SENSOR TECHNOLOGY AND DESIGN

Verifying Sensor Response to
Difficult Chemicals
with a New Test Chamber Concept

Chemicals with low volatility tend to present special problems when it comes to calibrating and verifying sensors that will be used to detect them. The challenge goes up a notch or two if the sensors must be tested under real-world conditions. But it can be done.

A. David Maughan, Jerome C. Birnbaum, and Kathleen M. Probasco
Pacific Northwest National Laboratory

Some chemical compounds are so difficult to measure that it is nearly impossible to evaluate and calibrate the sensors designed to detect them. These tricky substances include less- or semivolatile compounds found in pesticides and herbicides, the higher boiling polyaromatic hydrocarbons in diesel exhaust, and linked polyurethane foams used in products ranging from hiking boots to acoustic ceilings. They typically appear in low concentrations and evaporate very

These dual 9500 L test chambers, shown here in a fisheye view, are constructed of 2 mil Teflon film. Chambers can be built with smaller volumes as well to satisfy the requirements of the particular test.

slowly. These and other heavy chemicals are rarely measured accurately because they stick to surfaces and sampling equipment, making them hard to reliably sample or deliver to analytical detectors.

Researchers at Pacific Northwest National Laboratory (PNNL) have developed an atmospheric calibration chamber (see image at right), a controlled environment in which to certify sensor accuracy under the conditions in which the detectors are expected to operate. The facility is designed to handle and measure chemicals, including semivolatiles, at ambient levels. It can simulate releases ranging from industrial vents with high concentrations to those from surfaces, soils, and/or vegetation where the concentrations are in parts per trillion (ppt). Sensors placed in the chamber are evaluated against established standards in the presence of changing humidity, temperature, sunlight, photochemical oxidants, particulate loads, and interfering chemicals. Biological organisms, which are generally collected as particulates and cultured before identification, can also be included.

Chamber Components
The testing facility (see Figure 1) incorporates simulation or exposure chambers ranging in volume from a few liters to 9500, a pure air generator, a preconcentration system, chemical monitors, and highly sensitive analytical equipment.

Figure 1. A top-down schematic of the chambers and auxiliary equipment shows the various components of the test system.

All sampling lines and tubes in contact with test chemicals are heated, and have highly polished inside surfaces coated with a specialty glass. The larger chambers with mixing fans are constructed of virgin Teflon film and operate at laboratory pressure and temperature; smaller chambers are made of Pyrex glass that can be heated to >100°C.

Figure 2. The analyzers are adjacent to the chambers. Sample lines between the chambers and analyzers are coated with specialty glass and are heated.

Reproducible tests must also meet the requirements of the sensor. For example, it needs an adequate and known supply of the test chemicals or mixtures. The facility’s heated inert sampling surfaces allow researchers (see Figure 2) to dial up and hold environments at desired concentrations to evaluate sensors for accuracy, precision, reliability, and interference issues. The large enclosures minimize the interaction of test chemicals with chamber walls, circulating fans, and sampling ports.

To mirror real-world conditions, the chambers provide for natural chemical interactions. Light energy, moisture, and background contaminants can be added to simulate ambient conditions outside. Ultraviolet lamps lining the side walls provide radiant energy equivalent to the sun at noon, simulating the natural photolysis or degradations that occur in nearly all chemicals. The walls, ceilings, and floors are mirrored to bounce this light energy as evenly as possible throughout the chambers.

The dual-chamber setup permits simultaneous controls or duplicate analyses. The sensor under test can be placed in one chamber while the other provides a control or reference. The chamber volumes are great enough to sustain long sampling periods—hours to days, if needed. As samples are extracted, the chambers collapse on themselves to maintain a constant pressure that ensures unchanging test mixture concentrations over the sampling periods. A pressurized air purification system strips moisture and all contaminants exceeding ~¹/₂ ppb out of the make-up air supplied to the chamber, leaving only nitrogen, oxygen, argon, and carbon dioxide. Thermal oxidizers eliminate all hydrocarbons including methane, and the capability exists to remove carbon dioxide. To the pure air remaining can be added, as necessary, background concentrations of urban pollutants, photochemical oxidants, interfering chemicals, particulates, and others.

Testing to 1 ppt
A typical gas chromatograph-mass spectrometer (GC–MS) can identify chemical concentrations as low as ~1 part per million. The PNNL system can take that value down to 1 ppt or lower, particularly for the sticky semivolatile species. A large-volume sample preconcentrator captures tens of liters of test mixtures from the chambers on a precolumn, whose contents are then thermally desorbed for subsequent GC–MS analyses. The concentrations can thus be multiplied by factors of one thousand to one million or more, depending on the volumes preconcentrated.

