Carla Frege, Felipe Lopez-Hilfiker, Liang Zhu, Ben Bensaoula
TOFWERK, Thun, Switzerland
Semiconductor manufacturing often involves hundreds of processes. Between manufacturing stages, wafers are transported and stored in specialized plastic enclosures; Front Opening Unified Pods (FOUPs). Wafer defects have been related to increases in the time between processes (“queue times”) and the interaction of wafers with compounds that outgas from the surfaces inside the FOUPs . Precise and sensitive monitoring of outgassing compounds guides process adjustments to decrease defects and optimize FOUP cleaning processes. Measurements also inform the development of new FOUPs, using novel polymeric materials, and new surface treatment procedures . This work demonstrates use of a Semicon AMC Monitor continuously monitoring FOUP outgassing after standard cleaning procedures.
FOUP outgassing (~50 liters) was monitored using a Semicon AMC Monitor (Figure 1). The monitor directly sampled the air and instantaneously reported concentrations of trace organic and inorganic compounds.
Experiments were conducted by spraying a solution containing hydrochloric acid (HCl), hydrobromic acid (HBr), formic acid (CH2O2), acetic acid (CH3COOH), and nitric acid (HNO3) into the FOUP and then flushing with nitrogen to sanitize. The equivalent mass deposited from the solution ranged between 0.15 µg to 1 µg. Hydrofluoric acid (HF) was introduced using a permeation tube with an emission rate of 125 ng/min. The internal volume was flushed with a constant flow of N2 (2 L/min) to ensure the FOUP interior was well mixed and to simulate the cleaning of the FOUP container. This resulted in a FOUP ventilation rate of < 60 minutes.
The measurement protocol had three steps: (1) measure the FOUP background for 5 minutes, (2) place the HF permeation tube inside the FOUP for two minutes and immediately inject the acid solution, (3) continuously measure the mixing and subsequent decay of the compounds until concentrations return to background values.
After injecting the acid solution, the mixing took approximately 3-4 minutes (including evaporation of the injected solution) before the flushing initiated a decay of analyte concentrations. Figure 2a shows the decay of nitric acid and the reproducibility of acetic acid decay between repeat experiments (Figure 2b). All compounds showed a double exponential decay, with sticky compounds persisting at trace concentrations (10-100 pptv) even 100 minutes after injection. The double exponential fit (Equation 1) was used to retrieve the compound dependent time constants, which represent the flushing timescales of each compound from the FOUP. τ1 in equation 1 represents the e-folding time for the fast decay (gas volumetric exchange in the FOUP) and the second time constant (τ2) represents the slower outgassing from FOUP surfaces. The latter is significantly longer and depends on the interactions of the acid with the FOUP surfaces. Figure 2a shows an example of the double exponential fit for HNO3 which has a significant interaction with the walls of the FOUP and persists for much longer than the other acids.
Figure 3 shows the response of HF, HBr, HCl and HNO3 to nitrogen flushing over the first 45 minutes after reaching stable concentrations. Table 1 summarizes the time constants (τi) from the double exponential fits in Figure 3. Most of the acids have similar response in the first few minutes when volumetric flushing dominates, however, as shown in Figure 4, on longer timescales some acids persist at trace concentrations (10-30 pptv) for hours. Formic, acetic, hydrofluoric and hydrobromic acids all reached near background concentrations (90% decrease) in the first 60 minutes, implying no severe attenuation or memory on the inner surfaces (Table 2).
Table 1. Decay time constants (τ) for each acid shown in Figure 3. The values of τ2 are calculated according to the fit when the concentration starts to stabilize.
When using a Semicon AMC Monitor it is easy to estimate the optimal end point of a FOUP cleaning process. The slow decay of nitric and hydrochloric acids suggests that cleaning processes which are not optimized for slow acid outgassing, or that cannot sufficiently detect at low concentrations, could suffer from continued contamination risk. Table 2 summarizes the performance of Semicon AMC Monitors for the detection of organic and inorganic acids.
|Name||Formula||1 s LOD (pptv)||1 min LOD|
The capabilities of these monitors are well suited for both multi-port facility monitoring or mobile measurements. Outstanding detection limits and simple, autonomous operation presents a paradigm shift in the ability to quantify airborne and surface-bound AMCs at low concentrations as line widths are pushed to smaller dimensions.
- Jeong et al. Control of Wafer Slot-Dependent Outgassing Defects during Semiconductor Manufacture Processes. 2019. DOI: 10.1109/ASMC.2019.8791794
- Gonzalez-Aguirre et al. Control of HF Volatile Contamination in FOUP Environment by Advanced Polymers and Clean Gas Purge. 2015. DOI: 10.4028/www.scientific.net/SSP.219.247