TECHNICAL SUPPORT FOR ACCOUNTING FOR DESTRUCTION OR REMOVAL EFFICIENCY FOR ELECTRONICS MANUFACTURING FACILITIES UNDER SUBPART I       Office of Air and Radiation, U.S. Environmental Protection Agency
                                       
                                  August 2012
                                       





1. Introduction
This document provides technical support for the proposed changes to the provisions to account for controlled emissions associated with point-of-use (POU) abatement systems under subpart I, Electronics Manufacturing, of the Greenhouse Gas Reporting Program (GHGRP). Specifically, the Environmental Protection Agency (the EPA) is proposing the following changes to subpart I: 
      1)       revising the default destruction or removal efficiency (DRE) value for semiconductor manufacturing facilities to establish new default DREs grouped by gases and process types (etch and chamber cleaning); 
      2)       providing facilities multiple options to directly measure abatement system DREs;
      3)       revising the provisions that specify the number of abatement systems that must be tested, and the required frequency of abatement system testing to determine fab-specific DRE values (i.e., the random sampling abatement system testing program, or the RSASTP); and 
      4)       expanding the provisions for ensuring proper abatement systems installation, operation, and maintenance to include an on-site a maintenance plan for abatement systems.
This document was developed using information from two sources. The first is the Technical Support Document for Process Emissions from Electronics Manufacture (e.g., Micro-electro-mechanical systems, liquid crystal displays, photovoltaics, and semiconductors): Proposed Rule for Mandatory Reporting of Greenhouse Gases, Revised  -  November 2010 (hereafter, the November 2010 TSD), available in docket EPA-HQ-OAR-2009-0927. The second is non-confidential materials submitted to the EPA by the Semiconductor Industry Association (SIA).  These materials are listed at the end of this document.
2. Revised Default DRE Values
The December 2010 subpart I rule provided all electronics manufacturing facilities the option to use a default DRE value (60 percent) so long as abatement systems for which the default DRE was being applied were designed to abate fluorinated greenhouse gases (F-GHGs) and/or nitrous oxide (N2O) and the systems were properly installed, operated, and maintained (see Section 4). The default DRE of 60 percent was developed using data from the EPA DRE measurement studies done through the EPA PFC Reduction/Climate Partnership for the Semiconductor Industry. The data set from these studies was limited and relatively small in comparison to the DRE data set SIA submitted to the EPA after subpart I was finalized (see November 2010 TSD and SIA, 2012a for the data sets). The SIA data set represents the most comprehensive DRE database the EPA has seen to date. The DREs provided by SIA were based on measurements made by two semiconductor device manufacturers. The DRE measurements for the most part (126 tests) were done in 2011 by one company according to Method 1 of the EPA Protocol for Measuring Destruction or Removal Efficiency (DRE) of Fluorinated Greenhouse Gas Abatement Equipment in Electronics Manufacturing (hereafter the EPA DRE Protocol, available in docket EPA-HQ-OAR-2009-0927).  A small portion of the DRE data (19 tests) submitted by SIA represents tests done over the time period 2003 to 2011, using a method comparable to Method 2 in the EPA DRE Protocol. SIA analyzed its DRE data to determine if there were differences in DREs due to the measurement method used. Based on this analysis the SIA concluded that "at most, any test method difference would appear to be very small." (SIA, 2012a)
Using the SIA DRE dataset and available EPA DRE data, the EPA is proposing to revise the subpart I default DRE for semiconductor manufacturing processes. The revision would provide multiple default DREs grouped by gas and process type as opposed to one default factor for all gases and process types. The EPA is proposing this change for several reasons. First, the EPA now has access to a larger DRE database that covers more gas and process combinations. Second, some gases are more difficult to abate than others due to differences in the strength of chemical bonds, which can now be reflected in the default DREs due to the broader DRE database. For example CF4 is known to be harder to abate in combustion systems (the most commonly used abatement systems per SIA, 2012a) than other F-GHGs, which will typically result in a lower DRE for CF4 (Beu, 2005). Finally, abatement system DREs are dependent on inlet and outlet system flows and it is known that process flow rates are higher for chamber clean processes than etch processes (larger precursor flows and dilution factors), which -- for a given type of abatement system -- will typically result in lower DREs for chamber cleaning than for etch.
