&EPA
United States
Environmental Protection
Agency
Version 1
Protocol for Measuring Destruction or Removal Efficiency (DRE) of
Fluorinated Greenhouse Gas Abatement Equipment in Electronics
Manufacturing
March 2010
Office of Air and Radiation
Office of Atmospheric Programs, Climate Change Division
EPA 430-R-10-003
-------
Table of Contents
List of Tables 3
List of Figures 3
PREFACE 4
1. INTRODUCTION 5
1.1 Protocol Purpose 6
1.2 Protocol Objectives 7
1.3 Protocol Scope 7
1.4 Development of the Protocol 8
1.5 Expected Results 8
2. MEASUREMENT PLAN, PREPARATION, METHODOLOGY, AND DATA
ANALYSIS 8
2.1 Prepare Measurement Plan 8
2.1.1 Description of Experimental System 9
2.1.2 Sampling Configuration 9
2.1.3 Required Resources 9
2.1.4 Measurement Schedule 10
2.1.5 Safety 10
2.1.6 Quality Assurance/Quality Control 11
2.2 Measurement Methodology 11
2.2.1 Method 1—Dilution Adjusted Concentration Measurement 12
2.2.2 Method 2—Total Volume Measurement 12
2.2.3 Equipment Needed 13
2.2.4 FTIR and QMS Protocols 14
2.2.5 Calibration Curves 14
2.2.6 Flow and Dilution Measurements 16
2.2.7 F-GHG Measurements 17
2.3 Data Treatment and Analysis 18
2.3.1 Abatement System Dilution 19
2.3.1.1 Total Volume Flow 19
2.3.1.2 Dilution Factor 21
2.3.2 Abatement System DRE Estimate and Relative Errors 22
2.3.2.1 Method 1—Dilution Adjusted Concentration Measurements 22
2.3.2.1.1 Concentration Measurements 22
2.3.2.1.2 DRE 23
2.3.2.2 Method 2—Total Volume Measurements 24
2.3.2.2.1 Total Volume Measurements 24
2.3.2.2.2 DRE 24
3. BENCHMARK RELATIVE ERROR 25
3.1 Method 1—Dilution Adjusted Concentration Measurements 25
3.2 Method 2—Total Volume Measurements 25
4. DOCUMENTATION AND REPORTING 26
5. REFERENCES 28
-------
Appendix A-History of and Revisions to the Protocol 30
Appendix B- Sample Calculation 32
List of Tables
Table 1. Measurement Study Equipment Needs 14
Table 2. Acceptable Gases for Monitoring Total Abatement System Flows 17
Table 3. Minimum detection levels (CMDL) and threshold F-GHG concentrations (C*) for
measuring relative error of total fraction emitted when sample C < C* 18
List of Figures
Figure 1. Sampling Schematic for Single and Multiple Chamber Inlet Flow(s) 10
Figure 2. Sample Calibration Curve 16
-------
PREFACE
The U. S. Environmental Protection Agency (EPA) initiated preparation of this Protocol as part
of its commitment to assist the PFC Emissions Reduction/Climate Partnership for the
Semiconductor Industry (the Partnership) to achieve its voluntary fluorinated greenhouse gas (F-
GHG) emission reduction and reporting goals. F-GHGs include the most powerful and often
persistent greenhouse gases such as perfluorocarbons (PFCs), hydrofluorocarbons (FIFCs), sulfur
hexafluoride (SF6), and nitrogen trifluoride (NF3). The Partnership's reduction goal is to
decrease F-GHG emissions to 10 percent below the 1995 base-year amounts by 2010. Several F-
GHG-reducing options are available, with reduction-efficiencies ranging from 10 to 99 percent
or higher. Practical reduction strategies comprise modifying manufacturing processes—
optimizing processes and switching to alternative gases with lower emissions potential—as well
as installing point-of-use (POU) F-GHG abatement systems. Detailed methods for measuring F-
GHG emissions from manufacturing processes were standardized in 1999 and are documented
and continually updated in the industry's International Semiconductor Manufacturing Initiative
(ISMI) Guidelines. Standard methods for measuring and documenting the destruction or removal
efficiencies (DREs) from POU abatement systems have lagged those for measuring process
emissions. One objective of this document (the Protocol) is to align measurement of emissions
from manufacturing processes with those from abatement systems.
The Protocol builds on three DRE measurement studies of POU F-GHG thermal abatement
systems conducted onsite at three U. S. semiconductor manufacturing facilities, each a member
of the Partnership. Air Products performed and documented the measurements under subcontract
to ICF International, who was contracted by EPA to support development of the Protocol. In
total, four abatement systems were tested and all were connected to plasma-enhanced chemical
vapor deposition (PECVD) or etch process equipment. During the studies, DRE measurements
were made with the plasma on (during wafer processing) and with the plasma off. A key feature
of these measurement studies was development and testing of a reliable method for measuring
the dilution of F-GHGs during abatement.
This is the third version of the Protocol. It reflects two rounds of external reviews, one each for
the first and second version. The first version, prepared after measurements were made at two
facilities, was circulated to Semiconductor Industry Association (SIA), Japan Electronics and
Information Technology Industries Association (JEITA), and Taiwan's Industrial Technology
Research Institute (ITRI) as well as to abatement manufacturers and industry consultants
experienced in measuring the DRE of F-GHG abatement systems. The comments and responses
to those comments were documented and discussed with reviewers. This discussion led to the
third and latest onsite measurement study, as well as to a second version of the Protocol. The
second version of the Protocol was again circulated to a subset of the reviewers to the first
version. The comments and responses to those comments were also documented and discussed
with reviewers.
A consistent theme that emerged from EPA's testing and all the reviews of the Protocol is the
diversity of manufacturing practices and conditions across the electronics manufacturing sector.
Therefore experienced and resourceful metrologists are central to successful DRE measurements.
This Protocol is aimed for use by experienced metrologists—experienced not only in the use of
-------
advanced analytical measurement systems such as FTIR and QMS, but also in performing in-fab
measurements of F-GHG emissions. Moreover, the Protocol does not constitute a recipe—a
single approach for all circumstances—for making DRE in-fab measurements. Instead, properly
construed, the Protocol permits flexibility in measurement practice provided the measurements
achieve the performance standard that is integral to this Protocol
1. INTRODUCTION
Members of the U.S. Environmental Protection Agency's PFC Emissions Reduction/Climate
Partnership for the Semiconductor Industry (the Partnership) operate under a voluntary
agreement to reduce perfluorocompound (PFC, henceforth called fluorinated greenhouse gas, F-
GHG) emissions to 10 percent below the 1995 base-year amounts by 2010. A variety of practical
F-GHG-reducing options—with reduction-efficiencies that range from 10 to >99 percent—are
available. These options include process optimization techniques, switching to alternative gases
with lower emissions potentials, and point-of-use (POU) F-GHG abatement. While there are
industry-standard guidelines that set forth methods for measuring and documenting process
emissions (ISMI, 2006), that is not so for measuring the destruction-removal efficiencies (DRE)
of POU F-GHG abatement systems, although published reports of DRE measurement studies are
available (Beu et al., 1994; Li et al., 2001, 2002 and 2004; Lee et al., 2007).
The U.S. Environmental Protection Agency (EPA) initiated preparation of this document (the
Protocol) as part of its commitment to assist the Partnership in achieving the F-GHG emission
reduction and reporting goals. The development of this Protocol builds on three DRE
measurement studies of POU F-GHG abatement systems conducted onsite at three U.S.
semiconductor fabrication facilities (fabrication facilities are henceforth called fabs). (EPA
2008a, 2008b, and 2009)
-------
1.1 Protocol Purpose
The purpose of the Protocol is to provide a practical and reliable method for measuring DREs of
POU abatement systems of F-GHG gases used during the manufacture of electronics products,
specifically semiconductor systems, micro-electro-mechanical systems (MEMS), thin film
transistor (TFT) arrays and amorphous silicon (a-Si) and tandem amorphous silicon/
nanocrystaline silicon (a-Si/nc-Si) thin-film photovoltaic (PV) panels.1
The Protocol sets forth two specific methods for measuring abatement system inlet and outlet
flows, and hence DREs, for single or multi-chamber process tools. These methods address
measuring the mass or volume flows of F-GHGs entering and leaving the abatement system.
