Summary: EPA proposed a rule in September to guide states as they implement the National Ambient Air Quality Standards for fine particulate matter (PM2.5). The proposal establishes a hierarchy of precursor pollutants: SO2 is always considered a precursor, NOx is presumptively a precursor, and volatile organic compounds and ammonia are presumed not to be precursors. It outlines the steps states must take to demonstrate that they are achieving "reasonable further progress" in attaining the standards. It details the states' obligations regarding reasonably available control measures, reasonably available control technology (RACT), and PM2.5 program-specific rules for New Source Review. EPA proposes that in states that achieve their required reductions under the Clean Air Interstate Rule (CAIR) entirely from electric generating units, CAIR will satisfy SO2 and NOx RACT requirements, provided that for NOx RACT, existing selective catalytic reduction control operate year-round in nonattainment areas beginning in 2009. The states have their work cut out for them: even with CAIR in place, of the 36 areas designated as nonattainment, as many as 22 areas are projected still to be in nonattainment in 2010.

On September 8, 2005, EPA proposed requirements for state implementation of the National Ambient Air Quality Standards (NAAQS) for fine particulate matter (PM2.5). By April 14, 2008-three years after areas were designated as nonattainment-each state with nonattainment areas must submit to EPA a State Implementation Plan (SIP) demonstrating how the areas will attain the PM2.5 standards as expeditiously as practicable by reducing emissions of PM2.5 and its precursors.

More than 90 million people-approximately one-third of the country-live in 39 areas that EPA designated as nonattainment on April 14, 2005. (Since the original designation, three areas have been reclassified as attainment, dropping the total to 36.) Many of these nonattainment areas have a long way to go to reach attainment: over 40 percent have design values that are more than ten percent above the annual standard of 15 micrograms per cubic meter (µg/m3).

While national programs such as the Acid Rain Program and the Clean Air Interstate Rule (CAIR) will bring many areas closer to attainment, states nevertheless retain primary responsibility for ensuring attainment. Furthermore, air quality modeling demonstrates that even with various national programs in place, including CAIR, of the 36 areas currently designated as nonattainment, between 19 and 22 areas (with over 50 million people) are projected to remain so in 2010, as will 14 in 2015; hence the need for local action, some of which may also have higher benefits per ton of emission reduction, particularly in dense urban areas.

EPA emphasizes the enormous benefits of even small reductions in PM2.5, suggesting that the monetized health benefits of reducing emissions are three to 30 times greater than the costs. Moreover, it is more efficient to target direct PM2.5 as opposed to precursor emissions, since only a fraction of precursor emissions are converted to PM2.5.

The PM2.5 Implementation Rule describes the requirements and deadlines that states must meet in coming into attainment with the PM2.5 NAAQS. In particular, the rule specifies how states should treat PM2.5 precursors, and lays out measures states must take to demonstrate that they are achieving reasonable further progress, that they are implementing reasonably available control technology and reasonably available control measures, and that they are subjecting affected sources to the stringent requirements of the New Source Review program.

For some other NAAQS, EPA has developed a classification system for nonattainment areas. For example, ozone nonattainment areas are classified in one of five categories from "marginal" to "extreme," depending on the severity of the problem. However, for PM2.5 EPA prefers not to establish a classification system. Despite its preference, EPA does indicate that it may decide to establish a two-tier system, with areas classified as either "moderate" or "serious" based on the severity of nonattainment (e.g., serious areas would have design values above a certain threshold), attainment date (e.g., serious areas would be those with attainment dates 2011 or later), or other criteria.

Attainment demonstrations and SIPs are due by April 2008, or three years after re-designation if an area was originally designated attainment and then re-designated nonattainment. The attainment date is five years after designation, or April 2010 for most areas.

However, states may receive two types of extensions. First, a state is eligible for two one-year extensions if it has complied with all requirements and commitments, is achieving progress towards attainment, and has only a minimal number of exceedances in the preceding year.

