Overview

Process gas quality used in semiconductor manufacturing has always been of high concern. However, low volatility, high boiling point molecular contamination is rarely tested for in process gas quality control tests. Because this type of contamination can directly impact wafer and optical surfaces, it is extremely important to monitor and understand molecular contamination that may be contributed by process gases. The AiM-FT is a real-time, high sensitivity monitor that quantifies molecular contamination in process gases and distribution systems that affect critical product surfaces and equipment components.

Why is Monitoring Molecular Contamination in Process Gases Important?

Integrated circuits (ICs) are fabricated using complex equipment, gases, and chemicals. Although there are many steps involved, most of the steps and recipes are repeated many times throughout the process of constructing an IC. Because of this, however, if a process gas is contaminated with molecular contamination, wafers may be exposed to this contamination hundreds of times during the IC formation. Because these gas-borne imperfections can destroy circuits at any step in the process, gas "control" becomes perhaps the most critical contamination control issue that exists.1

For this reason, a strong degree of concern exists around process gas quality used in semiconductor processing. Ultra High Purity (UHP) process gases are very tightly controlled for certain contaminants such as water vapor, methane, carbon monoxide, carbon dioxide, helium, nitrogen oxide, etc. Yet significant consequences in process stability and yield result from process gas contamination containing much higher molecular weight contaminants. These gas phase contaminants have been labeled Airborne Molecular Contamination (AMC).

What is the difference between Airborne Molecular Contamination (AMC) and Surface Molecular Contamination (SMC)?

Airborne Molecular Contamination is a term that has been used as a "catch-all" for molecular contamination problems. It really represents contamination that exists in the gas phase. In general, it is represented as a gas concentration value (such as ppbv or µg/m3), but it does not give any indication as to how those gas-phase contaminants interact with critical product surfaces, such as semiconductor wafers, optical components, hard disk drive media, or TFT-LCD glass panels. If a very high level of contaminant exists in the gas-phase, but never interacts with product material or sensitive equipment, what yield or process stability benefit exists in monitoring that contaminant? However, if that contaminant does affect product material, there is significant reason to monitor the effect that contamination has on critical surfaces. Because the term and representation of AMC does not give any indication as to how molecular contamination affects these surfaces, Surface Molecular Contamination (SMC) has been adopted as the term that quantifies how gas-phase contamination interacts with critical product surfaces. SMC describes the adsorption, desorption, and reaction of contamination with these product surfaces, and thus is typically represented as a mass per unit area (ng/cm2 or ng/cm2/min). As an example, a 34-pptv gas concentration of dibutyl phthalate (DBP) in a CDA source may appear to be very clean. However, the daily adsorption of that concentration of DBP on a SiO2 surface is as high as 24 ng/cm2/day.2

The importance of monitoring for SMC caused by process gases is often recognized too late, only after processes begin shifting or equipment degrades. Effects of SMC on critical products include causing haze on optical components and process wafers, initiating corrosion reactions due to acidic species in process gases, and triggering detrimental interactions with wafer resists and thin-films. For example, it has been shown that residual organic contamination has a detrimental impact on electrical device performance during gate oxide formation. The updated 2003 International Technology Roadmap for Semiconductors (ITRS) current recommended limits for residual carbon from organic contamination prior to gate oxide is on the order of 0.36 ng/cm2. Based on the 24 ng/cm2/day contamination rate of 34-pptv DBP above, gate oxide breakdown voltages would begin to negatively shift after an exposure time of 21 minutes.3 Even trying to preserve the gate oxide integrity through the use of a nitrogen purge gas in a storage container (SMIF/FOUP) has been shown to be problematic. 4

Another area where molecular contamination in process gases is extremely damaging is in photolithography applications. For example, this contamination becomes a much larger concern as fabs transition from 248 to 193 and 157 nm exposure wavelengths. The recommended limit of purge gas contaminants significantly decreases based on the lithography wavelength as specified in the International Technology Roadmap for Semiconductors (ITRS 2003 Edition).

As discussed earlier, these ITRS recommended limits are based on gas-phase concentration and it is difficult to determine if these limits will truly protect optical and wafer surfaces without direct surface contamination measurements. Based on a linear scale of contamination rates, lithography optics exposed to 10-pptv (0.01-ppbv) of DBP would adsorb 7 ng/cm2/day, quickly leading to contamination at 157 nm.

In this case, interaction of organic or inorganic molecules with high-energy wavelengths can induce photochemical reactions that lead to the deposition of carbon or silicon based films on critical optical components. This is disturbing given the multi-million dollar replacement cost of photolithography optical systems, or the expense of equipment downtime in order to clean sensitive optical components.