Both methods account for the dilution that occurs in thermal abatement systems.2
The first and simplest, Method 1 - Dilution Adjusted Concentration Measurement, involves
measuring DRE when the process tool is off. This approach will produce reliable results when F-
GHG byproducts are not formed, for example, during chamber cleaning with CF4, SF6 or with
NF3 when carbon films are not present. A variation of Method 1 may be used that imitate
byproducts formation. This involves taking appropriate steps to add proper amounts of byproduct
F-GHG to the influent abatement-system flow of unutilized F-GHG(s).
The second method, Method 2 - Total Volume Flow Measurement, is the preferred method for
measuring DRE when byproducts are formed. Method 2 measures abatement-system DRE under
actual process conditions. One of two approaches may be used when applying Method 2,
dependent upon if measurements are being taken sequentially or simultaneously.
Method 2, requires introducing a chemically stable material - a tracer or spiking agent that
neither affects the performance of nor is affected by the abatement system—into the process
line(s). Using this method will provide a reliable estimate of the effective dilution associated
with in-fab thermal POU abatement systems3 and, in turn, the DREs. This chemical-spiking
approach is a special application of ASTM E 2029 - 99 (Re-approved 2004), which provides a
method for measuring volumetric or mass flow rate of a gas in a duct, pipe, etc. using a tracer
dilution technique (ASTM, 2004). ASTM E 2029-99, like this Protocol, addresses irregular and
non-uniform flow conditions where conventional pitot tube or thermal anemometer velocity
measurements are difficult or inappropriate due to the absence of a suitable run of duct/pipe
upstream and downstream of the measurement location.
Establishing DREs based on this Protocol will (a) assure reliable comparisons of vendors'
abatement systems; (b) assure third-parties of the reliability of reported emissions; (c) improve
understanding of appropriate and necessary system maintenance; and (d) serve as a starting point
1 The electronics manufacturing sector includes, but is not limited to those industry sub-sectors that are listed. This
Protocol is applicable to all electronics sectors; but it was developed based on work at a semiconductor
manufacturing facilities. However EPA sees no limitation on why the Protocol may not be applied when measuring
DREs at other electronics manufacturing facilities.
2 Testing for the development of the Protocol focused on POU thermal abatement systems because dilution is
greatest. It is expected the Kr tracer method is suited to POU plasma systems. However, EPA was not aware of
fully operational plasma abatement systems in production fabs at the time of the Protocol's publication.
3 See footnote 2.
-------
for emerging carbon trading opportunities (i.e., measuring baseline and actual emission
reductions).4
1.2 Protocol Objectives
The objectives of the Protocol are to specify:
• a method for measuring and reporting F-GHG-specific DREs of installed, in-fab
abatement systems that are operating during normal production processing;
• a method for measuring and reporting F-GHG-specific DREs of abatement systems that
are offered to manufacturers of electronic products that use F-GHGs during product
manufacturing; and
• guidance to fab Environmental Safety &Health staff and to providers of POU abatement-
testing services on best-practice procedures for measuring, documenting and reporting of
in-fab determinations of F-GHG abatement system DREs.
1.3 Protocol Scope
The Protocol is applicable to measuring and reporting DREs of installed POU systems whose
input flows may contain CF4, CHF3, CH2F2, C2F6, C3F6, C3F8, C4F6, c-C4F8, C4F8O, C5F8, NF3
and SF6. These high-GWP gases may be used directly in fab processes and, in some
circumstances, may be formed during those processes.
The Protocol is intended to complement the guidance prepared by International SEMATECH
Technology Transfer # 06124825A-ENG (December 2006); this Protocol adopts in its entirety
Appendix A of SEMATECH TT # 06124825 A-ENG (December 2006), which explains best
known methods for measuring, among other things, F-GHG emissions. The principal difference
between the 2006 updated SEMATECH guidance and this Protocol is the emphasis given here to
POU abatement system DRE measurement and to measuring dilution across the abatement
system.
The Protocol institutes two practices to establish the veracity of reported results. The first is
adoption of a benchmark metric together with a performance standard for that metric. The
benchmark metric is the relative error, which is a precision metric, in the reported true fraction
emitted of which the DRE is its complement. The performance standard the relative error must
meet is ±5 percent relative error.5'6 The second practice is the requirement for metrologist
4 The Executive Board of the Clean Development Mechanism has final authority for approving methods for
measuring emissions of greenhouse gases, for purposes of establishing a CDM project. However, CDM does not
have authority in the United States..
5 The fraction emitted is chosen as the performance metric rather than the DRE because the relative error of the
measured DRE is a function of the true fraction emitted. As the true fraction emitted approaches one (DRE
approaching zero), the relative error becomes very large even when the relative error in the true fraction is
acceptably small. Similarly, when the true fraction approaches zero (DRE approaches one), the relative error in DRE
can become small even when the relative error in the true fraction emitted is large. When the post-abatement
concentration approaches or falls below minimum detection an alternative calculation is provided (vide infra).
6 Experience demonstrates that ±5 percent or better is readily achievable in a fab environment. In three separate
studies (EPA 2008a, 2008b, and 2009), the estimated relative error for the true fraction emitted for four POU
-------
certification, certifying that the methods set forth in the Protocol were followed, including that
the conditions under which the measurements were made (with the plasma off or actual process
conditions).
EPA wishes to strike an appropriate balance between assuring reliable results and flexibility. In
the remainder of the Protocol, where EPA anticipates the need for flexibility it has provided
examples to demonstrate the nature and extent of the permitted flexibility. EPA will consider
updating the Protocol given the availability of and access to additional testing experiences and
well-documented results.
1.4 Development of the Protocol
The Protocol has gone through five stages of development: conceptualization, onsite tests, initial
drafting, and two informal peer review processes. During the informal peer review processes
EPA received comments from both national and international parties, including semiconductor
manufacturers, equipment manufacturers, gas suppliers and analytic service providers. Appendix
A discusses some of these comments, as well if and how the Protocol was revised to reflect
them.
1.5 Expected Results
This Protocol results in an estimate for the DRE of F-GHG abatement systems, where the
relative error of true fraction emitted estimate (the benchmark metric), measured under in-fab
production processing circumstances and typical (in-fab, as-installed) POU system operating
circumstances, is less than ±5 percent (the performance standard). The value of relative error of
the true fraction emitted is determined by the methods outlined in the remainder of this Protocol.
2. MEASUREMENT PLAN, PREPARATION, METHODOLOGY, AND DATA
ANALYSIS
2.1 Prepare Measurement Plan
A measurement plan serves two main purposes. First, it provides background on the tools,
processes, and abatement systems to be tested; and second, it facilitates coordination between
facility personnel and testing service providers. An understanding of the tools, processes, and
abatement systems provides testing service providers with background on the experimental
system in preparation for the study. Coordination with facility personnel reduces the likelihood
of encountering unexpected problems of measurement execution during the study. Testing
service providers should prepare a measurement plan and submit it to facility personnel at least
two weeks prior to the measurement study. At a minimum, the measurement plan should contain
detailed descriptions of the experimental system, the sampling configuration, required/expected
systems attached to either etch or CVD process tools was below ±3 percent. Adoption of ±5 percent as a
performance standard assures the accuracy of the measured true fraction emitted will be ±5 percent or better, under
the reasonable assumptions that the accuracy of calibration gases and mass flow controllers are each ±3 percent or
better.