A state may also receive an extension for a period of one to five years if it demonstrates in its SIP that it is impracticable for an area to attain the standard within five years of designation, due, for example, to the severity of the nonattainment problem or the lack of available or feasible control measures. In these circumstances, the state must demonstrate that all local control measures that are reasonably available and technically feasible are being implemented. In addition, the state must offer an alternative attainment date that comes "as expeditiously as possible, but in no case later than ten years after the designation date."

While precursors are often important contributors to PM2.5 formation, it is difficult to determine the impact of source-specific precursor emissions on ambient PM2.5 air quality concentrations. Furthermore, EPA believes that it may be appropriate to treat each PM2.5 precursor differently, based on evidence as to the degree to which the its emissions contribute significantly to ambient PM2.5 concentrations in a particular location.

In all aspects of the implementation rule (e.g., general nonattainment planning, New Source Review), EPA proposes to require states to address SO2 and NOx as PM2.5 precursors in all areas, but also to allow states to exempt NOx (but not SO2) by demonstrating that NOx emissions from stationary sources are not a significant contributor to an area's PM2.5 problem and that the area is not a source of a PM2.5 transport problem. On the other hand, states are not required to address ammonia or volatile organic compounds (VOCs) as PM2.5 precursors. However, a state may regulate these pollutants if it can demonstrate that they are significant contributors to an area's PM2.5 problem.

Section 172 of the Clean Air Act provides that SIPs for nonattainment areas must require "reasonable further progress," defined as "annual incremental reductions in emissions" that will ensure attainment by the deadline. EPA notes that it is important that reductions be achieved not only in the last year or two prior to the attainment date but, rather, in a fashion that provides "generally linear progress toward attainment."

RFP demonstrations are due by April 2008. However, states must submit demonstrations for areas not expecting to meet the standard by 2010, showing how the areas will make reasonable progress over periodic three-year milestones (i.e., 2010 and 2013). For the reasonable progress determination, EPA proposes to offer states flexibility to address either PM2.5 or its precursors as the states choose, so long as they achieve the net air quality improvements that are required.

Because of the different transport characteristics of PM2.5, NOx and SO2, EPA is proposing a different geographic reach for RFP demonstrations for each pollutant. In particular, because direct PM2.5 tends to deposit relatively locally, EPA proposes that the RFP demonstration include direct PM2.5 emissions within the entire nonattainment area, but not farther afield. However, because emissions of NOx and SO2 from outside the nonattainment area may affect the area's air quality, EPA proposes to consider sources located up to 200 kilometers outside of the nonattainment area for purposes of the RFP demonstration. An RFP demonstration should also include a motor vehicle emissions budget for each milestone year, but the area would not get credit for any motor vehicle emissions reductions achieved outside of the area.

States must also develop contingency measures for nonattainment areas, which are additional control measures to be implemented in the event that an area fails to meet RFP requirements or fails to attain the standards by its attainment date. It must be possible to implement these measures quickly, without significant further action or additional rulemaking. EPA emphasizes that contingency measures should consist of control measures beyond those required to attain the standards.

SIPs must "provide for the implementation of all reasonably available control measures (RACM) as expeditiously as practicable," including such reductions as may be obtained through the adoption of reasonably available control technology (RACT). RACT signifies the lowest emissions limitation that a source is capable of meeting by applying control technology that is reasonably available, considering technological and economic feasibility. A control measure is considered technically feasible if, in light of its availability, cost, and effectiveness, it provides a net benefit to public health and the environment. EPA is not proposing a fixed dollar per ton cost threshold for RACT or RACM, on the theory that the cost threshold should vary based on the severity of the nonattainment problem.

To demonstrate RACM, a state should show that there are no additional measures reasonably available that would advance the attainment date by at least one year or contribute to RFP for the area. States must submit RACT rules to EPA within three years of designation, at the same time as the attainment demonstration (April 2008). Although states must implement any measures determined to be RACT "expeditiously" (generally at least one year before the attainment deadline), EPA will allow for more time when necessary for the implementation of innovative control measures.