Process gases are a typical source for this type of contamination. Process gases are not the only source however; another major source of contamination is often found in the gas distribution system. Contamination within the gas distribution system will contaminate otherwise pure process gases. The first challenge for users is to determine the required gas purity level for the process, and have that purity actually delivered to the site. The second challenge is then taking that known pure gas and storing, transporting, processing or using it without introducing contamination along the way.5 Therefore it is important to monitor both the process gas source, as well as the point of use (POU) location in order to identify baseline shifts or spikes of molecular contamination in process gases.

The AiM-FT molecular contamination monitor by Particle Measuring Systems is a real-time, high sensitivity monitor for SMC detection in process gases. Incorporating the use of a flow-through channel and a unique diffusion cell, process gas contamination can be identified with a time resolution of 1 minute. Sample sensitivity is 0.02 ng/cm2/Hz, allowing the detection of sub-ppb levels of molecular contamination.

AiM-FT Principles of Operation

The gas flow channel of the AiM-FT is uniquely designed to allow for gas phase molecules to diffuse into the sensor chamber and equilibrate with the Surface Acoustic Wave (SAW) sensor. These same contaminant molecules that are adsorbing onto the surface of the SAW sensor are also interacting with product wafers and photolithography optical components.

When contamination adsorbs onto the exposed SAW sensor surface, the frequency of oscillation is reduced; that frequency reduction is identified by comparing the exposed sensor frequency to a hermetically sealed SAW sensor in which no contamination can adsorb. In this manner, SAW sensors allow for real-time monitoring of increases or decrease in molecular contamination, allowing engineers to better understand how this gas-phase contamination affects product integrity and yield.

When monitoring molecular contamination in ultra-clean manufacturing areas, including process bays and process gas lines, guidelines exist for determining the cleanliness based on the daily molecular contamination rate.

Clean areas in fabs or POU gas locations are characterized by less than 0.2 ng/cm2/day of total adsorption of molecular contamination on a SiO2 SAW sensor.

Case Study: Semiconductor Quality Process Gases

Through the use of scrubbers, chemical filters, and gas purifiers, as well as the use of electro-polished tubing and orbital welding techniques, molecular contamination contribution from process gases to fabs is significantly less than in years past.6 These techniques used to minimize contamination are a necessity in order for semiconductor manufacturers to approach and continue through the 90 nm technology node.

An evaluation was undertaken at two major semiconductor manufacturers' in order to better understand the quality of process gas supplied to sensitive process equipment in the fab. In this case, the same company manufactured and supplied the process gases for each of the semiconductor manufacturers.

The experimental setup for this evaluation dictated installing the AiM-FT downstream of the process gas source and decontamination equipment (scrubbers, filters, purifiers, etc.) and upstream of the POU location. In this monitoring location, the process gases should be in their cleanest state. A section of piping was teed off and a pressure regulator and an AiM-FT were connected to the teed line. Nitrogen process gas quality was monitored at both gas delivery systems. Argon and Oxygen were monitored only at one gas delivery system.

As seen in the Figure 2, Gas Delivery System A has 145% more molecular contamination in the N2 gas being delivered to their process tools than Gas Delivery System B. In fact, the contamination in the N2 gas at location A delivers significantly larger contamination levels than what would be seen in a clean fab with ambient air. The contamination conditions of the process gases used at location B would be considered clean to moderately clean. By understanding that their POU process gases are not as clean as a competitor, engineers at location A may use this data to track down the source of impurity and eliminate it, thereby achieving process gas contamination parity with their competitor. Very small decreases in process gas molecular contamination at location A could also have a dramatic effect on yield and therefore the bottom line savings.

Case Study: House N2 vs. Bottle N2

Another evaluation was undertaken to evaluate the difference between a house N2 supply and a bottle N2 supply at the POU location. The two sources of N2 are purchased from different manufacturers.

The house N2 is supplied from a Liquid N2 storage tank behind the cleanroom fabrication facility, and has a 99.998% purity rating (four nines).

Bottled N2 is obtained from a local manufacturer with a 99.9995% UHP quality rating (five nines). The bottled N2 is contained and stored at high pressure (³ 2000 psi), and according to the manufacturers website, the impurities within this high-pressure gas N2 source are comprised of the following components listed in Table 2.