-------
resources/support from the fab, a measurement schedule, address safety and a quality
control/quality assurance plan (QA/QC) as described in the following sections. In preparing the
plan, attention to Appendix A of the 2006 ISMI Guidelines is essential to assuring best in-fab
FTIR and QMS measurement practice is achieved.
2.1.1 Description of Experimental System
An understanding of the tools, processes, and abatement systems to be tested prepares the testing
service provider for the onsite measurement study. For example, the experimental system setup
informs the study's sampling configuration, whereas the expected dilution, flows, and DRE
inform the type of analytical tools for data treatment needed for the study. To ensure that testing
service providers are adequately prepared for conducting onsite measurements, information on
the following aspects of the experimental system should be included in the measurement plan:
• Description of the process tool and abatement system;
• Configuration of the process tool and abatement system;
• Presence of vacuum pump purges and additional purges;
• Process operating conditions;
• Measurement conditions (i.e., plasma off or actual process conditions);
• Gases/flow rates (when measuring with the plasma off) and/or process recipe (when
measuring during actual conditions);
• Number of process chambers served by abatement system;
• Nominal abatement system reactor temperature;
• Nominal abatement system exhaust flows;
• Nominal process exhaust flows;
• Nominal dilution factor (Note: Nominal dilution factor is defined as the ratio of the
nominal abatement system exhaust flow to the nominal process exhaust flow); and
• Nominal DRE(s) for F-GHG(s) being tested.
2.1.2 Sampling Configuration
An example schematic of the sampling configuration, which illustrates the configuration of the
process tool and abatement system, should be developed by testing service providers prior to
onsite measurements. In the schematic the planned/proposed sampling and measurement layout
in relation to the process tool and abatement systems, the process tool and abatement system
effluent sampling locations, and the type of sampling port (e.g., tie-in type) should be clearly
labeled (see Figure 1).
2.1.3 Required Resources
One of the primary purposes of the measurement plan is to facilitate coordination between the
testing service providers and the facility personnel. The plan should include a sufficiently
detailed description of all resources needed from the facility including sampling ports and
fittings, gas supplies, electrical outlets, and assistance of facility personnel including roles,
responsibility, and an anticipated schedule. Advance preparation by facility personnel is essential
for maximizing the utility of testing service providers' time at the facility.
-------
To Scrubbed Exhaust
Effluent from Process
Pump to Scrubber
AH* native S'et Lip for
Multiple !ii!> i F.t.'.v
Simplf I'ump
Figure 1. Sampling Schematic for Single and Multiple Chamber Inlet Flow(s) Source: EPA, 2009.
Note: Metrologists need to be mindful of recirculation issues when returning the sampled gas flows
to exhaust lines.
2.1.4 Measurement Schedule
In order to minimize any interference with facility production, a measurement schedule should
be agreed upon prior to conducting the onsite study. The measurement schedule should clearly
indicate the sampling start and end times. Additionally, the tool operator should approve and be
familiar with the measurement plan and experimental system including the process
tools/chambers and abatement systems, the process conditions during sampling (i.e., plasma off
versus actual process conditions), and the gas flow rates/process recipe to be supplied to the tool.
2.1.5 Safety
Safety of the measurement study participants is of highest priority. Prior to conducting any work
onsite, testing service providers will be instructed by appropriate fab personnel about all
pertinent safety requirements and practices. For example, appropriate personal protective
equipment (PPE) should be worn while conducting measurements in a fabrication environment
(e.g., head protection, hearing protection, eye protection, and foot protection). Additionally,
testing service providers are responsible for acquainting themselves with the hazards that exist in
the 1C fabrication environment (e.g., hazardous gases, chemical spills, and heavy objects) and
taking all steps to avoid adverse impacts. Special consideration should be given to the benefits
and risks of mixing pyrophoric/flammable and oxidizing gases. Most abatement systems are
10
-------
designed to prevent such mixing prior to combustion and measurement methodologies that defeat
manfuacturers' design goal should be thoroughly considered.
2.1.6 Quality Assurance/Quality Control
In preparation for the measurement study, testing service providers should gather information on
the experimental system and its expected characteristics, such as expected reactor temperature,
exhaust flows, etc (see Section 2.1.1). This information serves as a check to ensure the
experimental system is functioning as expected. Testing service providers should also be aware
of any leaks or instrument malfunctioning that could nullify results of the study.
2.2 Measurement Methodology
This Protocol describes two methodologies for obtaining DRE measurements: Method 1
(Dilution Adjusted Concentration Measurement), which is performed with the plasma off, and
Method 2 (Total Volume Measurement), which measures emissions during actual production
processing conditions. Two approaches may be followed when using Method 2 to measure the
DRE: Sequential Single-Chamber Process Inlet Abatement System Flow Sampling (SSPISF) or
Multi-Chamber Process Inlet Abatement System Flow Sampling (MPISF). Proper application of
these methods (or approaches) will ensure that the DRE measured has a relative error below ±5
percent. The relative error of the estimated true fraction emitted provides one benchmark for
assessing the quality of the sampling methodology, and is the metric used in this Protocol to
determine the acceptability of the DRE measurement methodology. If the relative error of the
true fraction emitted is less than ±5 percent at one standard deviation the methodology used is in
accordance with the best practices defined in this Protocol via Methods 1 and 2.
Whether Method 1 or Method 2 (SSPISF or MPISF) is most applicable depends on several
considerations, including the experimental system to be tested, measurement study objectives,
and safety (cf, Sec. 2.1.5). All named methods and approaches require similar sampling and
analytical methodologies, which are described in detail in this section of the Protocol. Where the
methods and approaches differ is Method 1 requires sampling when the process tool is off while
Method 2 (SSPISF or MPISF) requires sampling under normal process conditions. The
differences in sampling conditions lead to differences in data analysis and DRE calculations for
Methods 1 and 2. Using alternate methodologies to the ones described here is acceptable, as
long as the relative error in the true fraction emitted measurement is less than ±5 percent. If an
alternate methodology is used, the burden falls on the testing service provider to demonstrate that
the relative error requirement (±5 percent) is met, to provide detailed descriptions of the
methodology used, and to provide the appropriate supporting data analysis.
In instances when concentrations exiting the abatement system are below FTIR detection limits a
modification of the performance metric is required. Because the concentrations exiting the
abatement system cannot be measured in these instances, a predefined F-GHG concentration is
used (which exceeds the lower limit of FTIR detection) to demonstrate proper measurement. To
calculate the performance metric, the relative error of that concentration is used, with the result
compared to the performance standard of ±5 percent.
11
-------
Methodologies and approaches described in this Protocol make extensive use of Fourier
Transform Infrared (FTIR) Spectroscopy and Quadrupole Mass Spectrometry (QMS) techniques.
Testing service providers are directed to the guidance prepared by International SEMATECH
Technology Transfer #0612485A-ENG for detailed descriptions for best practice FTIR and QMS
Protocols (December, 2006).7
2.2.1 Method 1—Dilution Adjusted Concentration Measurement
In Method 1, the process tool plasma (RF) is off while data are collected, so DRE measurement
is made with the F-GHG and diluent gases entering the abatement system. An advantage of
Method 1 lies in controlling the flow rates of the F-GHGs measured, resulting in steady-state
conditions, which make the subsequent data analysis and DRE calculations simpler than dealing
with non-steady state flows. However, when the process tool is off, by-product gas DRE's will
not be measured, requiring modification of the process tool setup to imitate by-product gas
flows. Method 2 overcomes this difficulty and is, therefore, preferred.