EPA is proposing three approaches to implementing the PM2.5 RACT requirement. The first option would require states to mandate reasonably available control options for all affected stationary sources in the nonattainment area. The second would require RACT for stationary sources only to the extent necessary for expeditious attainment or to meet the RFP requirement. Yet a third option would require the second option for areas with attainment dates within five years, and the first option for areas with attainment dates beyond five years. EPA's view on this score is that certain areas will attain the standards by 2010 mostly as a result of federal programs, and that such areas should not be required to impose controls on local sources if doing so would not hasten attainment.

EPA suggests that states keep a number of items in mind when analyzing whether or not RACT will help an area achieve attainment or meet RFP requirements. The Agency emphasizes that many of the PM2.5 nonattainment areas are not former PM10 nonattainment areas; as a result, some stationary sources may currently be uncontrolled or under-controlled for PM2.5 and PM2.5 precursors. Even in areas that were designated nonattainment for PM10, EPA believes that certain stationary sources of PM2.5 precursors may be poorly controlled, giving rise to substantial control opportunities. In addition, many sources may have installed emission controls years ago, and cost-effective opportunities for deeper control may have become available in the interim. Improved monitoring methods may also have improved the ability to detect equipment failures.

Finally, existing controls may focus on filterable PM, but may not effectively control emissions of condensable PM. Condensable PM forms from material that is not particulate matter at stack conditions, but that condenses or reacts to form particulate matter upon exiting the stack and cooling in the ambient air. The formation of condensable particulate matter occurs within a few seconds after discharge from the stack, and is largely composed of PM2.5 (as opposed to larger particulates). Condensable PM is not measured in stack tests and has not been considered in the past for permitting purposes. EPA is currently in the process of developing guidance on how to treat condensable emissions for RACT.

EPA proposes to allow states to accept for RACT purposes RACT analyses conducted previously (for VOCs or NOx under the one-hour ozone standards, or for direct PM10 emissions under the PM10 standards), provided that the previous determination currently represents an appropriate RACT level of control for PM2.5. States may also rely on information gathered for New Source Review purposes from prior best available control technology (BACT) or lowest achievable emission rate (LAER) analyses, if the information is sufficiently current. In addition, if a state has determined VOCs to be a significant contributor to PM2.5 formation, compliance with MACT standards for VOCs will generally be sufficient for compliance with RACT. Finally, EPA is proposing to require a new RACT determination even if a previous RACT analysis for a source concluded that no controls were necessary, since new technologies may be available.

Because ambient PM2.5 concentrations are a year-round problem and NOx emissions have a more significant role in PM2.5 formation in cooler temperatures, EPA has determined that year-round operation of NOx emission controls (but not SO2 controls) that sources have already installed for NOx SIP call purposes is necessary for RACT. In most cases, the additional costs of running the controls year-round would be relatively cost-effective compared to controls for many other sources subject to PM2.5 RACT. EPA estimates the cost of year-round operation of these controls at $500 per ton in the CAIR region.

EPA proposes that in states that achieve their CAIR emissions reductions entirely from electric generating units (EGUs), CAIR will satisfy SO2 and NOx RACT requirements, provided that for NOx RACT, existing selective catalytic reduction (SCR) controls in nonattainment areas are required to operate year-round beginning in 2009. EPA does not believe that requiring source-specific RACT controls for EGUs will provide deeper SO2 and NOx reduction beyond what CAIR alone will achieve. EPA argues that source-specific control requirements may affect the temporal distribution of emissions (by reducing banking) or their spatial distribution, but that they will not affect total emissions.

In states not achieving all of their CAIR emissions reductions from EGUs, CAIR would not satisfy RACT. Finally, EPA notes that states have authority to conduct their own RACT analysis for any source, and that EPA's proposed approach to CAIR would not prevent any state from requiring beyond-RACT controls.