To assess how to establish new default DREs for semiconductor manufacturers the EPA analyzed the DRE data and supporting information submitted by SIA and data from the November 2010 TSD. 
 Etch Process Type Default DREs
Table 1 presents a summary and analysis of the available etch DRE data SIA submitted to the EPA and DRE data available through the EPA DRE measurement studies (SIA, 2012a and November 2010 TSD). The EPA analyzed the etch DRE data by looking at the mean and the standard deviations of the data by gas. To analyze potential default DREs the EPA evaluated the value of arithmetic mean minus two standard deviations of the population. This conservative method was employed to account for the limited representativeness of the SIA DRE dataset: while the number of data points is relatively large, the data comes almost exclusively from one abatement system manufacturer (i.e. approximately 97 percent of the systems tested were manufactured by one abatement system supplier). Based on SIA abatement systems inventory information, the EPA estimated that over 70 percent of systems installed in the US are manufactured by that one supplier. Five other suppliers are also in the market in the US; about three percent of the field DRE performance data are for systems from some combination of these five manufacturers (SIA, 2012a). DRE performance can vary depending on the specific design of each manufacturer, which is often proprietary. Consequently, a conservative approach was used to estimate the DRE.

Table 1. Summary and Analysis of All Available Etch DRE Measurement Data (SIA, 2012a and 2012e and November 2010 TSD)
                                      Gas
                                Mean (Percent)
                         Standard Deviation (Percent)
                        Number of Data Points Available
                  Mean Less Two Standard Deviations (Percent)
Etch 
CF4
                                     63.93
                                     32.19
                                      34
                                     -0.45
C2F6
                                     99.00
                                      NA
                                       1
                                     99.00
C4F6
                                     99.06
                                     0.28
                                       5
                                     98.50
C4F8
                                     98.85
                                     0.66
                                      14
                                     97.53
NF3
                                     99.27
                                     0.47
                                      14
                                     98.32
SF6
                                     99.14
                                     0.53
                                       4
                                     98.08
CH2F2
                                     98.94
                                     0.73
                                      16
                                     97.49
CHF3
                                     99.18
                                     1.54
                                      19
                                     96.09
 NA: Standard deviation could not be calculated because only one data point was available.
Due to the similarities of the calculated etch DREs (as seen in Table 1) for many gases used in etch processes, the EPA is proposing to group C4F6, CHF3, C4F8, NF3, SF6, and CH2F2 used in etching into one DRE category and assign one average DRE of 98 percent. A DRE of 98 percent was selected as it was the approximate average of all of the aforementioned gas DREs.
The available DRE measurements for CF4 emissions from etch processes varied across a wide range of values as indicated by the large standard deviation for the total available CF4 etch DRE data set seen in Table 1. This variability was due to underperforming abatement systems resulting from operating outside manufacturers' specifications.  Following the same guidelines used to establish the initial subpart I DRE (see November 2010 TSD), the EPA is proposing to remove some of the available data from the CF4 etch data set because the data represent the DREs of systems that were not properly installed, operated, and/or maintained. Based on an EPA request, SIA identified the CF4 measured DREs that were associated with systems that were not properly operated. Approximately half of the available CF4 measured DREs were removed from the dataset based on SIA input (SIA, 2012f). This resulted in 15 of the 34 CF4 etch DRE measurements summarized in Table 1 being removed from the dataset that could be used to develop a default DRE. Given that such a high percentage of data were collected under operating conditions that do not meet the requirements in the current subpart I (i.e., systems that were not abating CF4 as expected because of improper operation and maintenance), the EPA has concerns about the reliability of the entire data set. Therefore, the EPA is proposing to maintain the current DRE of 60 percent for CF4 for the etch process type. 
