Final Report: Ultrasensitive Acrolein Sensor for Environmental MonitoringEPA Contract Number: EPD06023
Title: Ultrasensitive Acrolein Sensor for Environmental Monitoring
Investigators: Richman, Bruce
Small Business: Picarro, Inc.
EPA Contact: Manager, SBIR Program
Project Period: March 1, 2006 through August 31, 2006
Project Amount: $69,284
RFA: Small Business Innovation Research (SBIR) - Phase I (2006) RFA Text | Recipients Lists
Research Category: Air Quality and Air Toxics , Particulate Matter , SBIR - Air Pollution , Small Business Innovation Research (SBIR)
Accurate, continuous measurement of acrolein in both indoor and outdoor air is essential for establishing and enforcing air quality standards that protect human health and the environment. Acrolein is extremely toxic to humans and is listed by the U.S. Environmental Protection Agency (EPA) as 1 of 33 hazardous air pollutants. Acrolein inhalation at levels as low as 90 ppb (parts per billion) for 5 minutes has resulted in eye and respiratory irritation with increasing concentrations leading to more severe symptoms (Toxicological Review of Acrolein, Integrated Risk Information System, May 2003). Symptoms also are more severe in asthmatics, infants, and children (Prioritization of Toxic Air Contaminants - Acrolein, Children’s Environmental Health Protection Act, October 2001). Acrolein has been reported as lethal in concentrations greater than 150 ppm for 10 minutes. A current concern is long-term exposure of children to 1 ppbv level concentration of acrolein. Although the Occupational Safety and Health Administration permissible exposure limit (PEL) for adults is 100 ppbv for 8 hours per day, toxicology experts now suspect that 1 ppbv may pose significant harm to children.
Acrolein is used principally as a chemical intermediate in the production of acrylic acid and its esters. Acrolein is used directly as an aquatic herbicide and algaecide in irrigation canals, as a microbiocide in oil wells, as a liquid hydrocarbon fuel, and as a slimicide in the manufacture of paper. It also is used to cool water towers and water treatment ponds. Combustion of fossil fuels, tobacco smoke, and pyrolyzed animal and vegetable fats contribute to the environmental prevalence of acrolein (Prioritization of Toxic Air Contaminants - Acrolein, Children’s Environmental Health Protection Act, October 2001). Acrolein is a by-product of fires and is one of several acute toxicants that firefighters must endure. It also is formed from atmospheric reactions of 1,3-butadiene.
Acrolein is highly reactive with a short half-life that complicates sample collection and measurement accuracy. The steps for making an acrolein measurement in ambient air, for example, involve drawing air at a specific flow rate through a cartridge coated with acidified dinitrophenylhydrazine (DNPH), coldpacking the cartridge for shipment, then desorbing the cartridge and a comparison blank in the laboratory using high performance liquid chromatography (HPLC). A report published by the International Programme on Chemical Safety states, “technical difficulties in the measurement of acrolein in air include possible interference of propionaldehyde-DNPH and acetone-DNPH derivatives with the acrolein-DNPH derivative during gas chromatography or HPLC high-performance liquid chromatography and potentially low recovery of acrolein from DNPH-coated silica gel” (Concise International Chemical Assessment Document 43, World Health Organization, Geneva 2002).
A simple, accurate, real-time measurement of acrolein is required to better understand the health impact of acute and chronic human exposure to acrolein. Such an instrument could be placed in the field to continuously monitor and report acrolein levels in targeted locations. Picarro, Inc., has developed a sensitive, real-time acrolein measurement instrument that can detect acrolein. The all-optical approach for our sensor uses cavity ringdown spectroscopy (CRDS), which has been clearly established over the past decade as a robust, real-time, highly sensitive and selective technology for measuring trace gas concentrations.
In Phase I, Picarro built a CRDS system to target the acrolein absorption band at 1623 nm and determined whether it can detect acrolein with a sensitivity of 1 part per billion in the presence of H2O, CO2, CO, and other potential sources of interference in the atmosphere and in combustion exhaust. We chose the 1623 nm absorption band for acrolein because most other gases in the atmosphere and combustion exhaust have windows with little or no absorption features with the acrolein band.
Cavity Ringdown Spectroscopy
CRDS is a technique where a gas sample is introduced into a high-finesse optical cavity. The CRDS analyzer continuously generates high-resolution optical spectra of a gas stream, where each spectrum is comprised of optical absorption loss as a function of optical frequency. As data are acquired, continuous analysis of optical absorption spectra is performed resulting in reported acrolein concentration levels.
Figure 1 shows a bock diagram of the Picarro CRDS system. Light from a laser is injected into the cavity, and light exiting the cavity is monitored by a photo-detector. By bringing the laser and cavity into resonance, the intracavity light intensity builds up. The incident light is then rapidly turned off, the intracavity intensity decays exponentially (“rings down”) with a time constant, τ, that depends on the losses due to the cavity mirrors and the absorption of the sample at the laser frequency. The light falling on the photo-detector is directly proportional to the circulating cavity intensity, and so the loss may be determined via an analysis of the ringdown waveform. Figure 2 shows a ring-down curve, which plots the intensity of light inside the cavity as a function of time (in microseconds) after the laser shuts off. After shutting off the laser, most of the light remains trapped within the cavity for a long period of time (40 microseconds or more), producing an effective path length of tens of kilometers through the sample. Much like a multipass cell, this long effective path length gives CRDS its high sensitivity. Ringdowns are collected over a range of laser optical frequencies. A high-resolution absorption spectrum then is obtained from which the constituent quantities of the gas sample may be determined.