When using Method 1, F-GHG flows entering the abatement system should be 120 percent of the
F-GHG flows entering the abatement system during normal production processing to ensure the
abatement tool is being tested under conditions that may occur during processing. For example,
if the process flow for an F-GHG, f, is x seems during production processing (which corresponds
to y g/sec) and the corresponding default utilization is Uf, then the flow entering the abatement
system during testing should be 1.2 * y * (1-Uf) g/sec of gas f. In a circumstance when
byproducts are formed, the sum of the imitated emissions of the F-GHG and byproduct gas(es)
should equal 120 percent of the value obtained using the appropriate IPCC 2006 default emission
factors (Volume 3, Table 6.3).
2.2.2 Method 2—Total Volume Measurement
Method 2 involves collecting data during actual process conditions. Method 2 requires no
modifications to the process tool for either process or byproduct gas DRE measurements.
However, collecting data during actual process conditions may not result in steady-state
conditions, making the data analysis and DRE calculations more involved than Method 1 (see
section 2.3 for detailed discussion of data analysis). Sampling to measure total F-GHG volume
flow leaving the process tool and the abatement system may be accomplished through two
approaches, which are applicable to process tools with single or multiple process chambers
(reference Fig. 1).
One approach consists of sequentially measuring the F-GHG volume flow entering the
abatement system from each chamber while also measuring the corresponding F-GHG volume
flow leaving the abatement system. This sampling approach is called the Sequential Single-
Chamber Process Inlet Abatement System Flow (SSPISF) method. The second sampling
approach consists of simultaneously measuring the volume flow from all operating chambers that
7 See Appendix A - Technical Protocols of Air Emissions Characterization of SEMATECH Technology Transfer
#0612485A-ENG (December 2006). This Appendix augments EPA Test Method 320, Measurement of Vapor Phase
Organic and Inorganic Emissions by Extractive Fourier Transform Infrared (FTIR) Spectroscopy, as
it might apply to measuring F-GHG emissions in waste streams produced during electronic device manufacture
12
-------
is entering the abatement system while also measuring the corresponding F-GHG volume flow
exiting the abatement system. This second sampling approach is called the Multi-chamber
Process Inlet Abatement System Flow (MPISF) method. Both approaches employ sampling
during actual process conditions. The SSPISF method uses a slip-stream of the effluent from a
single process chamber as its sample and the MPISF method uses multiple slip-streams of the
effluent (one for each operating chamber with each representing the appropriate proportion of the
total process flow from all operating chambers) that are then mixed to account for actual process
conditions. Sampling to measure the total F-GHG volume flow entering the abatement system
may be accomplished using the MPISF method provided appropriate safety precautions are taken
when mixing the flows of combustible gases.
2.2.3 Equipment Needed
Two FTIRs are optimal for measuring F-GHG concentrations in the effluent streams.8 One FTIR
is used to sample from the process effluent and another to sample from the abatement system
effluent. Additionally, a QMS, inert gas supply, and mass flow controller (MFC) are needed to
measure the dilution that occurs through the abatement system. When measuring DREs in
systems that do not abate CF4 and/or SF6, it is possible to measure dilution using an FTIR instead
of a QMS/MFC/inert gas supply system. When a tracer other than a noble gas is used, such as
CF4 or SF6, the metrologist should provide documentation demonstrating that less than 5 percent
of the tracer gas is destroyed. For purposes of testing using such tracers, the tracer gas is treated
as undestroyed. Two examples of an acceptable demonstration are: (1) the manufacturer of the
POU abatement system provides results, based on the methods of this Protocol, that the DRE of
the chemical tracer is <5 percent under the operating conditions being tested at the fab using the
QMS and the noble gas tracer methodology or (2) in-fab tests showing <5 percent destruction of
the chemical tracer on a POU abatement system (based on the QMS and noble gas tracer
methodology of this Protocol for one abatement system in the fab) may be used on abatement
system in the facility of the same make and model installed and operated in the same manner as
the abatement system tested.
Metal bellows sampling pumps should be installed after each FTIR and/or QMS to collect
effluent samples. Installing adjustable flow rate valves to control the sample flow rates and
capacitance manometers to monitor the sample line pressure are recommended. A sample filter
should be used in the abatement system sampling line to prevent particulate emissions from the
abatement system from damaging the FTIR or QMS. Lastly, when sampling using the MPISF
method, mass flow controllers (MFCs) are needed—one for each operating chamber during
sampling—to create a representatively proportional slip stream composition of the total flow.
Table 1 provides a summary of the equipment required and Figure 1 shows a sample schematic
of the experimental setup.
One FTIR can be used, but it would be necessary to alternate abatement system inlet and outlet measurements.
Steady-state behavior would have to be verified by sampling for a sufficiently long period of time. It must be shown
that sufficient precision can be achieved with just one FTIR system. Using one FTIR will require knowing with
confidence what the abatement system loading is. If sufficient inlet data is acquired to accomplish this, and then the
FTIR is switched to the outlet of the abatement system and the same process is run on the tool then the Protocol will
work just as well as using two FTIRs. The emission factors for a particular process run on the same chamber are
usually reproducible. Sufficient data can be collected to achieve the desired precision.
13
-------
Table 1. Measurement Study Equipment Needs"
Equipment Type
Quantity Required
FTIR
QMS
Inert gas supply/mass flow controller (MFC)
Metal bellows sampling pumps
Adjustable flow rate valves
Capacitance manometers
Sample filter
MFC for exhaust streams
1-2°
0-1
0-1
2-3
2-3
2-3
1-2
No. of chambers tested
a Not exhaustive
b When using 1 FTIR, it must have variable path length
2.2.4 FTIR and QMS Protocols
FTIR measurements should be taken following the best practice guidelines specified in the
guidance prepared by International SEMATECH Technology Transfer #0612485A-ENG
(December, 2006).9 The FTIR should be operated at a resolution of 0.5 cm"1 or 1.0 cm"1,
provided that calibrations match resolution used during sampling. The sampling frequency for
the FTIR should be less than 3 seconds.10 The FTIR absorbance range should be between 0.1 and
1.0; however, lower and higher absorbencies may be used provided sample data are bracketed by
calibration data. Alternate short- and long-path gas cells should be used, to ensure that the
measured absorbance for process effluent concentrations and abatement system effluent
concentrations falls within the range 0.1 to 1.
QMS measurements should be taken following the best practice guidelines specified in the
guidance prepared by International SEMATECH Technology Transfer #0612485A-ENG
(December, 2006).u To account for fluctuations/drift in QMS sensitivity, the ion signals should
be normalized to the signal obtained for the nitrogen fragment N+, which is formed during
ionization of N2.
2.2.5 Calibration Curves
9 See Appendix A.2 - Fourier Transform Infrared (FTIR) Spectroscopy Protocol of SEMATECH Technology
Transfer #0612485A-ENG (December 2006).
10 In the tests EPA performed, 4 FTIR scans were co-added for one data point, which takes 2.2 sees at 0.5 cm"1
resolution and provides adequate temporal resolution to follow time-varying process emissions. Thus, each data
point requires 2.2 sec, so 40 data points requires approximately 1.5 minutes.
11 See Appendix A.I - Mass Spectrometry Protocol of SEMATECH Technology Transfer #0612485A-ENG
(December 2006).
14
-------
The Protocol permits either onsite or offsite calibration for FTIR measurements of F-GHGs, with
both governed by the guidance provided in Appendix A.2 of SEMATECH Technology Transfer
(December 2006). In either circumstance, the range in absorbance for the calibration curves
should bracket the corresponding absorbance range for the onsite measured concentrations, with
a preferable range for the absorbance between 0.1 and 1. In situations where the onsite measured
absorbance lies outside the 0.1 to 1 range, either (a) a calibration must be available at the high or
low absorbance, (b) an alternative absorption band should be used that lies, for that specific
molecule, within the 0.1-1 absorbance, or (c) an alternate FTIR gas cell should be used.12
Further, while the measured absorbance of large molecules, like the F-GHGs of concern here, is
not expected to be sensitive to reasonable differences in resolution (say, 0.5 vs. 1
cm"1), this concern is mitigated by requiring the use of the same resolution for calibration
sampling.