Other instruments used to characterize test mixtures include a Fourier transform infrared spectrometer with a 20-meter path length, and high-sensitivity commercial detectors for ozone, oxides of nitrogen, and fine particles. Chamber temperature, pressure, and relative humidity are monitored constantly.

Spent mixtures are exhausted from the chambers through scrubbers, and the chambers are flushed with purified air before the next experiments. Pure nitrogen and medical-grade air from a large-volume oil-less compressor are available for purging. If a chamber cannot be brought to baseline concentrations, a new one is built.

Developing a Test Protocol
The first step before running a sensor evaluation experiment is to make an educated guess as to what the device might be required to detect. The physical-chemical properties and documented releases for each of the test chemicals can be reviewed (e.g., from American Chemical Society sources and Beilstein) to determine background levels and predict chemical behavior, fate, and degradation in field environments. The information is applied in designing chamber tests, including sensor exposure times, expected starting concentrations, and other factors a sensor could encounter, such as sunlight, urban- and transportation-related pollution, industrial emissions, photochemical oxidants, and changes in humidity.

PNNL staff prefer to work with the sensor developer or vendor when developing a test protocol. It is crucial that the protocol cover the full range (i.e., upper and lower concentration limits) of expected conditions. For example, a test protocol might specify that some sensors may be placed in a plume near a source where concentrations are high and variable, while others will be situated indoors or at locations where concentrations are lower and less variable.

The expected ranges in chemical concentrations and formed byproducts can be obtained using the following fate models:

• Dispersion modeling. A practical emission release rate for each test chemical is first estimated to predict drops in chemical concentration over time. Chemical fate models are used to look at the decline in chemical concentrations as a function of time and distance from a source, and to establish a realistic sensor exposure range for chamber experiments. If the release information is not available, it can be predicted from dispersion modeling or basic raw material (mass-balance) calculations. The degradation rate in the plume will generally be estimated using hydroxyl-radical interactions with the test chemical.
• Fugacity modeling. This modeling approach predicts the environmental media in which the test chemical will accumulate (e.g., air, soil, water, vegetation) after it is released into the environment, and the likely off-gassing concentrations of the chemical from those media. Fugacity modeling can also be used for any byproducts to which a sensor could respond.

For example, the semivolatile insecticide Malathion is designed to partition to or accumulate in plant leaves. Once in the leaves, it will slowly degrade and off-gas (see Figure 3).

Figure 3. A fugacity model run for Malathion produced these results.

Accordingly, a sensor designed to detect Malathion should be capable of measuring levels at ~40 ppb in the air adjacent to treated plants. For this application, a chamber experiment would be set up to expose a sensor to the off-gassing concentrations consistent with the fugacity model. When data are lacking on a particular chemical, PNNL relies on structural-functional group estimations.

Conducting a Test
To establish a detection limit and calibration curve, gaseous samples containing calibrated quantities of the test chemical are

Figure 4. Chambers can be built large or small to satisfy the particular test requirements. Here, an envelope-like chamber is suspended above several instruments being tested simultaneously. The setup includes test gas, flowmeters, manifold, and mixing chamber.

injected into the GC–MS. Detector response is then plotted against chemical concentration to produce the curve. The sensor is placed within a chamber of appropriate size (see Figure 4) as determined by the length of the experiment, concentration of the chemical of interest, and other substances to which the device will be exposed. A test typically proceeds as follows:

1. A nonoperating sensor is placed in the darkened chamber and a pretest is run to determine the chemical’s reactivity to the inert environment. The chemical is injected under nitrogen, and the circulatory fans are switched on for a short time to mix the test chemical with the pure make-up air. The concentration is monitored by the GC–MS to ensure the chemical’s concentration does not drop more than 5% in two hours.
2. The sensor is screened to determine whether it can detect and measure the chemical. A single midpoint concentration is added while the sensor is in operation to test its response time and ascertain how it performs with respect to the manufacturer’s range of detection capability.
3. Detailed tests are run to ascertain the sensor’s range of detectable concentrations. The sensor is then pushed to its operating limits relative to the sampling environment (e.g., changes in sunlight, temperature, and relative humidity).
4. Interference tests are conducted to expose the sensor to common atmospheric chemicals (ozone, hydroxyl radical, oxides of nitrogen, etc.) at ambient concentrations and to other chemicals that could be present in its operating environment.
5. The data are summarized and reported.

Summary
The PNNL atmospheric chemistry chamber facility is the only known system that can work with sticky, slowly evaporating, semivolatile chemicals without significant loss to sample line surfaces during testing. Its multiple chambers permit the creation of test mixtures of persistent chemicals. The sample preconcentrator and connected analyzers make it possible to evaluate sensor precision, accuracy, and performance in interfering environments. The facility is available to government, industry, aca- demia, and others for the measurement of persistent or volatile chemicals under nearly any real-world conditions.

We Love Feedback

Sensors® and Sensors Expo® are registered trademarks of Questex Media Group