• EPA proposes to allow NOx and SO2 "RACT averaging": a state would identify presumptive RACT for a set of emissions sources, and then propose a program that would assure aggregate emissions reductions equivalent to or greater than the emissions reductions that would be achieved if the presumptive RACT level were met by each individual source.
• EPA has determined 100 tons per year to be an appropriate size threshold for applying RACT.
• EPA interprets RACM as requiring the collective evaluation of measures under assessment, and instructs states not to reject a single measure under consideration as RACM simply because the emission reduction benefits from that measure standing alone would not advance the attainment date by one year.

EPA is proposing to amend the New Source Review (NSR)/Prevention of Significant Deterioration (PSD-but we use the term NSR to refer to both programs unless otherwise indicated) regulations for the PM2.5 standards. The proposed rule for PM2.5 would not supersede existing PM10 NSR requirements, so sources in some areas would be subject to both sets of requirements.

For nonattainment NSR, EPA proposes the major source size threshold of 100 tons per year of direct PM2.5 emissions and PM2.5 precursor emissions for all source categories. This threshold is higher than the major source threshold for PM10 nonattainment areas, which is 70 tons per year. For attainment areas (PSD), the Agency proposes the size threshold of 100 tons per year of direct PM2.5 emissions and PM2.5 precursor emissions for EGUs with a heat input over 250 million British thermal units per hour and for various other large industrial sources (e.g., kraft pulp mills, portland cement plants, petroleum refineries, iron and steel mills, chemical process plants, and primary zinc, lead, and copper smelters). Sources in all other categories in attainment areas would have a threshold of 250 tons per year - the same as in PM10 attainment areas.

In addition, EPA proposes to clarify that condensable emissions must be included when determining whether a source is subject to NSR, since condensables can be a significant component of PM2.5 emissions, and sources often exclude them when projecting their emissions increases.

A significant emissions increase is the amount (in tons per year) by which a source must increase its emissions in order to trigger NSR. EPA proposes ten tons per year as the significance threshold for PM2.5 emissions.

EPA is also proposing significant emissions increase levels for PM2.5 precursors, even though (as discussed above) two of the precursors - VOCs and ammonia - will be regulated under the PM2.5 NSR program only on a case-by-case basis. EPA's preferred option is to use the existing NSR significant emissions increase levels: 40 tons per year for NOx, SO2, and VOCs. (Ammonia is not currently regulated under NSR.) As a second option, EPA proposes to use the same levels for precursors as for direct PM2.5 emissions (i.e., ten tons).

The PSD program requires affected sources to perform an ambient air quality impact analysis to show that the emissions from the source will not cause or contribute to a PSD increment, a NAAQS violation, or an adverse impact in a Class I area (a federally protected area, such as a national park). The Agency is considering establishing a PM2.5 "significant impact level" (SIL), below which a source would be exempt from PSD. The Agency has already established SILs for SO2, NO2, and PM10. EPA believes that it would be difficult to establish SILs for PM2.5 precursors, because of the limited capability of existing models to attribute PM2.5 formed from precursors to specific stationary sources.

In addition, PSD requires continuous preconstruction ambient air quality monitoring for criteria pollutants emitted in significant amounts. A source can be exempted from this requirement if the maximum net emissions increase is modeled to be less than the significant monitoring concentration (SMC), or if the existing monitored ambient concentrations are less than the prescribed SMC. EPA is reconsidering its current approach, and offers a number of options to replace its current policy on PSD preconstruction monitoring.

The NSR program requires that new emissions be offset by emissions reductions, by an amount that is at least equal to the emissions increase. EPA proposes to establish an offset ratio (the emissions reduced compared to the emissions increased) of "at least 1:1 on a mass basis" for direct PM2.5 emissions. If EPA designates a PM2.5 precursor as an NSR pollutant, then that pollutant will also be subject to the offset requirement. Accordingly, because EPA plans to regulate SO2 and NOx as precursors, the Agency proposes that SO2 and NOx be subject to the offset requirement, while VOCs and ammonia would be subject to the requirement only if they are designated as precursors. The offset ratio for precursor pollutants would also be at least 1:1.

