"Executive Summary: The principal measurement methods for sampling particulate matter (PM) from stationary combustion sources generally consist of a sampling train that includes a heated filter maintained at a variety of elevated temperatures@ followed by a series of aqueous impinge solutions used to collect material that is in the vapor phase within the stack@ but could condense and form PM immediately after entering the atmosphere. These two fractions are commonly referred to as ""filterable"" (e.g. fly ash or catalyst fines) and ""condensable"" (e.g. sulfuric acid@ semi-volatile organics@ or ammonium sulfates)@ PM. Since the early 1970s@ the EPA has proposed a number of methods to determine PM emitted from combustion sources and these are discussed in this report. Most of the attention was initially centered on measuring filterable PM as this material was considered to comprise the major fraction of PM emissions subject to removal by control technology. The EPA methods and their variants differ primarily in the temperature of the collecting filter@ leading to accumulation of different amounts of material on the filter (depending upon the species present in the stack gas). A summary of the operating temperatures and impact on condensables accumulation indicates the biases that can occur between the principal PM measurement methods employed today. After the introduction of the National Ambient Air Quality Standards (NAAQS) for fine particulate matter@ regulatory interest shifted to methods that also measured the condensable fraction (consisting of PM having aerodynamic diameters equal to and less than 2.5 microns). Accurate measurements of these emissions became important as regulators sought to ensure attainment of PM ambient air quality standards by controlling source emissions. This paper discusses the technical basis of the resulting biases for both filterable and condensable PM for the various test methods The American Petroleum Institute recognized the importance of this issue in 1997@ initiating a review of the appropriate test method before undertaking a comparative testing program for PM 2.5 emissions from refinery sources. Measurements were made using a conventional cyclone/filter/impinger method (EPA Methods 201A and 202) along with a newly developed dilution system sampler. This new sampler seeks to represent atmospheric conditions by diluting and mixing stack gas with clean ambient air@ and collecting the resulting PM on filters similar to those used in ambient air sampling. Initial results from this program showed the cyclone/filter/impinger method had a significant positive bias that overstated the emissions of condensable PM because some of the stack gas SO2 was converted to sulfate PM in the impinger solutions. This program was expanded in 2000 with the participation of several other organizations: US Department of Energy@ Gas Research Institute@ California Energy Commission and the New York State Energy Research and Development Authority. By the conclusion of the test program in 2004@ the emissions of over a dozen combustion units had been measured. These studies confirmed the initial findings and demonstrated that the dilution system method provided more accurate and reproducible emissions data than those from filter/impinger methods. In addition to providing more accurate PM emissions data@ the dilution system readily provides a means of chemically characterizing these emissions in terms of their metals and organic content. Emissions from gas-fired sources were found to be significantly lower than those obtained using conventional test methods. The particulate matter collected was found to consist mainly of semi-volatile organics with no significant contribution from any poly-nuclear aromatics or metals. Emissions from catalytic cracking units were dominated by catalyst fines with their characteristic metals content@ sulfuric acid@ and to a lesser degree semi-volatile organics. While dilution system sampling has been endorsed by a committee of the National Research Council and the EPA@ the sampling devices are not widely available at present. In the interim@ a series of alternative options are provided as guidance to refinery source operators. These recommendations are: ? Use EPA Method 202 cautiously; it has a serious problem with positive bias caused by SO2 to sulfate conversion. While the EPA recognizes this@ it is not clear that all state and local regulatory agencies do. Therefore@ it is important to ensure that these agencies are also aware of the deficiencies of this method when used for compliance testing. ? The EPA has accepted a revised version of Method 202 to minimize the formation of artifact sulfate that appears promising. However@ it has not been extensively field tested and the presence of NH3 in combustion emissions may lead to a positive measurement bias ? PM emissions from gas-fired process heaters@ boiler@ and IC engines can adequately be determined by a method that only collects the filterable PM. This is supported by extensive testing@ which shows that the condensable fraction for these sources is negligible. ? PM emissions from catalytic cracking units are probably best treated on an individual basis as their operating characteristics differ widely@ leading to a range of combustion gas compositions. API studies have shown that emissions of semi-volatile organics are only a minor component compared to catalyst fines and condensable sulfur species (e.g.@ sulfuric acid and sulfate salts). Consequently a prudent approach to measuring these PM emissions is to focus on the major species using a combination of EPA Method 17 and the controlled condensation sampling system."