Little (i.e., one data point) or no DRE data were available for four gases emitted from etch processes, C2F6, CH3F, C3F8 and C5F8. Therefore, the EPA is proposing to assign default DREs for these gases used in the etch process type on the following basis:
   * C2F6: one DRE measurement of 99.0 percent is available for this F-GHG; however, in the absence of additional DRE measurements, the EPA is proposing to assign a default DRE of 60 percent to C2F6, the same DRE for CF - 4, because C2F6, like CF - 4, is also known to be difficult to abate. Both are highly fluorinated with no C-H bonds or C=C double bonds (Tsang et. al., 1998), making these compounds relatively harder to abate than most other F-GHGs. 
   * CH3F, C3F8, and C5F8: no DRE data are available for these F-GHGs, so the EPA is proposing to maintain the previous subpart I DRE value of 60 percent.
Chamber Clean Process Type Default DREs
As can be seen by the large standard deviations in Table 2, a wide range of NF3 DREs for chamber clean were measured due to underperforming abatement systems that were not properly installed, operated, and/or maintained (SIA, 2012e).  The EPA analyzed the chamber clean DRE data to determine a default DRE using the mean DRE less two standard deviations for the same reasons as stated above for the etch default DREs.  
Table 2. Summary and Analysis of All Available NF3 Chamber Clean DRE Measurement Data (SIA, 2012a)
                                      Gas
                              (Cleaning Sub-type)
                                Mean (Percent)
                         Standard Deviation (Percent)
                        Number of Data Points Available
                  Mean Less Two Standard Deviations (Percent)
Chamber Clean
NF3 (All sub-types combined)
                                     88.72
                                     14.61
                                      49
                                     62.22
NF3 (In-Situ Plasma)
                                     88.67
                                     16.71
                                      22
                                     61.04
NF3  (In-Situ Thermal)
                                     98.90
                                      NA
                                       1
                                      NA
NF3 (Remote Plasma)
                                     88.37
                                     13.08
                                      26
                                     62.22
   NA: Only one data point available, standard deviation could not be calculated. 
As was done for CF4 etch DREs, the EPA requested that SIA identify the NF3 chamber clean DREs that were associated with abatement systems that were not properly installed, operated, and/or maintained. Removing the four DRE data points that represented underperforming systems (as detailed in SIA, 2012e) resulted in the DRE values shown in Table 3 below.

Table 3. Analysis of Revised NF3 Chamber Clean DRE Dataset (SIA, 2012a and SIA, 2012e)
                                      Gas
                              (Cleaning Sub-type)
                                Mean (Percent)
                         Standard Deviation (Percent)
                  Mean Less Two Standard Deviations (Percent)
NF3 (All sub-types combined)
                                     92.18
                                     8.63
                                     74.93
NF3 (In-Situ Plasma)
                                     93.23
                                     7.94
                                     77.35
NF3  (In-Situ Thermal)
                                     98.90
                                      NA
                                      NA
NF3 (Remote Plasma)
                                     91.03
                                     9.27
                                     72.48
           NA: Only one data point available, standard deviation could not be calculated.
In reviewing the results displayed in Table 3, the EPA decided to propose a default DRE for NF3 of 75 percent for the entire chamber clean process type, and to not propose specific default DREs for process sub-types. This was done for simplicity and because of the close proximity of the DREs for in-situ plasma and remote plasma clean, which are the most commonly used chamber clean processes in the industry.  
For all other gases used in chamber clean, the EPA is proposing to maintain the previous subpart I default DRE of 60 percent because no other data are available to support proposing revised DREs. 
N2O Default DRE
Because no data are available to support establishing a default DRE for N2O used in chemical vapor deposition or other processes, the EPA is proposing to maintain the previous subpart I default DRE of 60 percent for N2O emissions.
