(Download this paper for all tables and figures) (712.5 KB)

Introduction

The purity of DI water and chemicals used in the semiconductor industry has improved to such an extent that they are almost particle-free environments. The ITRS roadmap regarding DI water for the future shows continued decreasing line widths and a corresponding decrease in the critical particle size.

The increasing cleanliness of DI water and other liquids and the decreasing critical particle size has caused a corresponding increase in particle counter capability and cost. It is, therefore, important to balance cost against features. The critical issue when purchasing a particle counter for DI water or ultra clean fluids is whether the objective is to monitor only for alarming of extreme out-of-spec conditions or to monitor both for alarming as well as trend analysis.

(Download this paper for all tables and figures) (712.5 KB)

Today, particles are typically monitored at 0.05 microns in DI water and 0.065 microns in liquid chemicals. There are relatively few particles and it is not uncommon to encounter particle concentrations of these sizes at less than 1 particle/mL or even less than 0.2 particles/mL. The high sensitivity particle monitors required to detect these very small particles have a significant drawback in that they only examine a very small percentage of the total fluid, such as DI water, flowing through their sample cells. The smaller the amount of DI water sampled, the less data you collect, and the longer it takes to gather sufficient data to be statistically significant. Therefore, if the purpose of monitoring the DI water is for alarming purposes only, a lower flow particle counter such as Particle Measuring Systems' HSLIS M50e DI water monitor is adequate. But if the desire is to monitor data trends, then a higher flow particle counter such as Particle Measuring Systems' Ultra DI® 50 DI water monitor is required.

Monitoring for Alarming Only: HSLIS M50 DI Water Particle Counter

Instruments with small sample volumes can be used to detect extreme out-of-spec problems in DI water. The HSLIS M50e DI water particle counter (Figure 1) has a flow rate of 100 mL/min but only examines 0.25% of this flow. This results in a measurement sample volume of 0.25 mL/min. Various competitor products in the market examine even less, typically 0.1 mL/min. This sampling rate affects the length of time required to gather results and, therefor,e the amount of DI water sample monitored over a specific period.

Assuming the particle concentration of the DI water is 0.5 particles/mL >0.05 microns, what size of sample is needed to collect meaningful data? In other words, what DI water sample size is representative of the DI water tested? Sample size and analysis time are inextricably linked. If the analysis time is too short and very few particles from the DI water are counted, then the sample-to-sample repeatability will be very poor. If the analysis time is too long, then the analyst will be blind to changes in the process between sample intervals.

These concerns can be addressed with a simple examination of expected purity of the DI water or other liquid chemical, the sample volume of the particle counter, and particle counting statistics. Particle counting statistics predicts the repeatability when a certain number of raw events are measured. The precision expected is the square root of the number of raw counts divided by the total number of raw counts.

(Download this paper for all tables and figures) (712.5 KB)

Precision Percentage = (the square root of the number of raw counts) divided by the number of raw particle counts

For example, if a 20% relative standard deviation is desired, then 20 counts must be measured during the sample interval (square root of 20 is 4.47 and divided by 20 = 22%).

The conditions cited above (0.5 cts/mL concentration and a sample volume of 0.25 mL/mn) predict that we would need 160 minutes to detect 20 particles in the DI water. It should be noted that only 40 ml of DI water would be examined during this entire time. Since the time interval is too long to be practical, the HSLIS M50 DI water particle counter is more suitable for alarming, rather than trending.

Cosmic Rays

Cosmic rays also affect results. Cosmic rays impact the detector about once per minute and are indistinguishable from particles due to the nature of the optical particle counter (i.e.: examining light scattering produced from particles). Although the particle counter anticipates and corrects for this, the measurement accuracy is affected especially for units with low sample volumes. This further deteriorates the quality of the data for low sample volume counters and effectively eliminates them from providing useful information for monitoring data trends.

Particle Trend Analysis

Collecting repeatability data for particle trend analysis requires more raw counts per minute than low sample volume DI water particle counters can provide. DI water Particle counters with medium sample volumes of 1 to 10 mL/min can meet this need, however the Ultra DI50 DI water particle counter (Figure 2) with a sample volume of 3.75 mL/min is in this category. In the example cited earlier, the time to measure 40 mL of DI water or other fluid (for an average of 20 particles) is only 10 minutes with a Ultra DI50 DI water particle counter. This is contrasted to an alarming device such as the HSLIS M50 or the Ultra DI50 particle counter which requires 160 minutes to produce the same data (see the table below).

The table below highlights the differences between the technologies available today. (Download this paper for all tables and figures) (712.5 KB)

The ideal DI water particle monitoring strategy is to combine alarm devices such as Particle Measuring Systems' HSLIS M50 and higher precision particle monitors such as the Ultra DI50 DI water particle monitor.

Conclusion

Before purchasing a particle counter to monitor DI water it is important to understand the goal of particle monitoring. If the goal is to provide particle trend analysis and to be alerted in cases of out-of-spec conditions, then the particle monitor must not only have high sensitivity, but also a high sample volume. However, if the need is simply an alarm when the liquid conditions are extremely out-of-spec, it is more economical to purchase a particle counter such as the HSLIS M50e DI water monitor, with a smaller sample volume.

(Download this paper for all tables and figures) (712.5 KB)

Contact us if you need more information or have questions regarding choosing a DI water particle monitor or understanding the data from your particle counter.

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

Published: 2005

Author: Ed Terrell, Particle Measuring Systems