How Desiccants Control Moisture Interferences with Optical Particle Counters

K. Andy Young, Hach Company
Tags: particle counting, oil analysis

Particle counters are highly accurate, highly sensitive instruments, that by virtue of modern engineering and user-friendly interfaces, have become commonplace and accessible to a variety of industries and operators.

Because of this, anomalous particle count data can be difficult to diagnose for those not well-versed in the governing theories and principles of particle counting. This article presents a potential source for sample contamination: Moisture ingression into hydrophobic liquids via an expired desiccant chamber in Pacific Scientific’s HIAC Automatic Bottle Sampler (ABS-2).

The benefits of the desiccant chamber are often downplayed or ignored. However, experience in the field and in the laboratory shows conclusively that the effects of an expired desiccant can be quite dramatic. Two case studies are detailed that include data collected by customers in real-world applications.

Complementing one of these case studies is data collected in the supplier’s laboratory. The data shows that moisture contamination can adversely affect the accuracy and reproducibility of optical particle counters and that an expired desiccant chamber in the bottle sampler was the source of this contamination in these cases.

Because particle counting has become so accessible to industry, it is typically forgotten that it is still a sensitive tool susceptible to many disturbances and contaminants that may alter the sample and lead to erroneous readings.

Many of these disturbances and contaminants may not be obvious to the operator, or assumed to be benign; many people forget or are unaware that an optical particle counter has sensitivity into the range of parts per billion (ppb).

Poor sampling procedures and contaminated samples are a perpetual problem and it is believed that 80 percent of all reported particle counter problems can be traced to bad samples. When abnormal counts are encountered or when counts are not reproducible, it is natural to inquire about the accuracy of the sensors, samplers, counters, lasers and other sensitive electronic components.

However, experience suggests that, most often, the equipment is operating correctly and the counts reported are accurate (though, perhaps not representative of the real particles in the fluid). The source of the deviations will typically be found in sample preparation, counting procedures and other environmental factors.

Role of Desiccant in Controlling Soft Particles

The bottle sampler operates by using atmospheric air to generate a pressure gradient to drive the fluid through the sensor. This air is double filtered: Once through a desiccant to remove moisture and once through a borosilicate glass filter to remove particulates.

If the filter was faulty, then particles circulating in the ambient air would be introduced to the sample fluid. The result of adding solid particulates (called hard particles) via the pumped air is obvious - if particles are added to a fluid, the fluid’s particle concentration will increase. The resultant particle count will be elevated and will not be representative of the fluid.

If the desiccant is expired, then it will no longer remove the water vapor from the air and moisture would potentially be introduced to the sample fluid. The result of adding moisture to the sample is subtler than adding hard particles.

If the sample is hydrophilic (water loving), like some ester-based synthetic lubricants for instance, the added moisture is likely to be solubulized in the fluid without affecting the particle count. The amount of moisture added would be so miniscule that the volume of the sample and volume dependent fluid properties would be unchanged.

In the case of a hydrophobic (water hating) oil sample, such as a mineral-based lubricant, the moisture may interfere with typical particle counting. The water vapor introduced into the fluid may condense and form an emulsion (like fog), creating micron-sized beads of water dispersed within the sample.

When an immiscible fluid is introduced into a sample, it is called a soft particle. This nomenclature has been chosen because these fluids are seen by the optical sensor and counted as particles, but they are not particles in the traditional sense.

Soft particles cannot be filtered out of the sample in the same manner as hard particles; the emulsion would pass through a mechanical filter with roughly the same efficiency as the supporting fluid.

Typically, the sample fluid will have an index of refraction different from water, which renders the water beads visible to the particle counter. That is, the water will scatter or block the laser light due to the fluids’ contrasting optical properties. This creates false counts due to the water contamination of the sample fluid.

In the case of a mineral-based oil sample, a faulty desiccant could lead to the introduction of moisture into the oil. Because water is both immiscible in oil and denser than oil, the moisture would emulsify and gravitate to the bottom of the sample chamber. As the sample is drawn through the sensor, the water beads would register as particles and contribute to the overall sample count.

The natural question is: How susceptible to significant moisture contamination are hydrophobic fluids and the condition of the desiccant to such errors? The following cases explore data gathered in industry and the lab in an attempt to better characterize the effects of water contamination on hydrophobic fluids.