If additional DRE data are submitted to the EPA in response to this proposal, they will be evaluated in a similar manner to the DRE analyses above.
Revised Default DRE Summary
A summary of the proposed, expanded semiconductor manufacturing default DREs is provided in Table 4.
Table 4. Proposed DRE Default Values for Semiconductor Manufacturing Facilities 
Process Type/Gas
Proposed Default DREs
Etch
CF4, C2F6, C3F8, C5F8, CH3F (and all other F-GHGs not listed in the subsequent row)
60%
CHF3, CH2F2, C4F8, NF3, SF6, C4F6
98 %
Chamber Clean
NF3
75 %
All other gases
60 %
N2O
CVD and all other N2O-using processes
60 %
The EPA is proposing that the revised default DRE values only apply to semiconductor manufacturing facilities because the EPA did not receive any additional data for other electronics manufacturing facilities to support a change. For the use of abatement systems at other electronics manufacturing facilities, the EPA is proposing to maintain the previous subpart I default DRE of 60 percent. 
3. Measurement of Fab-Specific DREs 
As an alternative to using default DREs, the December 2010 subpart I provides all electronics manufacturing facilities the option to measure and develop fab-specific DREs to use in place of the default DREs. The EPA is proposing to amend the provisions for measuring DRE to be more flexible and to require less frequent abatement system testing.  
   3.1. Methods for Measuring Fab-Specific DREs
The November 2010 TSD and the EPA DRE Protocol contain detailed discussions on the justification for the required adherence to the EPA DRE Protocol for proper measurement of fab-specific DREs. However, based on new information that SIA submitted to the EPA on an alternate method for measuring total volume flow (TVF) of an abatement system, the EPA is proposing to amend the provisions for measuring fab-specific DREs for all facilities covered under subpart I. The EPA is proposing to allow facilities to use a method for determining TVF of gas through the abatement system that is an alternative to the method outlined in the EPA DRE Protocol which suggests the use of quadrupole mass spectroscopy (QMS). The proposed alternative method would consist of injecting a non-native polar tracer gas into the abatement system outlet stream and measuring the downstream tracer gas concentration using Fourier transform infrared (FTIR) spectroscopy to determine TVF. This proposed alternative, which is a lower cost option than using a QMS-based method (ISMI, 2010), was tested by SIA on three abatement systems at one semiconductor manufacturing facility. The results of the tests comparing the use of the FTIR-based and QMS-based methods are presented in the International SEMATECH Manufacturing Initiative (ISMI) Technology Transfer #10095115A-ENG Comparison of Fourier Transform Infrared (FTIR) and Quadrupole Mass Spectroscopy (QMS) Methods for Determining POU Abatement System Effluent Flow.  The EPA reviewed the method comparison and concluded that the results of these tests adequately showed that using the suggested alternative FTIR method was suitable for determining dilution across an abatement system, provided that the injected tracer gas is well-mixed at the abatement system exit stream sampling location (ISMI, 2010), or that the sampling procedure accounts for any stratification (lack of mixing). Therefore the EPA is proposing to include this alternate FTIR method as an option in Appendix A to subpart I, with an additional procedure to ensure that the sampling location and procedures account for the fact that the tracer gas may not be well mixed.
As discussed above, ensuring adequate mixing is essential to provide accurate results, but can sometimes be difficult to achieve due to the short duct lengths commonly found downstream of abatement systems. Ideally, sampling would be done well downstream of the abatement system to ensure complete mixing. 
To demonstrate complete mixing in the abatement system exhaust stream, or to account for any stratification in the exhaust if it is present, the EPA is proposing that facilities using the alternative method to determine dilution should use an adaptation of section 8.1.2 of EPA Method 7E, Determination of Nitrogen Oxides Emissions From Stationary Sources (Instrumental Analyzer Procedure), in 40 CFR part 60, Appendix A-4. This adaptation, which is also included in the proposed Appendix A to subpart I, includes the following requirements: 
   * The concentration of the tracer would be measured at three traverse points at 16.7, 50.0, and 83.3 percent of the diameter of the duct. 
