Simplicity in AMC monitoring

Over the past five years, airborne molecular contamination (AMC) has moved to the forefront as one of the hot topics in the semiconductor industry. Leading edge semiconductor manufacturers processing 300mm wafers and using 193nm exposure wavelengths for photolithography are seeing some nasty effects from this gas-phase form of contamination.

Ammonium sulphate hazing on 193nm optical elements and reticles are wreaking such havoc, that a competitive advantage is gained by effectively being able to monitor for and reduce this form of contamination.

Over the past several years, different sensor and detection technologies have been adapting themselves by increasing their complexity, to meet the requirements needed for effective AMC monitoring.

Chemiluminescence, ion chromatography, pulsed fluorescence, cavity ring-down, and ion mobility spectrometry are but a few of the technologies that have been implemented for the detection of AMC.

While each of these techniques has its strengths and weaknesses, one critical attribute stands out when considering AMC monitoring techniques — simplicity.

Complex designs only add to the uncertainty in the measurement by introducing opportunity for the sample to be altered prior to detection, or creating potential hardware failure points which require regular preventative maintenance.

Design attributes

In evaluating the simplicity of each of these AMC monitoring techniques, table 1 provides a direct convenient comparison of the relevant design attributes of a contamination monitor. The features listed are considered in detail here.

Sample tubing before the detector: Gas-phase molecules are in constant motion at standard temperature and pressure. They are constantly colliding, reacting, or interacting with other molecules or the container in which they are enclosed, which, in this case, would be the sample tubing upstream of the monitor.

For a given size diameter of sample tubing, longer lengths of tubing have a larger surface area for molecular contamination to interact with by either collecting (or sticking) to the wall surface, reacting with the material the tube is made of, or reacting with other contamination that has collected on the tube walls.

Consider a monitor that uses 6 inches (15.4cm) of 1/8" (3mm) outer diameter tubing (such as an IMS analyser) compared with a molecular analyser pulling from a sample manifold using 100 m of 1/4" (6.35mm) outer diameter tubing. The surface area inside of the IMS analyser is about 1.2 in2 compared with a surface area of 2326.4 in2 for an analyser with a sample manifold and 100 m of tubing. Those three orders of magnitude difference can have a detrimental impact on the quality of results that the analyser is expected to report.

From a monitor perspective, longer sample tubing, or greater surface area, ultimately lead to longer response and cleardown times, or even missed contamination events. For example, consider a 0.5 ppbv ambient NH3 concentration sample being pulled into an AMC monitor. If that ambient air changes quickly from 0.5 ppbv of NH3 contamination to 15 ppbv of NH3 contamination, the monitor should immediately read 15 ppbv. However, because sample tubing upstream of the detector needs time to equilibrate to the new conditions, the monitor will detect an exponential decay of concentration up to the 15 ppbv level, which will be proportional to the length of sample tubing. Thus, longer lengths of sample tubing will cause a delayed response to contamination changes, whether that change is from a response to a high concentration event or a cleardown due to cleaner environments.

Sample manifolds: As noted in table 1, sample manifolds have made their way into AMC monitoring techniques over the past few years. While there are benefits to sample manifolds, such as reducing the cost per sample point, there are many disadvantages as well. Most sample manifolds use a valve or combination of valves to sequentially sample from many different locations. Each sample is transported to a central location where the AMC monitor is located, and software outputs the concentration of contamination at that sample location prior to moving on to the next location. One drawback is monitor equipment downtime (scheduled or unscheduled).

By having a central analyser, when that analyser goes down due to equipment failure or preventative maintenance none of the sample points can be monitored. The high cost of many of these central analyser technologies do not allow for a spare analyser to be on hand for a "hot-swap".

Point-of-use analysers do not suffer from this issue, as each monitor is located at its actual sample point of interest. Should a point-of-use monitor need repair, it can be simply swapped with another analyser to ensure the critical point will always be monitored. Moreover, the low cost of point-of-use analysers allow for a hot spare to be available to rotate in or out should the need arise.

Purging time

Additionally, because sample manifolds are typically sequential, time must be allowed to purge the common sample lines when rotating from one sample point to the next. This is especially important when moving from high concentration sample areas to low concentration sample areas. Because there will always be some amount of “common” sample line in-between the manifold and the detector that all sample travels through regardless of which sample port is actively being sampled, time spent purging must be endured with manifold systems.

A typical purge time at each sample point is no less than nine minutes to allow thorough evacuation and equilibrium of the common lines and detection system. Because the measurement, may contain remnants of contamination from the previous sample location, time spent purging is meaningless and does not represent the true concentration of AMC at the sample location. After purging, a sample is typically analysed for a one-minute period of time, after which the manifold moves to the next sample point and continues the purge/sample sequence.

