(Download this paper for all tables and figures) (581.0 KB)

Introduction

Batch non-volumetric particle counters offer a rapid and cost effective requalification of wet process tools following chemical changes or filter replacement. In these applications, a particle counter is mounted on a cart and moved from one sample location to another. This mobile sampling approach maximizes the use of each particle counter, and provides particle contamination information in real-time for improved process tool utilization. Unlike traditional surface scan test wafers, non-volumetric particle counters characterize contamination in the individual baths being maintained. With a non-volumetric particle counter, samples can be taken directly from a process bath (Download paper to see Figure 1) (581.0 KB) without special modification of the process tool or lengthy down time. This article explains the use and benefits of portable liquid particle counters especially their application within the semiconductor and like manufacturing industries.

Batch Sampling Systems

Particle Measuring Systems' CLS 700T particle counter has successfully monitored a wide variety of processing chemicals. When sampling, chemical is drawn into the sampler under vacuum. Once proper fluid levels are reached the inlet valve is closed and the chemical is pressurized to eliminate bubbles1. After a user specified pressurization delay, the chemical is forced through the particle counter at a constant pressure and flowrate. Several different particle counters may be used in the batch system to match particle sizing requirements of the chemical or process being monitored.

The great flexibility provided by the batch sampling particle counter system allows the units to perform well in a number of varied applications. Their portability and bubble suppression capability are very useful in spot checking chemical delivery systems to verify chemical cleanliness before it can impact production yields.

Non-Volumetric Particle Counters vs. Test Wafers

Like any complex device, wet process tools must be maintained to replace consumables, repair worn parts or upgrade system components. The most frequent forms of maintenance are chemical and filter replacement. After performing this type of maintenance, the process tool is brought on-line to recirculate chemical through the integrated filtration system. Historically, after the recirculation system is running, particle test wafers are prepared and run through the various process baths. The test wafers are scanned with a surface particle counter and the number of particles added to their surfaces are recorded. If the total particle added are within quality limits, the process tool is qualified to run product again. If the total particles added exceed quality control limits, the test process is repeated before system troubleshooting begins.

Particle test wafer procedures are very resource intensive on personnel, equipment and raw materials. A typical surface scan procedure include the following steps:

1. Pre Scan
   2. Run Wafers Through Baths
   3. Post Scan
   4. Repeat if Test Failed
   

Particle test wafers are normally run through several process baths rather than the one or two which are being maintained. This procedure takes time and can mask problems in individual baths. The CLS 700T particle counter offers a faster and more cost effective approach to tool qualification. Following tool maintenance, a portable CLS 700T can be installed to sample directly from the process bath being maintained.

The CLS 700T particle counter can begin measuring particle concentration within the bath in a matter of minutes. When chemical begins flowing through the re-circulated filtration system, there is a marked decrease in particle contamination.

When the particle concentration drops below quality assurance limits, the tool can be qualified to run product again. As illustrated in figure 2, the CLS particle counter identified bath particle concentrations were within qualification limits well before surface scan data would typically be available. If a problem occurs in the recirculation system or bath, the particle concentration does not drop normally (figure 3).

Figures 2 and 3 (Download this paper for all tables and figures) (581.0 KB)

These problems are quickly identified with the CLS 700T particle counter and can prevent product contamination, or additional lost processing time spent rerunning test wafers.

The CLS 700T particle counter's ability to measure particle concentration changes in real-time naturally lend themselves to continuous process monitoring. As process monitors, CLS 700T particle counters are very helpful in detecting catastrophic failures in fluid handling systems (figure 4). Once notified of the system failure, operators can react to correct the problem before significant loss of product. Continuous process monitoring is best accomplished through dedicated CLS 700T particle counters rather than mobile units.

System Hardware Configuration

The portable CLS 700T particle counter system is mounted on a cart so that the sampler is above the fluid level in the sample vessel or bath. This positioning prevents the possibility of forming a siphon in the event of a catastrophic plumbing failure. The sampler component of the CLS 700T particle counter is placed in a double containment vessel as an added safety precaution. All sample tubing, including the inlet, outlet and purge should be contained to protect the operator and equipment from chemical contact.

While sampling, the system as a whole must be protected from disturbance. If the sample lines are jarred or otherwise disturbed, particles can be shed from the fittings or the tubing walls, producing artificially high particle readings. These precautions ensure operator safety and maximize data accuracy and repeatability.

System Operation

When sampling, the system should use the same flowrate at which the sensor was calibrated. Although the system may be calibrated at slower flow rates, most semiconductor processing chemicals have not required it. It is common for the sensor flowrate to vary with chemical viscosity and compression pressure. Flowrate is displayed on the controlling software and may be adjusted as needed. Flowrate deviations less than 10% from the calibrated flowrate do not cause significant particle sizing errors.