   * If the tracer gas concentration at each traverse point differs from the mean concentration for all three traverse points by no more than +-5.0 percent of the mean concentration, the gas stream would be considered un-stratified and the facility would be allowed to collect samples of the tracer gas from a single point that most closely matches the mean.
   * If the 5.0 percent criterion were not met, but the concentration of the tracer gas at each traverse point differed from the mean concentration for all traverse points by no more than +-10.0 percent of the mean, a facility would be able to take samples from two points and use the average of the two measurements. The sampling could be performed at any two of the three points spaced at 16.7, 50.0, or 83.3 percent of the line. 
   * If the concentration at each traverse point differed from the mean concentration for all traverse points by more than +-10.0 percent of the mean but less than +-20.0 percent, the facility would take samples of the tracer gas from three points at 16.7, 50.0, and 83.3 percent of the measurement line and use the average of the three measurements. 
   * If the gas stream were found to be stratified because the +-20.0 percent criterion for a three-point test were not met, the facility would be required to locate and sample the tracer gas from traverse points in accordance with Sections 11.2 and 11.3 of EPA Method 1 in Appendix A-1 to Part 60. The tracer gas concentration used for measuring flow would be determined from the average of the concentrations measured at each of the traverse points.
This method has not been used in practice at an electronics manufacturing facility, but it is expected to be sufficient for demonstrating complete mixing and ensuring that a concentration value used in calculations is representative of a well-mixed stream. This is because the method outlined above is based on the EPA's Method 7E, which has been used in practice in other applications. Also the method includes thresholds for differences in measured concentrations and contingencies for when a threshold cannot be met. This structure helps ensure that the effluent stream concentration used is representative of a well-mixed stream at a sampling location.
In addition, the EPA is also proposing to allow facilities to request approval to use an alternative sampling and analysis method that is not included in subpart I to measure abatement system DRE. The EPA is proposing that alternative test methods would have to be validated according to EPA Method 301 (Field Validation of Pollutant Measurement Methods From Various Waste Media). Under this proposal, prior to using an alternate test method, the facility would be required to submit the method along with a proposed test plan to the EPA for review and receive approval from the EPA. This process would ensure that any method outside of those specified in subpart I are viable and appropriate. Once the EPA approves an alternate method, the method could be used by other fabs that meet the same conditions under which the alternative method was approved. The EPA would specify those conditions in the approval of the alternative method. The regulatory language that addresses this option is included in the body of the text, not in the proposed Appendix A to subpart I.
Regardless of the method used to determine dilution (either using krypton and a QMS as specified in the EPA DRE Protocol, a polar species and a FTIR (as proposed in Appendix A of subpart I), or an alternative method approved by the EPA), facilities would still be required to meet the EPA DRE Protocol performance benchmark, which is that the relative error of the true fraction emitted must be less than or equal to five percent, and to follow all other aspects of the EPA DRE Protocol with one exception.  That exception is that facilities may use the FTIR and QMS Protocols specified in the proposed Appendix A of subpart I in lieu of the 2006 ISMI Guideline for Environmental Characterization of Semiconductor Process Equipment FTIR and QMS Protocol as required in the EPA DRE Protocol.  The FTIR and QMS protocols contained in the proposed Appendix A to subpart I are modeled after the FTIR and QMS Protocols in Appendix A of the 2009 ISMI Guideline for Environmental Characterization of Semiconductor Process Equipment and reflect advancements in experience in using FTIR/QMS in the industry. Although ISMI-SEMATECH is a common source of best practices for the semiconductor manufacturing industry, the EPA does not have any information on whether, and in what way, the modifications to the FTIR and QMS protocols in Appendix A of the 2009 ISMI Guideline would impact DRE measurements, compared to the 2006 ISMI Guideline.  Therefore, the EPA is specifically seeking comment on the appropriateness of using the 2009 ISMI Guideline as the basis for the proposed Appendix A to subpart I. 