EPA is proposing to allow inter-precursor trading to comply with the offset requirement. Specifically, EPA proposes to allow increases in emissions of direct PM2.5 to be offset by decreases in emissions of a PM2.5 precursor, and to allow increases in PM2.5 precursor emissions to be offset by decreases of a different PM2.5 precursor or by direct PM2.5 emissions. Such trades would be allowed with EPA's approval, but only if the state demonstrated that the trade would be beneficial in reducing overall ambient concentrations of PM2.5. EPA acknowledges that this program could be difficult to administer, but advocates it nonetheless.

Since April 5, 2005, the effective date of nonattainment designations, states have been required to issue NSR permits that include PM2.5 requirements. Between that time and this rule, EPA has permitted states to use the PM10 nonattainment NSR program as a surrogate for a PM2.5 program, but once the PM2.5 implementation rule is finalized, states will no longer be allowed that option. Instead, states will be required to implement a transitional PM2.5 nonattainment NSR program under Appendix S of 40 CFR part 51. In general, this means that new and modified sources will have to meet LAER and obtain offsets for PM2.5 emissions.

EPA is still determining how to deal with precursors for NSR purposes during the interim period. As stated earlier, while EPA is proposing to require that SO2 be treated as a precursor, the Agency also proposes to allow states to exempt NOx from its PM2.5 NSR program in a specific area under certain circumstances. However, EPA notes that during the SIP development period states will face substantial hurdles in making the required determination regarding precursor contribution, because they will still be in the initial stages of gathering information. For that reason, EPA is considering whether NSR applicability to precursors should be stayed during the SIP development period.

EPA offers three possibilities for the PSD program during the interim period. One of the options - the one EPA seems to favor - is to continue implementing the 1997 guidance to use PM10 as a surrogate for PM2.5. However, to use this option, states must require sources to demonstrate that their emissions will not cause or contribute to a violation of the PM2.5 NAAQS, and must include condensable particulate matter emissions in determining PSD applicability and control requirements. EPA adds this second requirement because, although PM10 is an adequate surrogate for PM2.5 in most respects, this is not the case if a source emits significant amounts of condensable emissions that would not be counted under the state's PM10 PSD program.Ð

EPA proposes that in states that achieve their CAIR reductions entirely from EGUs, CAIR will satisfy SO2 and NOx RACT requirements, provided that for NOx RACT, existing SCR controls operate year-round in nonattainment areas beginning in 2009. This aspect of the rule may come as a surprise to companies that made the decision to install SCR controls earlier, and which are thereby disadvantaged.

According to EPA, it has not determined the level of control that would constitute source-specific RACT, but it concludes, nonetheless, that CAIR will satisfy RACT. In this regard, the Agency's conclusion appears to have been foreordained.

The proposed implementation rule signals a significant change in the way in which sources are required to treat condensable particulate matter. Condensable PM is not measured in stack tests and has not been considered as part of a plant's emissions. However, because condensables often make up a large share of a plant's PM2.5 emissions, the new rule proposes to include them in the determination of a source's particulate emissions. Doing so will have serious implications.

The science of precursors and PM2.5 is still evolving. While direct PM2.5 can be attributed to a specific source, it is virtually impossible to ascribe to a specific source PM2.5 that forms from precursors. In light of this problem, it is difficult to imagine that inter-precursor trading will fly.

The PM2.5 Implementation Rule will only further complicate the already labyrinthine NSR program. For example, the rule leaves for case-by-case resolution the Agency's approach to PM2.5 precursors. Different nonattainment areas may have different requirements; some may even have to regulate ammonia, which up to now has avoided regulation as an NSR pollutant.