Customer #1 - Diesel Fuel Samples

A manufacturer of diesel fuel accessories recently encountered problems maintaining consistent particle counts with its optical automatic particle counter. The Connecticut-based company had been using the equipment for some time with no problems. In late summer, the particle counts in the diesel fuel showed an increase on successive run trends for a given sample.

Figures 1 and 2 show the data collected by the customer plotted on log-log graphs. Typical particle count distributions yield a linear fit on log-log graphs, which is what is seen in the first run. However, as the run number and the time increase, the particle counts become elevated and anomalous.

The shape of the curve combined with experience in data analysis suggested that the anomoly was due to water or air. Because the samples were degassed and diesel fuel is hydrophobic, water was the most likely source of the inflated particle counts in the larger size channels.

A careful analysis of the customer’s sampling procedures and particle counting techniques ruled out most typical sources of sample contamination. Upon close inspection of the equipment, it was discovered that the bottle sampler had an expired desiccant chamber.

The plant is located in the northeastern United States, a region where extremely high relative humidities are common during the summer months. It was recommended that the desiccant chamber be replaced or recharged before proceeding with any additional data collection. After recharging the desiccant, the company had no further erratic readings from its particle counter.

Customer #2 - Motor Oil and Dilution Fluid Samples

A construction equipment company that routinely uses particle counters to assess the cleanliness level of motor oils was looking for a relatively clean fluid to use as a diluent. The application was designed to effectively mask or assist in dissolving additives that were contributing to the particle counts of the oil.

On the recommendation of its oil supplier, the company investigated a paraffinic lamp oil. However, they noticed that they could not achieve reproducible particle counts with the neat lamp oil. At times, the lamp oil appeared extremely clean, with an ISO code of 12/10.

However, on successive runs the counts fluctuated wildly. There appeared to be no pattern to the data, except that the data was constantly changing.

In an attempt to assist the customer, the supplier conducted similar tests with the lamp oil at its laboratory. It can clearly be seen in Figure 3 that the data quickly became erratic and unpredictable. It was not immediately clear to either the customer or supplier what could have caused these irregular data trends.

Further tests were conducted at the supplier’s laboratory using a nonpolar hydrophobic solvent (QED) and a hydrophobic aviation hydraulic fluid (MIL-H-5606B) used for calibration. Similar erratic results were observed. Below is the list of normal suspects when erratic readings are observed, especially on calibration fluid:

Because each lot of clean oil and calibration fluid is verified against industry standard concentration limits, these fluids were an unlikely problem.

Similarly, due to the experienced technicians and the high quality instrument involved, it was also unlikely that there was something wrong with the sensor. PSI was, therefore, still trying to identify the phenomenon that was creating such abnormal behavior.

Hydrophilic Isopropanol Reveals the Problem

Up to this point, the investigated fluids have been hydrophobic and nonpolar. Though isopropanol (IPA), a hydrophilic polar fluid, had been used intermittently throughout the testing as a cleaning agent, no particle counts had been performed on the fluid. Like QED, IPA has the advantage of being relatively clean (ISO codes of 9/8) straight out of the bottle. Therefore, it was expected that relatively low levels of contamination would be seen in the IPA.

The data collected from the IPA tests are shown in Figure 4. Unlike the previous fluids, IPA showed expected low particle counts and stable, reproducible counts. The cause of the abnormal particle counting behavior was not affecting the IPA. Subsequent testing on lamp oil, QED and 5606 showed identical behavior to those seen previously. The particle counter was simply behaving differently with hydrophobic than hydrophilic fluids.

Finally, upon close inspection of the instrument, it was determined that the bottle sampler’s desiccant chamber was expired. The desiccant was immediately changed and the tests were repeated on all the fluids under investigation. The data shown in Figure 5 are in sharp contrast to earlier data. The particle counts for the lamp oil, QED and 5606 were extremely low and stable after the expired desiccant was replaced.

The customer’s efforts to find a suitable diluent for their application and their faith in particle counting was severely damaged by the strange behavior and the lack of an immediate explanation.

However, confidence was restored due to a quick response leading to a clear explanation of the problem and a permanent and effective solution. This case highlights several important points about particle counting:

This knowledge actually gives the customer a deeper appreciation for particle counting, helps them to become more comfortable with the science and leads them to develop better and more comprehensive particle counting procedures.

Certainly, desiccant chamber maintenance is vital to the particle counting. To ensure accurate and reliable particle counting results with optical particle counters, the desiccant chamber must be replaced immediately upon expiration.