   3.2. Abatement System Sampling Protocol
The EPA is proposing to modify the random sampling abatement system testing program (RSASTP) in 40 CFR 98.94(f)(4)(ii) to make the required testing frequency less burdensome for facilities that elect to measure fab-specific DREs. Table 5 below summarizes the elements of the RSASTP as finalized in the December 2010 version of subpart I, the proposed changes to the RSASTP, and the technical justification for the modifications.
             Table 5. Summary of Proposed Revisions to the RSASTP
                           December 2010 Provisions
                          Proposed Revised Provisions
                    Justification for Modification/Addition
Required Testing Frequency
         * Each year, test three systems or 20% of installed abatement systems for each class.
         * Do not repeat measurements for systems measured in the previous year.
         * If a DRE was measured for a particular system during the previous 2 years, the facility must use the most recently measured DRE.
         * If the DRE was not measured for a particular system during the previous 2 years, use the simple average of the measured DREs for that class.

         * For each process type/sub-type and gas combination, test 10% of systems in each fab annually for two years (20% total) or 20% in the first year to set a baseline average DRE for that process type/sub-type and gas combination. 
         * Use default DREs until DREs have been measured for 20% of systems. 
         * For every 3-year period after, test 15% of the systems in the fab for each process type/sub-type and gas combination. A fab can choose to test 15% in the first year of a 3 year period but must test at least 5 percent of the systems each year until 15 percent are tested. 
         * The fab-specific DRE would be calculated as the average of the first 2 years (or first year if 20% are tested) and thereafter as the average of the most recent t 30% tested. 
The EPA is proposing to modify the RSASTP because:
         *    The previous definition of "class" combined with the permutations of process and abatement type resulted in testing requirements representing an average of 45 percent of the total number of abatement systems in a fab, which is higher than the 20 percent required to be tested by the RSASTP. (SIA, 2012a)
         *    The proposed changes would reduce testing burden by requiring fewer tests compared to current subpart I, yet provide sufficient information to ensure that the DRE values used represent current system performance as abatement systems age (see DRE versus abatement system age information in SIA, 2012a).
Definition of Class
Class means a category of abatement systems grouped by manufacturer model number(s) and by the gas that the system abates, including N2O and CF4 direct emissions and by-product formation, and all other F-GHG direct emissions and by-product formation. Classes may also include any other abatement systems for which the reporting facility wishes to report controlled emissions, provided that class is identified.
Delete the definition of "class." 

Require that DREs be determined for abatement systems based on the gas(es) that the systems abate and the process type connected to the systems, including etch/wafer clean, chamber clean, and all N2O using processes. 
The EPA is proposing to delete the definition of class because:
   *    Testing by gas and process type/sub-type combination would simplify apportionment and uptime calculations, and recordkeeping, and hence reduce burden.
   *    Abatement system categories based on gas and process type combinations would be consistent with calculation methods for subpart I default emission factor based methods.
   *    Broader categories of abatement systems would reduce the number of systems that need to be tested. 
Accounting for Under-Performing Abatement Systems in Calculating Emissions or Average DRE
Not specifically addressed.
(If a facility uses the default DRE of 60%, the facility must certify that the system is installed, operated, and maintained according to the manufacturer's specifications.

If a facility measures the DRE of a system, they must use that measurement in calculating emissions.)
   * If proper maintenance and operation procedures are followed, and the measured DRE of an abatement system is below the manufacturer-claimed DRE, the data from the low DRE test would be included in calculating the fab-specific average DRE for that gas and process combination. 
   * If proper maintenance and operation procedures are not followed, then the appropriate operational change or system maintenance (per the manufacturer instructions or the fab maintenance plan) would need to be implemented, and a retest would be required that year to replace the measured value. Alternatively, the fab could use the measured value in the DRE determination and then include the same system in next year's abatement system testing in addition to the testing of randomly selected systems.