The AiM-FT molecular contamination monitor was placed at the point of use location, and the gas was distributed to the inlet of the AiM-FT in two manners. For the house N2, the distribution path consisted of the liquid N2 tank followed by a significant distance of brazed copper tubing, and just prior to the AiM-FT, a pressure regulator and 5-feet of Teflon tubing to reach the AiM-FT inlet. For the bottle N2, the distribution path consisted of the bottle supply followed by pressure regulator and 5-feet of Teflon tubing (the major difference between bottle and house N2 distribution systems being the long distance of brazed copper tubing the house N2 gas had to travel through). There are no chemical filters or gas purifiers in either case.

The contamination rate differences between the two sources of N2 are shown in Figure 3. Significant amounts of contamination are being delivered to the POU location in both cases, and are over two orders of magnitude higher than molecular contamination adsorption values typically seen in semiconductor process bays. Had either of these gas sources been purging optical components within a lithography tool, carbon bonding and deposition would occur rapidly on the surface in the presence of high-energy radiation. This type of contamination will increase equipment downtime and if conditions are severe, require replacement of multi-million dollar optical components.

In general, it would be expected that a higher purity gas would exhibit less molecular contamination than a lower purity gas. In addition, with the purity specifications of these two sources of N2, it may normally be assumed that the molecular contamination contribution would be minimal. Data analysis from the evaluation above tells otherwise. Engineers cannot assume that a higher purity gas will always contribute less molecular contamination, as it is shown that the House N2 is 30% cleaner than the Bottle N2, even though the Bottle N2 contamination specifications are purer. It also cannot be assumed that process gases that are labeled "UHP" are pure from a molecular contamination adsorption viewpoint, as both of these gases deliver about 100 times more contamination than that seen in the Semiconductor Quality Process Gas Case Study. If not quantified, molecular contamination in process gases will contribute large amounts of contamination that adsorbs to product material and sensitive equipment components, resulting in process instability and potential yield loss.

A number of components could contribute to the extremely high contamination rates seen over the course of this evaluation. For example, in the case of the bottle N2, the bottles are cleaned and washed on a set schedule, but not each time they are filled. Molecular contamination will accumulate on the inner lining of the bottle over time with repeated fills. Even the cleaning process can contribute molecular contamination if the cleaning solutions/solvents are not removed adequately. In addition, the quality of the internal bottle surface (surface passivation techniques and conditioning treatments) plays a role in how much internal contamination is adsorbed/desorbed from the bottle lining. The house N2 may be contaminated in a number of ways as well. For example, the fluxes and brazing alloys may still remain in the house N2 distribution system. These are generally hydrocarbon-based compounds, and they can take an extremely long time to be purged from a system (years). In addition, valves and regulators in the house N2 distribution system will contribute molecular contamination generally due to the wetting or lubricating compounds used in these components. These are some of the likely sources of additional molecular contamination, but there are undoubtedly many others.

Summary

Process gases and distributions systems can deliver large amounts of molecular contamination to sensitive wafer or optical surfaces, jeopardizing process stability and yield. Almost every step in the fabrication of integrated circuits utilize these process gases, either to provide purging environments, or for use during wafer processing. It is therefore critical to be able to quantify any molecular contamination contributions from process gas sources, distribution systems, and POU locations. In order to achieve this identification and quantification of molecular contamination, a low cost, high sensitivity, real-time SMC monitoring technique for process gases has been identified with examples from evaluations. This technique utilizes SAW sensor technology in order to monitor molecular contamination as it adsorbs not only on the sensor surface, but also on critical product surfaces, such as semiconductor wafers and photolithography optics.

References

1. Hogan, H. A Whole New State of Purity, Cleanrooms, Oct. 2002.
2. Akiyama, T., Takahashi, H., and Gomi, H., Monitor System for Clean Dry Environment Using Surface Acoustic Wave Sensor. 21st Annual Tech. Meeting on Air Cleaning and Contamination Control, Japan Air Cleaning Association, Apr. 2003, 46-48.
3. 2003 Updated International Technology Roadmap for Semiconductors, Front End Processes, 18.
4. Veillerot, M., et al, Testing the Use of Purge Gas in Wafer Storage and Transport Containers, Micro, Aug/Sept. 2003.
5. Lyons, M., Detecting Sources of Atmospheric Contamination in High Purity Gas Lines, Advancing Application in Contamination Control, Dec. 2002.
6. Ostrander, C., and Solcia, C., The Presence of Impurities in Ultrahigh-Purity Gas Distribution Systems: Case History Studies. Semiconductor Fabtech, 13th Edition.

AiM® is a registered trademark of Particle Measuring Systems, Inc.