The sample compression pressure should be set to the lowest pressure required to eliminate bubbles. This pressure setting will vary for different chemical combinations, dilutions and temperatures. To determine an appropriate compression pressure follow these steps while sampling a relatively stable sample source:

1) Set at least 5 particle size bins covering the sizing range of your CLS 700T particle counter (0.2, 0.3, 0.5, 1.0, 2.0) or (0.3, 0.5, 0.7, 1.0, 2.0).
2) Increase the compression pressure to 60 PSI and adjust the flow to remain constant.
3) Run the system long enough to collect a statistically significant number of samples (20 or more).
4) Reduce the system compression pressure by 5 PSI.
5) Repeat steps 3 and 4 until you observe increased counts in the larger size channels.

Bubble formation can be identified by an observed particle size distribution shift.

As illustrated in Figure 5 (Download paper to see all Figures) (581.0 KB), bubbles appear as particles in the larger size bins where none were present with higher compression pressures. If you observe a particle size shift you are below your optimal compression pressure. Increase the compression pressure until a normal particle distribution is observed (figure 6). If the system has been used on dirtier processes, increasing pressure can cause some initial particle shedding as tubing and fittings expand. Allow the particle counts to stabilize before adjusting the pressure.

If bubbles are still apparent even with maximum sampler compression pressure, lengthen the compression delay. A longer compression delay gives the bubbles more time to return to solution. Compile a table of appropriate compression pressures for each of your chemicals, and use it to prepare the system for sampling.

Care must be taken when choosing a sample point within a bath. If the bath is recirculating, areas of increased flow, or near filter effluent cleanup faster and therefore produce lower particle readings. A sample point should be chosen which is representative of what the product will come in contact with, or the worst case of what the product will contact with.

If the system will be sampling while the process tool is in operation, the sample tube must be positioned to prevent interference with wafer carriers. As shown in figure 7, the sample tube is normally positioned along the side of the bath mid-way to bottom. The drain and vent tubes are positioned to expel fluids into either a waste recovery system or the overflow weir on a recirculated chemical bath.

System Contamination

When the CLS 700T particle counter is not in-use, it should be plumbed into a clean DI source and set to flush mode. This process helps purge the system sample lines of any remaining chemicals and contamination, and prevents any organic growth. It is important that the system sample lines not be allowed to dry out. If chemicals are allowed to dry inside the sample lines of the CLS 700T particle counter, it becomes very difficult to clean the lines, and may require significant (and costly) factory maintenance. The exterior of the sample tube must be protected from contamination when not in-use. If you suspect contamination on the exterior of the sample tube, wipe it down before inserting it into the bath. Wipe away from the free end, not toward it.

Quicker clean-up times may be gained by dedicating samplers to specific chemistries. This is not the most cost efficient approach from an initial equipment expenditure point of view, but it will improve cleanup times and lengthen maintenance cycles. If this approach is too costly, dedicate one sampler to fairly dirty chemistries, and another to the cleaner chemistries. Figure 8 highlights the effect this approach has on sampler clean-up times.

The sampler dedicated to cleaner processes cleans-up much faster than the general purpose sampler. This approach will optimize equipment utilization and reduce time requirements for bench qualifications.

Safety

When changing sample chemistries, care must be taken to completely clean the system. The recommended procedure for removing caustic chemicals is to go through several increasingly lean dilutions before pure DI water is introduced into the sampler. If this is not practical, the sampler should be purged with pressurized nitrogen for at least 30 seconds, then pause for an additional 30 seconds. This sequence should be repeated twice, before drawing DI water into the sampler for rinsing. If insufficient cleaning is performed, the sample tubing may become "crazed" (this resembles fogging). The fog is actually many micro fissures, which act as particle traps and can seriously degrade a system's ability to clean-up quickly. If the crazing becomes serious, the fluid level detectors inside the sample may fail.

Conclusion

Using the CLS 700T particle counter from Particle Measuring Systems can maximize wet process tool utilization by reducing requalification times, and more accurately characterizing the cleanliness of individual baths. Modern non-volumetric particle counters can reliably overcome variances in chemical viscosity and bubble formation to accurately size and count particle contamination. Through proper equipment use, portable CLS 700T particle counters can provide an accurate and cost efficient means of testing many process tools.

(Download this paper for all tables and figures) (581.0 KB)

Contact us if you need more information or have questions.

References

1. L. Owens, R. Hendrix, B. Witowski, "The Use Of Insitu Liquid Particle Counting in Pre-Diffusion Cleaning, " Proc. of 41st Tech. Conf., Institute of Environmental Sciences, May 1995.

1. S. Rush, T. Ford, "Adopting a Particle Monitoring Program For Hot-Liquid Baths: A Case Study", Microcontamination, May 1994
2. J. R. Mitchell, B. A. Knollenberg, New Techniques Move In Situ Particle Monitoring Closer to the Wafer, Semiconductor International, Sept. 1996

Published: 1997