For onsite calibration, the metrologist must report/show the calibration graphs and report the
corresponding calibration and measurement conditions for all F-GHGs measured. The
measurement conditions to report are gas temperature and pressure, as well as the FTIR
resolution used during sampling and calibration. When reported DREs employ offsite
calibration, the metrologist must report/show calibration graphs as well as the conditions
(temperature, pressure and resolution) used for calibration and during onsite sampling. If
calibration and onsite temperatures, pressures or both differ then corrections must be made to
account for the change in gas number density. These corrections are often standard features of
modern FTIR software packages. The calibration curves for both onsite and offsite testing should
be generated using at least three calibration points such that expected measured absorbance range
is covered. Quality of calibration may be judged by proximity of R2 to 1.00. For non-linear
calibration curve, it's important that highest observed concentration is within not outside the
range of concentrations used to define the calibration curve.
Prior to conducting measurements of the process and abatement system effluents, onsite QMS
calibration should be completed for all compounds to be measured. The calibration curves
should be generated using at least 1 zero and 5 non-zero data points. The recommended ion
intensity range for the QMS should be within one decade. Calibration plots with the signal
intensity versus compound concentration data points and the line of best fit should be included in
the measurement report. Additionally, the slope, the relative error of the slope (one standard
deviation), the y-intercept, and the correlation coefficient (i.e., R2) of the correlation line should
be presented and each parameter clearly labeled (see Figure 2 as an example).
To ensure that the ±5 percent benchmark relative error is achieved, all calibration curves should
meet the following criteria:
• Slope must have an error below ±5 percent; and
• Correlation coefficient must exceed 0.98
12 Some FTIR quantitative analysis software packages (e.g. AutoQuant Pro used by Midac Corporation) has the
ability to shift to different absorbance peaks over several orders of magnitude depending on the concentration of the
material being measured. In this case, multiple gas cells may not be required. However, that shifting to another peak
may risk influence of interferent that also absorbs at that same wave number. Using multi-path cells is best practice
for this FTIR application.
15
-------
QMS Response to Kr and Xe
1DO&D3
9DOE-04
8DOE-04
7DD&Q4
'J5 6DOE-Q4
OJ
S 5DOE-04
1 4DOE-Q4
3DDE-D4
200E-04
1DO&04
ODQE+00 -I-
0.0
Ki-
20.0
40D eon 800
Concentration of Added Kr & He (ppmv)
100.0
120 fl
Figure 2. Sample Calibration Curve The R2 and relative error of the slope for both calibration
curves shown are >±99 percent and <±1 percent, respectively. Source: EPA, 2008a and 2008b.
2.2.6 Flow and Dilution Measurements
Determining dilution across the abatement system is a two step process. The first step is
determining the total volume flow of the process effluent, and the second step is determining the
total volume flow of the abatement system effluent. The total volume flow of the process effluent
can be estimated by flowing known quantities (flows) of F-GHG through the tool with the RF
turned off, while a FTIR system measures the F-GHG concentration of the process effluent. The
total volume flow of the abatement system effluent can be estimated by supplying known
quantities of inert gas into the process stream effluent, while a QMS system measures the inert
gas concentration in the abatement system effluent. Controlled volume flows of inert gas can be
supplied to the process effluent using a mass flow controller (MFC). When measuring DREs in
systems that do not abate CF4 and/or SF6 it is possible to measure total abatement system flow
using an FTIR instead of a QMS. In such systems CF4 or SF6 can be used in place of an inert gas
since their DREs are zero percent. Table 2 provides a list of acceptable gases for measuring total
abatement system flows, along with their use conditions.
16
-------
Table 2. Acceptable Gases for Monitoring Total Abatement System Flows
Tracer Gas
Kr
Xe
Ar
CF4
SF6
Analytical Equipment
QMS
QMS
QMS
FTIR
FTIR
Limitations
None
May be used in some etch
processes recipes.
Must subtract out air
concentration background;
cannot use if Argon is used
as a process gas.
DRE must be zero
DRE must be zero
To ensure that the ±5 percent relative error is met, the following conditions should be met:
• Accuracy of calibration gases in the concentration of the inert gas supplied to the
process effluent stream should be ±3 percent or less.
• 3-5 different flow rates should be supplied to the process tool and/or abatement
system, and at least 40 distinct measurements made for estimating the average
concentration at each flow rate.
• Appropriately sized MFC such that accuracy of lowest flow is ±3 percent or better
2.2.7 F-GHG Measurements
The Protocol suggests that using two FTIRs is optimal.13 One FTIR is used to measure F-GHG
concentration in the process effluent and another to measure F-GHG concentration in the
abatement system effluent. The concentration of F-GHGs in both effluent streams should be
recorded, as well as the gases/flow rates and/or process recipes that are fed into the process tool.
Absorption bands should be chosen to minimize the influence of interfering substances and to
hold absorbance within the range 0.1 to 1.
To ensure that the ±5 percent relative error is achieved, the following conditions should be met:
• Method 1: For each process gas tested, 3-5 different flow rates should be supplied to
the process tool, and at least 40 process tools and abatement system effluent
concentration measurements averaged for each flow rate.
• Method 2 (SSPISF and MPSIF): For each process gas tested, 3-5 data points should
be collected for total volume in and total volume out of the abatement system and at
least 40 process tools and abatement system effluent concentration measurements
averaged for each flow rate.
When the concentration exiting the abatement system approaches or falls below the minimum
detectable level the metrologist can demonstrate proper measurement practice by introducing
into the sampling stream a flow of that gas such that its concentration entering the abatement is
low but exceeds the minimum detection level. This concentration is denoted as C*. To
13
See footnote 8.
17
-------
demonstrate proper measurement of C , the relative error of the C measurement is used in place
of C0 (the exit concentration) to calculate the performance metric. The precision of C is
estimated with data from the FTIR used to measure the F-GHG exiting the abatement system.
Table 3 sets forth C*, in units of ppm-m, for the gases governed by this Protocol. Table 3
identifies the minimum detection (concentration) levels for the gases of interest at standard
temperature and pressure (STP), defined as a signal to noise of 3 for a HgCdTe liquid nitrogen
cooled detection, 0.5 cm"1 resolution. A single value for C* was chosen for simplicity. It is
evident from comparison of the second and third column that, for the selected C*, the signal to
noise ratio (S/N) will be much greater than 3 so the precision should be high. Further, for the
typical long-path used in such tests, the C concentration being sampled should correspond to
DREs in the range 95-99%.
Table 3. Minimum detection levels (CMDL) and threshold F-GHG concentrations (C*) for measuring
relative error of total fraction emitted when sample C < C', ppm-m
F-GHG
CF4
CHF3
C2F6
C3F8
c-C4F8
NF3
SF6
Typical CMDL, ppm-m
0.12
0.40
0.21
0.2
0.7
1.1
0.1
C*, ppm-m
10
10
10
10
10
10
10
Primary absorbtion
band, cm"1
1280
1150
1250
1150
965
910
943
Note: CMDL values based on calibration data using path lengths between 10 cm and 5 m,
resolution of 0.5 cm"1, gases at STP, HgCdTe liquid cooled detector and a S/N of 3 10"3
absorbance units.