These provisions would be added as part of requirements for DRE measurement because they would:
   * Ensure that the DREs used in calculations reflect actual performance.
   * Ensures measured values are used by not allowing a reporter to select the higher of the default or the measured DRE.

Testing of Systems That May Be Disruptive to Production
Not addressed.
(No provisions to allow deferring testing of abatement systems if the testing would be disruptive to production.)
Where testing of systems randomly selected for testing would be disruptive to production, such systems could be (a) replaced by another randomly selected system for testing, and (b) returned to the sampling pool for subsequent testing. If testing of a system is deferred, that system could be chosen for testing in one of the next two sampling years. The EPA is proposing that a system cannot be returned to the pool in more than three consecutive selections.
This provision would be added to address situations where testing of a specific abatement system would disrupt manufacturing productivity.
IF deferred, a system could be specifically selected for testing so that production and testing could be coordinated to avoid disrupting production.

4. Ensuring Proper Abatement System Installation, Operation, and Maintenance
As discussed in Section 3.6.1 (Proper Installation, Operation, and Maintenance) of the November 2010 TSD, proper installation, operation, and maintenance of the abatement systems in accordance with manufacturer specifications is essential to ensuring that abatement systems are performing as expected. This concept is supported by the circumstances described in Section 2 of this document for CF4 and NF3, in which some DREs were lower than expected as a result of system operational and/or maintenance issues. The  wide range of measured values for CF4 used in etching and NF3 used in chamber cleaning from the complete available DRE dataset, which includes data that was removed in the Section 2 analysis, can be seen in Figure 1 below (where `IS' denotes the In-Situ chamber cleaning technology and `RPS' the Remote Plasma System, a.k.a. the remote cleaning process). The broad ranges were not typical of the DRE data provided by SIA, but were not wholly unexpected due to issues the EPA has seen with CF4 DREs during measurement studies (see EPA 2008a and 2008b) and because of the known high flow rates of NF3 used during remote chamber clean processes (SIA, 2012c and SIA, 2012e).
                                       
Figure 1. Range in Measured DREs for CF4 Etch and NF3 Chamber Clean (SIA, 2012a)
SIA suggests the variation in flows through the abatement system that was outside of manufacturer specifications were the primary reason for the observed variation in NF3 DREs for chamber clean processes (SIA, 2012a and SIA, 2012b). Low CF4 DREs resulted from operating and maintaining the abatement systems outside the manufacturer specifications; for example, system inlet/outlet volumetric flows were at higher levels than recommended by the manufacturer (SIA, 2012b). More specifically, for the low CF4 DREs associated with etch processes, it was found that operators were not complying with the manufacturers' recommended set points for flow rate and/or pressure for fuel, oxygen, and air that should have been verified during abatement installation, and that ongoing maintenance to ensure that the systems performed as expected was not being performed. For systems that were operating as expected during testing for CF4, it was determined that the flow, fuel, and oxygen settings were properly set and the flow out of the abatement system was within the recommended manufacturer range (SIA, 2012b).
Abatement system manufacturers provide general guidance on operation and maintenance of their products. However, lessons learned from the EPA DRE measurement studies and from SIA's experiences highlight the need to expand the subpart I provisions to require facilities to take more specific actions to ensure abatement systems are properly operated and maintained over time. Therefore, the EPA is proposing to add a requirement for facilities to develop, follow, and keep on-site a maintenance plan for abatement systems.  The EPA is proposing to require that the plan must include a defined preventive maintenance process and checklist (built on the manufacturer's recommended maintenance program) and a corrective action process to follow whenever an abatement system fails to operate properly.
5. References 
Beu, L. (2005). Reduction of Perfluorocarbon (PFC) Emissions: 2005 State-of-the-Technology Report, TT#0510469A-ENG, International SEMATECH Manufacturing Initiative (ISMI), December 2005.  Available at: http://www.epa.gov/highgwp/semiconductor-pfc/documents/final_tt_report.pdf. 
