Final Report: Detecting Metals in Ambient Particulate Matter: X-Ray Fluorescence Analysis of High-Volume Impaction DepositsEPA Contract Number: EPD04019
Title: Detecting Metals in Ambient Particulate Matter: X-Ray Fluorescence Analysis of High-Volume Impaction Deposits
Investigators: Hope, Thomas J.
Small Business: Rupprecht & Patashnick Co, Inc.
EPA Contact: Manager, SBIR Program
Project Period: March 1, 2004 through August 31, 2004
RFA: Small Business Innovation Research (SBIR) - Phase I (2004) RFA Text | Recipients Lists
Research Category: Air Quality and Air Toxics , Particulate Matter , SBIR - Air Pollution , Small Business Innovation Research (SBIR)
Testing of the system as a whole first centered on the viability of providing significantly increased signal-to-noise ratios using the high volume impaction/concentration scheme. Samples of particulate matter containing smoke from an automotive safety flare were collected. The automotive safety flare smoke was used to provide elevated levels of elemental species not normally found in ambient PM samples, specifically strontium.
No specific control over specific strontium PM concentration was implemented for this test other than "least", "more" and "most" concentrated samples were collected and analyzed. Blank samples were also analyzed for this test for background subtraction purposes. The plastic sample holder used to contain and concentrate samples for analysis contained titanium. The spot size for the XRF equipment used for this experiment was oversized and hence illuminated not only the sample but portions of the sample holder as well. Later improvements to both the XRF equipment and sample holder eliminated the problem of sample holder titanium peaks showing up in the analysis spectra. Titanium peaks were present throughout the data indicating that further X-ray spot size optimization was necessary.
Spectra were collected for both concentrated and un-concentrated impaction samples. The air sampled contained approximately 1200 standard liters of "most concentrated" road flare smoke. The post collection concentration ratio of the sampled material in this experiment was approximately 10:1. The increase in signal strength for both the K α and K β XRF peaks of strontium on concentrated and un-concentrated media was approximately 7.4:1. The concentration improved signal-to-noise ratios almost a full order of magnitude.
Due to the nature of XRF analysis, analytical sensitivity for individual elemental species is variable. The variability arises from the physical interaction of the analysis system and it's environment, physical interactions between materials as well as energy limitations within the system. It will therefore be necessary to determine the system's analytical limits for each elemental species of concern in order to use the system to determine quantitative information on the elemental make up of sampled PM.
The system used to analyze all samples included in this report completed all analysis in air. The interaction of the analysis beam and the surrounding air will attenuate the signals from all of the lighter elements. Therefore, the system will not be able to determine the presence of elements below atomic number 12, magnesium. If the system completed the analysis function under vacuum, lighter elements would become visible.
The system employed to complete all measurements included in this report is capable of providing a limited amount of excitation energy. Due to the limitation, the innermost electrons for the heavier atomic species are unable to absorb enough energy to be ejected from orbit. Therefore as atomic number increases beyond a yet to be determined threshold, K band derived emissions from specific elements become unavailable. For heavier elements, it will be necessary to use L band emissions to identify their specific presence in an analytical spectrum.
Finally, the XRF system uses a silver (atomic number 47) anode tube. Interference can occur with elemental compounds close in atomic number to silver as the large peeks associated with the anode material and associated scattering "washes out" peaks with similar energy levels. Cadmium (atomic number 48) is a good example of an element where the K band peaks become disguised by anode material scattering. L band energy peaks identified the cadmium analyzed during this test.
Liquid standards each containing 1000 ug/ml of chromium, manganese, nickel, zinc, cadmium, mercury, lead and barium in a 5% nitric acid or hydrochloric acid solution were used to prepare samples spiked with known amounts of each element. Each solution was placed as one drop onto individual sample substrate pieces. Assuming one drop ' 0.05 ml approximately 50 ug of analyte material was placed on each sample substrate.
Identification of all the transition metals tested was easily accomplished with large signal-to-noise ratios, with the noted exception of cadmium as discussed above. Characteristic K band peaks identified chromium, manganese, nickel and zinc where as characteristic L band peaks were used to identify cadmium, mercury, barium and lead samples. Additional testing for all the elements of interest will be completed during the project's β phase.
Analytical sensitivity can be improved for the system as a whole through increasing the signal-to-noise ratio. Two different signal-to-noise ratio improvements were tested. The signal-to-noise improvements included using variable excitation voltage on the X-ray tube and a filter on the x-ray source filtering. Using different excitation voltage on the X-ray tube in effect uses all available energy to excite a lesser portion of the broad-spectrum range at the expense of other portions. Filtering accomplishes a similar effect by eliminating a portion of the energy emitted from X-ray tube before it reaches the target.