Of the several ways for preparing a sample stream with concentration C , two that are acceptable
are described here. The first is to ask the process tool operator to provide a sufficiently low flow
of gas such that after dilution by the pump purge, C* is achieved at the sampling point ahead of
the abatement system. For example, if the target concentration, C , is 2 ppm and the pump purge
is 50 1pm (determined previously), then 50 ppm is achieved with a process flow of C of 0.1
seem. For process tools with MFCs that cannot deliver sufficiently low flows, the sampled slip
stream can be diluted with nitrogen using either a separate MFC or needle valve. The actual
concentration sampled may be within ±20 percent of C , i.e., ±0.4 ppm in the example (vide
supra).
A second approach for preparing C* is for the metrologist to have available a gas cylinder with
the appropriate concentration to introduce ahead of the abatement system, which, accounting for
the pump purge flow, will produce C ±20 percent of the value taken from Table 3. The first
approach may be preferable because it doesn't require the availability of additional gas cylinders
of the appropriate concentrations.
2.3 Data Treatment and Analysis
18
-------
The procedures for treating flow and concentration measurements as well as calculating total
flows, dilution factor and DRE of the abatement system are described in this section. Also
described are the methods for consistently estimating the precision associated with measuring
gas concentrations, flows, dilution factors and DREs. The data treatment and analysis
component consists of two sections. The first section addresses data used to calculate best-
estimate values for flows and the abatement system dilution factor as well as their standard
deviations. The second section addresses the data used to calculate best-estimate of the
abatement system DRE and associated relative errors, using either Method 1 or Method 2. In the
following sections absolute error is referred to as a and relative error is referred to ass , and all
relative and absolute errors are presented at 1 standard deviation.
When an alternative methodology, such as the Physical Flow Measurement method, to the two
described in this Protocol is used to estimate DRE, the data treatment and analysis procedure
described below may not be applicable. In such cases, the testing service provider should
determine and use the appropriate procedures to calculate both the DRE and its corresponding
error.
2.3.1 Abatement System Dilution
2.3.1.1 Total Volume Flow
Total volume flow (TVF) is the total volume of gas that flows across a process tool or abatement
system, and is generally measured in standard liters per minute (slm). For both Methods 1 and 2,
measurements of TVF entering and exiting the abatement system are needed. TVF or F is
calculated using Equation 1:
sf
F= f—— (1)
Ce/xl(T6
where,
F = total volume flow based on a single concentration data point [slm]
Sf = spike gas flow into the process tool/abatement system [slm]
Cef = measured concentration of standard gas in the process tool/abatement system effluent
[ppmv]
To ensure that the error of TVF is low enough to meet the relative error requirement of ±5
percent, TVF measurements should be collected during at least three different standard gas flow
rates. During each gas flow rate, at least 40 concentration measurements (FTIR or QMS scans)
should be collected at the effluent side of the process tool/abatement system. For example, when
measuring the TVF exiting the process tool, a process gas such as C$6 should be fed into the
process tool at multiple flow rates (e.g., 0.2, 0.6, and 1.0 slm), and during each flow rate, a
minimum of 40 concentration data points of the process effluent should be collected using a
FTIR. Similarly, when measuring the TVF exiting the abatement system, an inert gas such as Kr
should be spiked into the abatement system at multiple flow rates (e.g., 0.010, 0.030, and 0.050
19
-------
slm), and during each flow rate a minimum of 60 concentration data points of the inert gas in the
abatement effluent should be measured using a QMS.
The measurements above will result in at least 120 TVF data points, which will need to be
averaged appropriately. A simple average, denoted as F determines the average TVF and
corresponding standard deviation during each flow rate applied, denoted by subscript m (see
Equation 2 and Equation 3).
= .
where,
Fm = simple average of TVF [slm] for the mth flow
n = number of concentration measurements taken while at a constant standard gas flow (e.g., 40
concentration data points collected during a 0.2 slm standard gas flow rate)
a- =]-Y(F1-FmJ (3)
Fm V n^—l\ ' ' ^ '
The TVF simple average (Fm) only applies to values that are collected during the same standard
gas flow rate. Therefore, if three different standard gas flow rates are collected, three different
simple averages must be derived.
When combining the multiple TVF simple average (Fm) values for each flow rate into a single
/\
average TVF, a variance weighted average must be applied (denoted as F ). The equations for
determining the variance weighted average and its corresponding standard deviation, are shown
in Equations 4 and 5.
M
V
'
where,
Fg = variance weighted average TVF [slm] for flow standard gas, g
M = number of flow rates used per standard gas, g
(5)
Fm J
20
-------
where,
n = number of concentration measurements
The variance weighted TVF average (F ) can only be calculated using average TVF values (F )
obtained using the same standard gas. For example, if two different standard gases are used,
then a variance weighted average is calculated for each standard gas.
Equation 5 assumes that the number of concentration measurements during each flow rate is
constant (i.e., n is the same for each simple average). If n varies for each simple average, then
Equation 5 must be altered to account for variations in n. Alternatively, when n varies, the
lowest n can be used in Equation 5 to generate a conservative estimate of the standard deviation
(i.e., the standard deviation will err on the larger side).
In the case when multiple standard gases (e.g., Kr and Xe for abatement system or C$6 and CF4
for process tool) are used to calculate TVF, then the variance weighted average (Fg ) for each
gas, g, can be combined to determine a TVF best estimate (denoted as F ) and its corresponding
error as shown below (see Equations 6 and 7).
•
where,
F = Best estimate for TVF [slm]
G = number of standard gases used to determine TVF
f G
a- =
F
£—
(7)
The absolute errors or standard deviations obtained from equations 3, 5 and 7 measure the
precision of each standard gas flow for each gas, for all flows for each gas, and for all gases,
respectively, when more than one gas is used (e.g., if Xe and Kr were both used). The overall
accuracy is determined by the accuracy of the calibration, calibration standard gas
concentrations, and mass flow controller. In order to obtain a DRE relative error of less than ±5
percent, the concentration standard gas should have a relative error of less than 3 percent of the
designated concentration, and the accuracy of the MFC should be less than 3 percent for the
lowest flow rate used during the measurements.
2.3.1.2 Dilution Factor
21
-------
The dilution factor describes the total volumetric dilution that occurs across an abatement
system. The dilution factor is determined by calculating the best estimate for TVF into (Fin)
and out of the abatement system (Fout). (See Section 2.3.1 for determining the best estimate for
TVF). The TVF exiting the abatement system divided by the TVF entering the abatement
system defines the dilution factor as shown in Equation 8. The standard deviation or precision of
the dilution factor is determined by the standard deviation of each TVF measurement, and is
estimated using error propagation techniques (See Equation 9).
Fm
where,
DF = Best estimate for dilution factor [dimensionless]
Fout = Best estimate for TVF exiting abatement system [slm]
Fin = Best estimate for TVF entering the abatement system [slm]
(8)
e(DF) =
where,
77
V out J
-------
During Method 1, the concentration of process gas entering and exiting the abatement system is
monitored, while the process tool is off, and while a known flow rate of process gas is fed into
the abatement system. For each flow rate, the concentration of process gas in the abatement
influent and effluent is in steady state. For each process gas at least three different process gas
flow rates should be used to collect a minimum of 60 concentration data points per flow rate, to
ensure relative error is achieved. As described in section 2.2, an FTIR should be used to monitor
concentration of process gases, and the FTIRs absorbance signal should be converted to a
concentration using a calibration curve. The relative error of each concentration data points
measured by the FTIR is equal to relative error of the calibration curve slope.14 Using the
concentration data points, the variance weighted average and corresponding error for
concentration can be estimated using a procedure analogous to that shown for total volume flow
(see Section 2.3.1.1; Equations 1 to 5). The variance weighted average is considered the best
estimate for F-GHG concentration entering and exiting the abatement system.