Burton (2007).  Assessing the need for FC abatement standards, Solid State Technology, January 2007.  Available at: http://www.electroiq.com/index/display/semiconductors-article-display/281422/articles/solid-state-technology/volume-50/issue-1/departments/industry-forum/assessing-the-need-for-fc-abatement-standards.html.
EPA (2008a). Developing a Reliable Fluorinated Greenhouse Gas (F-GHG) Destruction or Removal Efficiency (DRE) Measurement Method for Electronics Manufacturing:  A Cooperative Evaluation with NEC Electronics, Inc. (EPA 430-R-10-005). Office of Air and Radiation Office of Atmospheric Programs, Climate Change Division, U.S. Environmental Protection Agency, Washington, DC.  Available at: http://www.epa.gov/highgwp/semiconductor-pfc/documents/nec_report.pdf. 
EPA (2008b). Developing a Reliable Fluorinated Greenhouse Gas (F-GHG) Destruction or Removal Efficiency (DRE) Measurement Method for Electronics Manufacturing:  A Cooperative Evaluation with Qimonda (EPA 430-R-08-017). Office of Air and Radiation Office of Atmospheric Programs, Climate Change Division, U.S. Environmental Protection Agency, Washington, DC.  Available at: http://www.epa.gov/highgwp/semiconductor-pfc/documents/qimonda_report.pdf. 
EPA (2009). Developing a Reliable Fluorinated Greenhouse Gas (F-GHG) Destruction or Removal Efficiency (DRE) Measurement Method for Electronics Manufacturing:  A Cooperative Evaluation with IBM (EPA 430-R-10-004), Office of Air and Radiation Office of Atmospheric Programs, Climate Change Division, U.S. Environmental Protection Agency, Washington, DC.  Available at: http://www.epa.gov/highgwp/semiconductor-pfc/documents/ibm_report.pdf. 
ISMI (2010). Comparison of Fourier Transform Infrared (FTIR) and Quadrupole Mass Spectroscopy (QMS) Methods for Determining POU Abatement System Effluent Flow. International SEMATECH Manufacturing Initiative (ISMI) Technology Transfer #10095115A-ENG. October 30, 2010.  Available at: http://ismi.sematech.org/docubase/document/5115aeng.pdf
SIA (2012a). SIA Briefing Paper on Abatement Issues: Destruction Removal Efficiency (DRE), January 10, 2012. Available in EPA docket EPA-HQ-OAR-2011-0028. 
SIA (2012b). Questions Generated from SIA/EPA Conference Calls and Outstanding Questions from Work Plan Appendices 120329 SIA EPA Outstanding Questions, March 29, 2012. Available in EPA docket EPA-HQ-OAR-2011-0028. 
SIA (2012c). CVD Abatement units with flow data. March 6, 2012. Available in EPA docket EPA-HQ-OAR-2011-0028. 
SIA (2012d). Etch DRE testing with flow data.  March 6, 2012. Available in EPA docket EPA-HQ-OAR-2011-0028. 
SIA (2012e). Questions for SIA Response from April 19, 2012 SIA/EPA Call With April 25, 2012 SIA Response. April 25, 2012. Available in EPA docket EPA-HQ-OAR-2011-0028. 
SIA (2012f). Outstanding Questions / Topics from SIA / EPA Discussions as of April 26, 2012. May 9, 2012. Available in EPA docket EPA-HQ-OAR-2011-0028.  
SIA (2012g). Proposal to Incorporate ISMI 2009 Guideline References in Subpart I. May 17, 2012. Available in EPA docket EPA-HQ-OAR-2011-0028. 
Tsang et al. ( 1998) Tsang, W., Burgess Jr., D. R., and Babushok, V. (1998) "On the Incinerability of Highly Fluorinated Organic Compounds," Combustion Science and Technology, 139:1, 385-402.