Figure 1. Block Diagram of the ESP-1000 Cavity Ringdown Analyzer
Description of the CRDS Instrument
As our first objectives, we procured and assembled the components of the cavity ringdown sensor. The sensor basic design is similar to several cavity ringdown sensors Picarro already has assembled, tested, and delivered to government and commercial customers for other applications, including the detection of trace ammonia in air, trace hydrogen sulfide in vehicle exhaust, and atmospheric carbon dioxide (the CO2 application being of extremely high precision and accuracy).
The most important differences among the instruments are the choice of laser wavelength, the type of materials used in the gas handling components and the gas flow rate, and the spectral analysis software and calibration methods. For this project, we used a laser tunable around a wavelength of 1623 nm, appropriate for observing the absorption spectrum of acrolein in the near-infrared.
Figure 2. Ringdown Curve
We determined that the lower detectable limit (LDL) of our trace gas sensor in artificial air (mixed from pure components) is 2 ppbv in zero air (N2 and O2 only) in a 2-hour measurement time. The LDL is minimum at this measurement time (±1 hour). The LDL is approximately 10 ppbv in approximately 6 minutes in artificial air mixed from pure components (not improved by a longer measurement time). We define the LDL as three times the standard deviation of the acrolein measurement result at zero and low acrolein concentration. The artificial air consisted of a mixture of CO2 varied between 300 and 1,000 ppmv, CH4 varied between 1 and 4 ppmv, relative humidity varied between 20 to 80 percent (at room temperature), and balance of zero air (80% N2, 20% O2). We also attempted to measure low concentrations of acrolein (0 to 10 ppbv) mixed with air sampled from our laboratory room. The results of the room air test were inconclusive, but we estimate that the LDL is approximately 10 ppbv in 6 minutes without attempting to correct for the concentrations of other volatile organic compounds (VOCs) in the room air (increasing the measurement time does not improve this result).
Absorption Spectra of Acrolein and of CO2, H2O, and CH4
We measured the absorption spectra of acrolein, and of the possible interfering atmospheric gas species CO2, CO, H2O, CH4, and C2H4. Our purpose was to select spectral regions to use to determine the acrolein concentration, relatively free of absorption by the interfering species. In addition, we wanted to select at least one region with absorption by those species to determine their concentrations, so that we could apply corrections to the computed acrolein concentration. Figure 3 shows a measured spectrum of 1 ppmv acrolein in nitrogen at 500 torr pressure using our sensor. These scans were taken using high spectral resolution and cover almost the entire tuning range of the laser. The acrolein spectrum has no fine structure, such as resolved rotational lines typical of smaller molecules. Hence, its spectral shape is independent of pressure, and its amplitude scales linearly with pressure. Because the broad spectra have too much information to process for a real-time acrolein monitor, we need to select small regions of the broad spectra that contain enough information to determine the acrolein concentration in the presence of the other species. As such, we need to acquire data in spectral regions with both high and low acrolein absorption (so that the difference may be used to determine the acrolein concentration), and spectral data to quantify the concentrations of the interfering species (i.e., absorption peaks), to make corrections to the computed acrolein concentration.
Gas Concentration Calibration
Figure 4 shows the measurement of acrolein concentration steps in time, using zero air as balance gas. We also measured each of CO2, CH4, and water vapor individually in concentration steps using zero air as a balance gas.
We correlated the fitter output of each of these three gases with each of the other two gases, and with the baseline differences. These data were used to determine correction factors to the computed acrolein concentration from the baseline differences.
Figure 3. Measured Spectrum of 1 ppm of acrolein in N2 at 500 Torr Pressure After Subtracting Baseline
We measured zero air continuously for more than 24 hours to determine how averaging the baseline differences over time improved the standard deviation of the average. We define the LDL as three times this standard deviation. Figure 5 shows the trend of the standard deviation of the averages of the computed acrolein concentrations for zero air. The trend ideally follows a reciprocal square root, shown for comparison. In all real systems, the standard deviation of the average reaches a minimum at some number of averaged results, and increases for averaging more than that number. According to the data in Figure 5, this minimum occurs at greater than 256 scans. (We consider the 512-average point as uncertain because of too few 512-averages to compute the standard deviation.) Acquiring 256 successive spectral scans takes approximately 100 minutes.
Figure 4. Computed Acrolein Concentration During Concentration Stepping
Figure 5. Standard Deviations of the Averages of Successive Scans
In our experiments to determine the LDL of our acrolein sensor we found that the LDL in zero air with or without acrolein is approximately 2 ppbv in 100 minutes, in artificially prepared air (from pure components) is approximately 10 ppbv in 6 minutes (not improved by a longer measurement time), and in laboratory room air is approximately 10 ppbv in 6 minutes. In our measurement of artificial air, one of the three interfering gas concentrations was changed every 10 minutes (each gas changing every 30 minutes) in a psuedo-random concentration pattern. The CO2 concentration was varied from 300 to 1,000 ppmv, the CH4 concentration from 1 to 4 ppmv, and the H2O concentration from 20 to 80 percent relative humidity. The computed acrolein concentration values varied over a range of approximately 6 ppbv. From this, we estimate that our LDL in artificially prepared air is approximately 10 ppbv. Also, averaging only 16 scans (6 minutes) is sufficient to obtain an LDL of 10 ppbv according to these data. We also measured 2 ppbv and 4 ppbv of acrolein mixed with the artificial air. The statistical results were similar to those with no acrolein. We measured laboratory room air with zero and 1 to 10 ppbv of acrolein mixed in. We saw similar variations in the computed acrolein concentration as with the artificial air. An average of 16 scans was again sufficient to provide an LDL of approximately 10 ppbv.