The amount of time sampling with real-time point-of-use monitors as compared to manifold sampling systems is clearly illustrated in figure 1. For example, a 60-point manifold will only capture a true representative sample for three minutes a day, with the remaining time spent either purging that sample line or sampling from other locations. A 16-point manifold will only capture a true representative sample for nine minutes each day. The continuous point of use monitor on the other hand samples from a specific location for 1,440 minutes each day; other locations are also monitored for 1,440 minutes per day with dedicated point of use monitors.

Upstream components: Many AMC monitors have evolved from old monitoring techniques, with increasingly complicated systems put in place to make up for weaknesses in those techniques. Often, hardware must be installed upstream of the detection system in order for the sample to be properly conditioned or setup prior to entering the detection mechanism.

Maintenance

As seen in table 1, hardware can consist of valves, pumps, catalytic converters, etc. These components, and all hardware, are potential failure points within the monitor, and often require preventative maintenance. Mean time to failure (MTTF) and mean time between failures (MTBF) are often much shorter with additional upstream hardware, and mean time to repair (MTTR) is normally much longer. Being prepared for these failures and adding the cost of preventative maintenance can significantly increase cost-of-ownership of the AMC monitor.

With the sample interacting with each of these pieces of hardware, another opportunity exists for the sample to be changed prior to reaching the AMC monitor. Even more, these additional hardware items can contribute to contamination being detected in the air sample. In essence, with upstream components, rarely will the sample entering the AMC monitor be representative of the actual air sample in the environment due to effects caused by additional hardware upstream of the AMC detection system.

Calibration: AMC exists in the gas-phase. AMC monitors are supposed to detect contamination in the gas-phase, and so calibration of the AMC monitor should be carried out with traceable gas-phase contamination standards. Techniques such as Ion Chromatography (IC) or Mass Spectroscopy (MS) use liquid standards for calibration, and do not concern themselves with the collection efficiencies of the techniques that are used to convert the air samples into a liquid sample capable of being injected into these monitors.

Sampling Time: The less time it takes for the AMC monitor to detect the air sample, the better. Excess time only provides additional opportunities for contamination in the air sample to interact with itself.

Reactions may occur that create new molecules that are not representative of the actual air sample, contaminants within the sample may break down and so the monitor may not detect components that are actually there, or again the container the air sample is entrapped in may contribute to contamination in the stream that may then be falsely associated with the environment where the air sample was taken.

These concerns are especially valid with impingers or sorbent tubes, which collect air over a long period of time by trapping it in a medium. De-ionized (DI) water is a common impinger medium, and charcoal is common for sorbent tubes. Each of these techniques allow ample opportunity for the sample to be altered, due to the length of time taken to collect the sample, transport it to a laboratory for analysis, where it may wait for analysis.

Real-time results

Ion Mobility Spectrometry (IMS) is one of the most widely used techniques in the semiconductor industry to monitor AMC contamination. The preference for using IMS to monitor AMC contamination generally stems from the fact that it is a continuous real-time technique, it is highly sensitive (<70 pptv detection limits), and has a simple design.

Figure 2 illustrates the pneumatic and detection diagram of the AirSentry II IMS analyser from Particle Measuring Systems. It is apparent that the AirSentry II product has been simply designed, as there are no hardware components upstream of the IMS detection cell, and only a minimal amount of sample tubing is internal to the analyser (about 6 inches of 1/8" PFA tubing).

Because the AirSentry II is a point-of-use monitor, it is placed at the area of concern and does not rely on additional sample tubing to transport the sample from one location to another. There are no unnecessary hardware components that could act as failure points, and this design virtually eliminates preventative maintenance.

The simple design gives the AirSentry II advantages over other monitoring techniques. Because of minimal upstream sample tubing and hardware, the AirSentry II is able to achieve detection limits below 70 pptv. This limit of detection is defined as three times the noise at zero, therefore, zero noise is typically less than 20 pptv. In addition, there exists in the AirSentry II only one consumable item which needs to be changed at five year intervals, and so there are no mechanical hardware failure points and minimal service requirements.

Failure reduction

In summary, attributes of a given AMC monitor must be considered when implementing a monitoring strategy for molecular contamination control. Simpler designs have fewer hardware failure points, require less preventative maintenance, and provide less opportunity for the sample air to be altered prior to the monitor's detection system.

When considering an AMC monitoring strategy, it is the small things that usually count the most. Many AMC monitoring companies tout the limit of detection as the main technical point to focus on. However, while detection limits are a factor, it is also imperative to consider all aspects of a given monitor, including design attributes. A simple design often increases the effectiveness of the monitoring solution, reduces downtime, and provides a more accurate representation of the true AMC levels within critical lithography equipment, process bays, or cleanroom facilities.