ULTRAPURE FLUIDS

Evaluating the performance of dual PTFE filter assemblies in hot-acid etch baths

James Molinaro, SubMicron Systems, and Dennis Capitanio, Pall

Recirculating hot-acid baths are widely used in the semiconductor industry for wet-etching, stripping, and cleaning silicon wafers. To meet the stringent cleanliness requirements of the industry, particulate contamination generated during processing is removed from the baths by recirculating the etchant through filter assemblies using diaphragm or bellows pumps. Generally, the pump and filter assemblies are housed in secondary containment cabinets to ensure worker safety. The pumps, tubing, and filter assemblies are usually made from fluorocarbon compounds that are resistant to chemical attack, and the baths are fabricated from chemical-resistant materials such as quartz.

Because particles that adhere to the wafer surface during processing will increase the possibility of circuit failure and thereby reduce chip yield, the removal of contaminants from etch baths during the immersion of wafers is extremely important. The studies reported in this article examined the removal efficiency and temperature and viscosity independence of recently developed filter assemblies suitable for use in recirculating phosphoric and sulfuric acids. High-temperature (150°—180°C) phosphoric acid is used to etch silicon nitride layers on the wafer surface. Such layers act as a mask for selective oxidation of silicon or provide final passivation and protection for the integrated circuits.¹ Unprotected silicon nitride is selectively etched away by immersing wafers in a hot phosphoric acid bath. High-temperature (90°—150°C) 98% sulfuric acid is used in organic and hydrocarbon removal and in photoresist stripping or etching during wafer patterning. The common piranha etch consists of immersion in a sulfuric acid/peroxide bath at 150°C or in a sulfuric acid/ozone bath at 130°C. The stripping or etching speed is regulated by the bath temperature.

Experimental Setup

A Model SMS 178-02 hot-acid cabinet (SubMicron Systems, Allentown, PA) equipped with an all-fluorocarbon pump (Iwaki Walchem, Holliston, MA) was used for these studies. The cabinet was piped with 3/4-in. PFA tubing and contained patented ceramic fittings and a sensor that actuates an alarm in case of an acid spill. During each experiment two of the Super-Cheminert Kleen-Change filter assemblies being evaluated were mounted in parallel in the cabinet. Developed by Pall (East Hills, NY), the filter assemblies are composed of PTFE membranes and PFA hardware, which ensures their compatibility with the process acids. After passing through the filter assemblies the acid traveled through double-walled PFA tubing into the bottom of a quartz etch bath (Imtec, Sunnyvale, CA), and eventually flowed over a four-sided scalloped weir at the top of the bath. This bath design provides good turnover of the etchant and virtually eliminates any stagnant areas. The acid that overflowed the weir was returned to the recirculation pump in the acid cabinet for reuse. Acid temperature was controlled by a nitride etch controller (also provided by Imtec), and particles in the bath were sized and counted using a Model CLS-700 liquid sampler (Particle Measuring Systems, Boulder, CO). Featuring a sensitivity of 0.2 µm and 15 sizing channels with thresholds from 0.2 to 4.0 µm, the particle counter is able to draw a sample, size and count the particles within it, and return the sample to the bath every 2 minutes (±5 seconds). A schematic drawing of the system is presented in Figure 1.

Test Methods

Phosphoric Acid Study. To initiate the evaluation of the test filters' particle removal efficiency in phosphoric acid, two 0.2-µm filter assemblies were prewetted with isopropanol, flushed with ultrapure water, and mounted in the acid cabinet. The quartz etch bath was then filled with 21.5 L of 85% phosphoric acid (Ashland Chemical, Columbus, OH) and heated to 80°—100°C before the recirculating pump was turned on. The particulate contamination in the bath was then monitored until it reached a background level of <20 particles/ml 0.2 µm at 160°C.

For the first test, the bath was contaminated by a direct injection of 10 ml of a 0.01% dispersion of Georgia kaolin hydrite UF. This caused the contamination level to jump to 5000—20,000 particles/ml (0.2 µm). (The particle size distribution of the contaminant is illustrated in Figure 2.) Particle monitoring continued as the contaminant level decreased, and then for an additional 30 minutes after the previous background level was reached. The next phase of this study investigated the particle levels in the bath during nitride etching of wafers.

For the etch test, the background contaminant level of <20 particles/ml was achieved prior to the immersion of a boat containing six 6-in. wafers that had a 1200-Å-thick nitride coating. Etching continued for 40 minutes at 160°C, and particle levels in the bath were monitored throughout this time. After the wafers were removed, particle monitoring continued until the background level was again reached. A second boat of wafers was then introduced to the bath and particles were again monitored during the etching process. This testing provided a good indication of the contamination in the bath during nitride etching and the protection afforded by the filters against wafer particle defects.

Sulfuric Acid Study. Test methods for the sulfuric acid study were similar to those for phosphoric acid. In this case two 0.1-µm (rather than 0.2-µm) filter assemblies were prewetted with isopropanol and flushed with ultrapure water before being mounted in the acid cabinet. The quartz etch bath was then filled with 21.5 L of 98% sulfuric acid and heated to 80°C before the recirculating pump was turned on. For each test run, contaminant levels in the bath were monitored until they reached background levels of <20 particles/ml >= 0.2 µm at the temperature being studied.