Testing was completed to show that running the X-ray tube at voltages of 25, 38 and 40 kV produced spectral results that were improved for some portion of the overall analysis and worse in others. Specifically, the spectral results for barium were improved with the 25 kV excitation. Packing the energy available into the lower end of the excitation spectra boosts the signal for the barium L α and L β peaks.
The use of improved spectra at specific energy levels may be of use in the automated analysis system. In cases where desired analytical accuracy was not achieved the functionality of the system could be designed to recognize the poor accuracy and re-analyze at optimized energy levels tied to specific elements.
Testing was completed to show the increase in signal-to-noise gained by the introduction of an X-ray beam filter. The filter used was a thin aluminum foil filter. The filter was placed over the end of the collimator. The filter provides increased signal-to-noise through eliminating any primary (X-ray tube) energy at or below the level to which the filter attenuates. Any energy peaks present at or below the filtered levels come from excited sample material only and not from scattered primary X-rays.
The use of improved spectra with specific excitation energies removed may be of use in the automated analysis system. In cases where desired analytical accuracy was not achieved the functionality of the system could be designed to recognize the poor accuracy and re-analyze at optimized energy level through the most beneficial filter.
A series of ambient particulate matter samples were taken for variable durations. Average Ambient PM level for the duration of the sampling was extremely low. The New York State Department of Environmental Conservation (NYS DEC) Ambient Air Sampling Station, South Pearl Street, Albany, NY measured levels from ambient PM2.5 of 5 to 8 ug/m3 for the duration of the sample set. The sampling location (Castleton-on-Hudson, NY) was approximately 8 miles southeast of the NYS DEC Ambient PM Air Monitoring Station on Pearl Street in Albany, NY. The Albany NYS DEC station information was used to determine that the ambient particulate matter remained at very low levels for the overall area for duration of the testing. Samples of 0.5, 1, 2, 4 and 8 hour durations were collected at 186 l/min.
Samples were analyzed and normalized to 300 seconds of exposure time. The X-ray tube parameters for the entire sample set were 38 kV, 30 uA. Results showed increasing signal-to-noise improvement with increasing time and hence, deposition. Iron, calcium and sulfur were detected in all but the 0.5-hour duration sample.
The Alpha prototype was designed, built and successfully tested. The prototype uses high volume impaction/concentration technology to collect a sample and XRF techniques to analyze the collected sample. The sample concentration techniques employed offer significant signal-to-noise improvements. Concentration techniques provide a significant improvement in the analysis system signal-to-noise ratio. Testing shows that for a 10:1 compression ratio, gains in the signal-to-noise ratio of roughly 7.4:1 were accomplished.
The ambient PM collection device uses a proven technique efficient in rapidly collecting a particulate matter sample from ambient air with little loss. The collection train is based on existing R&P instrumentation.
Through combining high-volume collection and concentration, it becomes possible to place the particulate matter contained in 12 m3 of air into a 16 mm2 area, in an hour, for x-ray analysis. Using normal techniques, the PM contained in 1 m3 of air is placed onto a 1075 mm2 area. Assuming a 20-ug/m3 ambient PM level the prototype system is capable of depositing 15 ug/mm2 whereas the manual method system only produces 0.0115 ug/mm2 (assuming an industry standard 1m3/hour flow rate and a 47 mm diameter collection filter). It is the gross mismatch of PM concentration, provided by the two techniques that places enough material in a small enough area using a small X-ray spot size for XRF analysis to be feasible using the tested technique.
The X-ray measurement head was developed specifically for ambient PM speciation system. The measurement head consists of X-ray Tube (40 kV end window with a silver anode and collimator), X-ray tube controller, detector and digital pulse processor. The measurement head geometry was optimized specifically for use in the α prototype instrument.
The prototype is capable of collection and analysis of ambient particulate matter samples in near real time. All interactions between subsystems of the unit are completed manually. All spectra interpretation associated with the Alpha prototype was also completed manually.
Analytical sensitivity to the various elemental species that may be present in ambient PM is varied and will be studied in detail during Phase II of the development project. To show proof of concept, samples containing roughly 50 ug of barium, chromium, manganese, nickel, zinc, cadmium, mercury and lead were analyzed. Cadmium and Barium proved to be the most difficult elements to measure although certain useful analytical techniques, discussed above, make their identification easier.
Testing showed sensitivity enough to determine the presence of sulfur, calcium and iron in a 1, 2, 4 and 8-hour duration ambient PM samples even under extremely low ambient PM concentration conditions. Analytical techniques will be investigated during the β phase of development. Certain techniques may prove useful in providing better signal-to-noise ratios in the Beta unit. Filtering and/or variable excitation energy may provide added analytical sensitivity when required.