted the DO profile inside HAR structures. Figure 1c shows the plan for modeling the DO in order to predict the etch rate variability: a liquid flow model predicts the temperature and DO across the wafer then this model feeds into an additional model for the DO concentration inside the HAR feature. These models then report the etch rate as a function of wafer position and depth in feature, so long as the etch rate as a function of temperature and DO is known. Table 1 shows the equations for the wafer surface liquid model. Equation 1 calculates the temperature and volumetric flow rate across the wafer and equation 2 uses the results of equation 1 to calculate the DO concentration. Table 2 shows the chemical reactions and rate equations for modeling the DO concentration inside a HAR feature, where the DO was consumed as it diffuses down into the feature and reacts with the poly Si. The rate of reaction 1b is assumed to be much faster than 1a, so the important rate equations are 1a, 2, and 3 with their associated rate constants. Figure 1b shows the results of the wafer surface liquid model in predicting the DO concentration and resulting etch rate across the wafer. The process environment O2 was varied in which the atmosphere is either 0.1% O2, standard 20% O2, or a carefully controlled 6% O2 chosen to maintain a constant DO in the TMAH across the wafer, even with temperature variation. However as we will show in the full paper, center to edge etch rate uniformity is insufficient for good process control, as DO consumption within HAR features causes very different etch rates between the top and bottom of the HAR structure. We will then show that this difficulty can be solved by using a much more powerful oxidizer than DO such as H2O2, ensuring a uniform etch rate across the HAR. Figure 1