The bath was then contaminated by injecting 10 ml of a 0.01% dispersion of Georgia kaolin hydrate UF directly into it. The injection caused the contaminant level to increase rapidly to 15,000—20,000 particles/ml >= 0.2 µm. Particle monitoring continued following the injection until 30 minutes after the background level was reached. This test was repeated four times, at injection temperatures of 90°, 110°, 130°, and 150°C, respectively, to determine whether filter performance was consistent at various temperatures and acid viscosities (which vary with temperature).

Test Results and Discussion

Both of the studies described above were concerned with the reduction of particulates in a recirculating bath. The particle reduction performance of a filter in a recirculating bath when there is no continuous introduction of contaminants can be defined by the following relationship:

C = C_ie ^{— (QEt/V)}

where C is the concentration of particles in the bath at time t, C^_i is the initial concentration of particles in the bath (i.e., at t = 0), Q is the flow rate in liters per minute, V is the bath volume in liters, t is the time after contaminant injection in minutes, and E is the reduction value. The equation can be applied to the contaminant injection portion of the studies, during which the reduction of particles in the bath following a single injection was measured as a function of time. Because bath volume (21.5 L) and acid flow rate (12 L/min) are also known, the reduction values (range = 0—1) can be calculated by fitting the experimental data to the equation.

Phosphoric Acid Study. The reduction value for the 0.2-µm filter assemblies recirculating 160°C phosphoric acid was determined from the particle count data shown in Figure 3. Fitting the curves to the equation yielded an average reduction value of 0.99 with a standard deviation of 0.07.

Particle levels in the bath before, during, and after the etching of two boats of nitride-coated wafers are shown in Figures 4 and 5, respectively. The figures show an initial burst of particles as etching starts, followed by a rapid drop as the acid recirculates through the filter assemblies. The particle levels reached the background level during the first 4—6 minutes of etching and remained there until the wafers were removed from the bath. If it is assumed that the wafers were being etched at a rate of 80 Å/min, the 1200-Å silicon nitride layer on the wafers would take about 15 minutes to etch completely.² As seen in Figures 4 and 5, after the initial burst of particles the filter assemblies were able to reduce the particle levels in the etch bath to background level despite the continuation of the particle-producing etch process. This removal capability means that, with the filters in place, wafers would not be exposed to a high level of particles during the etch process, and therefore the risk of particulate-related damage to the wafer surface would be minimized.

Sulfuric Acid Study. The calculated average reduction values for the 0.1-µm filter assemblies at 90°, 110°, 130°, and 150°C are plotted in Figure 6. The reduction value for a comparable 0.1-µm filter cartridge at 20°C determined during an earlier study is also represented in the figure.³ A linear regression performed on these five data points found the slope of the curve to be equal to zero. The viscosities of 98% sulfuric acid at the temperatures studied are also included in the figure. These study results clearly demonstrate that the particle retention efficiency of the filter assemblies in sulfuric acid is independent of temperature and viscosity over this temperature range.

Conclusion

Both the 0.2- and 0.1-µm paired filter assemblies that were evaluated rapidly reduced the level of particulates in the etch bath following contamination. The 0.2-µm filter assemblies tested with 85% phosphoric acid at 160°C exhibited excellent particle removal capability (99% at 0.2 µm): contaminant levels of 5000—20,000 particles/ml were reduced to background levels within a 10-minute period. The particle retention efficiency of the 0.1-µm filter assemblies tested in sulfuric acid did not vary significantly over the temperature and viscosity ranges studied. Highly contaminated baths (15,000— 20,000 particles/ml) were again brought back to very low counts within 10 minutes of contaminant injection. The 0.2-µm filters' ability to achieve and maintain low particulate levels during the nitride etching process minimizes the possibility that wafers would be exposed to particulates and suffer particle-related surface defects. In turn, the reduction of surface defects would lead to higher yields in the manufacture of semiconductor chips.

References

1. Wolf S, and Tauber RN, Silicon Processing for the VLSI Era, Volume 1: Process Technology, Sunset Beach, CA, Lattice Press, p 191, 1986.

2. Van Zant P, Microchip Fabrication—A Practical Guide to Semiconductor Processing, 2nd ed, New York, McGraw-Hill, p 227, 1990.

3. Capitanio D, and Gotlinsky B, "Methodology for Evaluating Particulate Control in Aggressive Chemicals for the Semiconductor Industry," in Proceedings of the 39th Annual Technical Meeting of the IES, Mount Prospect, IL, Institute of Environmental Sciences, pp 218—224, 1993.

James Molinaro is senior vice president of sales, service, and technology for SubMicron Systems, Allentown, PA. He is one of two cofounders of the advanced semiconductor and silicon wafer automated process equipment company, which was established in 1989. He has a BS in mechanical engineering and robotics from Pennsylvania State University and completed the executive development program at Wharton University. (Molinaro can be reached at 610/391-9200.)

Dennis Capitanio, PhD is a senior staff scientist in the scientific and laboratory services department at Pall Corp., Port Washington, NY. He is responsible for technical support for the semiconductor industry and has published articles on filtration and contamination control. He received his BS in chemistry from the University of Massachusetts, Amherst, and his PhD in physical chemistry from Northeastern University, Boston. (Capitanio can be reached via e-mail at Dennis_Capitanio@pall.com or at 516/484-3600.)

MicroHome | Search | Current Issue | MicroArchives
Buyers Guide | Media Kit