2.3.2.1.2 DRE
np/7, -
\L ^ x Db }
^ C- in
where,
x 100 percent
(10)
DREi = Best estimate for Destruction and Removal Efficiency [percent]
Cout = Best estimate for concentration of F-GHG exiting the abatement system [ppm]
Cin = Best estimate for concentration of F-GHG entering the abatement system [ppm]
s(DREl)= =
where,
- ^ 01
Ar =-=-
Cin
2 ^= ^2
Cm
(11)
\ v_- out j \^ in /
a, = The standard deviation of Ac
a— = The standard deviation of DF
DP
14 The relative error of the calibration curve slope should not exceed ±5 percent (at one standard deviation) to ensure
the Protocol benchmark error of ±5 percent is obtained.
23
-------
a= = The standard deviation of Cout [pPm]
C0ut
a- = The standard deviation of Cm [ppm]
Cm
2.3.2.2 Method 2—Total Volume Measurements
2.3.2.2.1 Total Volume Measurements
When measuring the DRE using Method 2, the concentration of the gas in the abatement influent
and effluent is monitored while the process tool is on, and cannot be assumed to be at steady
state. Therefore, the total volume of process gas entering and exiting the abatement system must
be estimated by integrating the FTIR signal over time (The absorbance signal should be
converted to concentration using the calibration curve, prior to integration). When multiple total
volume measurements are collected for a single process, the best estimate and error of the total
volume measurement is equal to the simple average and its standard deviation, which are
analogous to Equations 2 and 3 in Section 2.3.1.1.
2.3.2.2.2 DRE
Vout
DRE 2 = | 1 - -==- \ x 100percent (12)
Vin
^.
where,
Vout = Best estimate for total volume of F-GHG exiting the abatement system [si]
Vin = Best estimate for total volume of F-GHG entering the abatement system [si].
Vi = \Vi(t} = ^FiCi]M]=Fi^Ci]M] (13)
where /' denotes either the outlet or inlet F-GHG gas, F denotes the corresponding inlet or outlet
flow, Ar denotes the integration interval, C.yis the concentration of F-GHG either entering or
exiting the abatement system and the sum is taken over the period, T, of production processing.
(14)
where,
^=-fe
Vm
Vout
\^V out J
a= = The standard deviation Vout [slm]
V out
24
-------
a- = The standard deviation of Vin [slm].
Vin
It can be shown thate^, the relative error in the true fraction of emitted F-GHG estimated using
Method 2 is, to a very good approximation, governed by relative error in the effluent process and
abatement system flows, or the relative error in the dilution factor. This simplification occurs
because, in calculating the volume flow, integrating the measured concentrations over the period
of production processing averages out the uncertainties in FTIR measured concentrations
compared to the corresponding measured flow rates.15 The relative error for Method 2 may be
estimated, to a very good approximation, as the relative error in the dilution factor, which is
given by Eq. 9.
3. BENCHMARK RELATIVE ERROR
For Protocol to be considered valid, the relative error for true fraction emitted must be less than
±5 percent. Presented below are the formulas for estimating relative error (i.e., s(TFE)) of the
true fraction emitted for both Method 1 (Equation 14) and Method 2 (Equation 15). All the
parameters used to estimate the relative error of true fraction emitted have been defined in
section 2.3, and therefore, are not redefined here. These formulas may not be applicable when
alternative methodologies to those presented in this Protocol are used, and in such cases, testing
service providers should determine and use the appropriate formulas to calculate the relative
error of the true fraction emitted.
3.1 Method 1—Dilution Adjusted Concentration Measurements
s(TFEl) = II -^- I + -=11- < ^percent (15)
(DF)
3.2 Method 2—Total Volume Measurements
a-
-^- +-^- < ^percent (16)
Vout j [^ Vin
which may be estimated using Eq. 9.
In those instances where the concentrations exiting the abatement system, Cout in the equations is
replaced with C*, so Ac is calculated using C*.
15 This can be shown using the expression for ^(/l^ ) given in Eq. 14 and noting that relative errors, Sy and Sy ,
are each the sums in quadrature of the relative errors of the outlet/inlet flows and outlet/inlet concentrations,
respectively. Using the method of simulation to estimate the relative error in the outlet or inlet concentrations, it
becomes evident that the relative error in measured outlet or inlet concentration is more than an order of magnitude
less than the relative error in the corresponding measured abatement system outlet or inlet flows. The reduced
relative error in the integrated concentration profile occurs because in the course of integrating over the measured
FTIR concentration profile the error in each FTIR measurement is reduced through averaging.
25
-------
4. DOCUMENTATION AND REPORTING
Proper documentation and reporting ensures transparency of the DRE measurement process,
allowing users of DREs and third-parties to review and understand the measurements collected
as well as the data treatment and analysis conducted. Therefore, it is important that testing
service providers adequately document their analyses, and present the results in a clear and
concise report. A suggested structure for the report is presented in the text box below.
The Protocol also requires certification by the testing provider. The final report should contain
the following language together with the appropriate signature.
The tester/metrologist principally responsible for the measurements and content presented in this
report and obtained during the period to at (name of facility) hereby certifies that the
methods and calculations set forth in the Protocol were followed and that material departures or
deviations therefrom, if any, are fully documented in this report dated, . The
tester/metrologist also certifies that the calculations and benchmark relative error presented in
this report properly and accurately reflect the actual measurements made at the facility during the
test period noted in this certification.
Signed
Date
26
-------
Final Report Structure
I. Introduction
a. General information about testing
II. Experimental Setup
a. Actual sampling/testing configuration
b. Description of how measurements are taken (reference to Protocol
encouraged)
c. Sampling Configuration
III. Data Analysis
a. Documentation
i. Formulas Used
ii. Figures showing calibration curves for each F-GHG and tracer
gas (the regression equation should be displayed for each.
calibration curve along with relative error of the slope and y-
intercept).
iii. Figures showing the FTIR spectra at the abatement inlet and
outlet, which each F-GHG peak clearly labeled.
iv. Figures showing the temporal concentration profiles of the
abatement inlet and outlet concentration for each F-GHG and
tracer gas.
v. Tables showing the means and standard deviations of the F-
GHG and tracer gas inlet and outlet concentrations and/or
volumes (see Section 2.3 for description of the means and
standard deviations appropriate for each quantity).
vi. Tables showing the mean and standard deviation of the dilution
factor and ORE for each F-GHG.
b. Comparison of estimated DRE error with the 5 percent benchmark
relative error.
IV. Discussion
V. Metrologist certification that reported results conform to Protocol.
Appendix A: Measurement Plan
27
-------
5. REFERENCES
40 CFR 60. 1998. Appendix A, Method 2, as amended. U.S. Environmental Protection Agency, Method 2 -
Determination of Stack Gas Velocity and Volumetric Flow Rate. U.S. Code of Federal Regulations.
40 CFR 52. 1998. Appendix E, Performance Specifications and Specification Test Procedures for Monitoring
Systems for Effluent Stream Gas Volumetric Flowrate. U.S. Code of Federal Regulations.
ASTM. (2004) Standard Test Method for Volumetric and Mass Flow Rate Measurement in a Duct Using
Tracer Gas Dilution. Designation E 2029 - 99 (Reapproved 2004).
Beu. Laurie S. International Sematech Manufacturing Initiative (ISMI 2005). Reduction of
Perfluorocompound (PFCF-GHG) Emissions: 2005 State-of-the-Technology Report. International Sematech
Document ID 05104693 A-ENG, December 1,2005.
Beu, L. et al.(1994)., Results of Delatech Controlled Decomposition/Oxidation (CDO) Unit Testing for
Perfluorocompound (PFC) Emission Abatement Applications, SEMATECH Technology Transfer
#94092543A, October 1994.
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.
http://www.epa.gov/semiconductor-pfc/documents/qimonda 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.
http://www.epa.gov/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 43 0-R-10-004), Office of Air and Radiation Office of Atmospheric Programs, Climate Change
Division, U.S. Environmental Protection Agency, Washington, DC. http://www.epa.gov/semiconductor-
pfc/documents/ibm_report.pdf
International Sematech (2006). Guideline for Characterization of Semiconductor Process equipment.
International Sematech, Technology Transfer # 06124825A-ENG, December 22, 2006. Note that this is an
update to previous guideline, TT from International Sematech # 01104197A-XFR, December 2001.
IPCC Guidelines for National Greenhouse Gas Inventories. 2006. Volume 3 Industrial Processes
and Product Use. Chapter 6, Electronics Industry Emissions.
Laush, Curtis, Mike Sherer and Walter Worth. International Sematech Manufacturing Initiative (ISMI 2006).
Guideline for Environmental Characterization of Semiconductor Process Equipment. International Sematech
Document ID 06124825A-ENG, December 20, 2006.
Lee etal. (2007). Evaluation Method on Destruction and Removal Efficiency of Perfluorocompounds from
Semiconductor and Display Manufacture. Bull. Korean Chem. Soc. 28 (8), 1383-1388 (2007).
28
-------
Li et al., (2001). Improved Technique for Evaluating Point-of-use Abatement Systems. Semiconductor
Fabtech, 14th Edition, June 2001.
Li et al., (2002). FTIR spectrometers measure scrubber abatement Efficiencies. Solid-State Technology, July
2002.
Li et al.,(2004). Default values appear to be overestimating F-GHG emissions from fabs. Solid-State
Technology, Sep. 2004.
Stuart L. Meyers (1974). Data Analysis for Scientists and Engineers. John Wiley & Sons, Inc., New York
1974. Page 146-147.
29
-------
Appendix A - History of and Revisions to the Protocol
The Protocol has gone through five stages of development: conceptualization, onsite tests, initial
drafting, and two informal peer review processes. During the informal peer review processes
EPA received comments from both national and international parties, including semiconductor
manufacturers, equipment manufacturers, gas suppliers and analytic service providers. EPA
addressed these comments directly with reviewers in verbal and written form.
Modifications were made to the Protocol to address this feedback. In addition to specific
technical issues that reviewers raised, an overarching concern emerged that the Protocol provide
as much flexibility as possible without compromising the integrity of the Protocol's
measurement process. Calls for flexibility stemmed from the diversity of fab environments, as
well as from the diversity of established measurement practices and technologies. International
reviewers were unanimous in seeking acceptable alternatives to QMS systems (used for
measuring abatement system dilution via the gas tracer method of chemical spiking) for onsite
testing, although central to the method favorably evaluated by Lee et al. (Lee, 2007) is the use of
a magnetic type precision gas mass spectrometer rather than a quadrupole mass spectrometer
(QMS). To provide for more flexibility and to address the technical issues that reviewers raised,
EPA considered and made substantive changes to the Protocol:
• Removal of suggested temperature limitations at which a POU system would not abate
CF4 or SFe in reference to instances where CF4 or SFe may be used as tracers gases to
measure flow across the abatement system
• Allowance of off-site Fourier Transform Infrared (FTIR) Spectrometer calibration
• Consideration of alternative DRE measurement methods. The alternative methods
considered were:
o The Multi-chamber Process Inlet Flow Sampling (MPIFS) Method
o The Physical Flow Measurement Method16
o The Abatement system De-tuning Method17
o The Post-abatement Flow Tracer Method
Of these listed methods, the Protocol expands Method 2 to include the MPIFS method. The
MPIFS method was tested in December 2008. The results using the MPIFS method agreed with
results obtained using the Sequential Single-Chamber Process Inlet Flow Sampling (SSPIFS).18.
The Physical Flow Measurement Study, because it is already an EPA method, is allowable under
16 Physical Flow Measurement Methods are allowable per the Protocol to measure the Total Volume Flow in order
to determine DRE. Acceptable flow measurement methods would be those that have pre-determined specified
measurement standards in place, such as EPA Appendix A Method 2 and EPA Appendix E. However, these
methods are not preferred because often they require laminar flow regimes, where that is often not the case in in-fab
testing. When using a physical method for measuring gas flow rate, the performance benchmark metric and standard
given in Section 3 applies.
17 The Abatement system De-tuning Method involves either turning off the methane or oxygen flow to the burner of
the abatement system, lowering the temperature of the abatement system, or a combination of the two. The Protocol
does not permit this method because to do so would defeat the purpose of the abatement system and intentionally
release F-GHG emissions to the atmosphere during testing.
18 EPA cautions, however, that the MPIFS method holds potential safety risks when mixing pyrophoric/flammable
and oxidizing gases.
30
-------
this Protocol; and the Abatement system De-Tuning Method in not permitted under this Protocol.
The Post-abatement Flow Tracer Method, while theoretically attractive, has not been compared
to the pre-abatement tracer method used in the Protocol and is therefore not permitted under this
Protocol.19
19 While the Post-abatement Flow Tracer Method is not permitted in the Protocol, EPA is interested in learning
more, through in-fab testing, about the validity of this method.
-------
Appendix B - Sample Calculation
Step 1 Determine the Mean Total Flow
Data
n =
Kr
Flow
(slm)
0.01
0.02
0.03
0.04
0.05
60
AvgKr
Cone
(ppm)
14.1
25.5
38.8
52.2
65.8
Step 2 Determine the Dilution Factor
Data
Out
In
TF
(slm)
765
15.5
OIF
(slm)
2
0.1
Calculations: Dilution Factor
OTF/TF
0.0026144
0.0064516
DF
Sum of (aTF/TF)2
Relative Error
(o-TF/TF)2
0.0000068
0.0000416
= 49.35483871
0.0000485
0.7%
Units
NA
NA
dimensionless
dimensionless
dimensionless
Calculations: Data Treatment
Measured effluent flows
Standard deviation in
Avg Effluent measured effluent
Flow, F
(slm)
709.2
784.3
773.2
766.3
759.9
flow, a
(slm)
80
52
34
28
28
Variance weighted mean of effluent flow
Variance in
measured effluent
flow, a2
(slm2)
6400
2704
1156
784
784
Sum of F/a2
Sum of 1/a2
Mean total flow
(Sum of I/a2)'172
Sqrt(1/n)
a of effluent flow
95% Cl
Upper bound of mean
Lower bound of mean
Relative error, %
1/a2
0.0001563
0.0003698
0.0008651
0.0012755
0.0012755
flow
flow
F/a2
0.110816
0.290057
0.668855
0.977402
0.969233
3.01636
= 0.0039421
765
15.927
= 0.129099
2
4
769
761
0.5%
Units
NA
NA
NA
NA
NA
slm'1
slm'2
slm
slm'1
dimensionless
slm
slm
slm
slm
dimensionless
32
-------
Step 3 For Method 1: Determine the Relative Error of the TFE
*The TFE error must meet the +/- 5% performance standard
Data
C0
c.
DF
Value
200
12,134
49.4
a
2
195
2
Calculations: A
a/Best Estimate
0.0100000
0.0160705
0.0405229
Ac
O-A
A*DF
1 - (A * DF)
Sum of (aCo/C0)2 and (aa/C|)2
OA/AC
(aA/Ac)2
Sum of (aA/Ac)2 and (aDF/DF)2
Relative Error A
ORE, %
Relative Error ORE
Relative Error of TFE
ORE, and True Fraction Emitted
(a/Best Estimate)2
0.0001
0.000258262
0.001642103
0.0167450
0.0001033
0.8264455
0.1735545
0.0003583
0.0061710
0.0000381
0.0016802
1 .9%
1 9%
20%
4%
Units
NA
NA
NA
dimensionless
dimensionless
dimensionless
dimensionless
dimensionless
dimensionless
dimensionless
dimensionless
dimensionless
dimensionless
dimensionless
dimensionless
33
-------
Step 3 For Method 2: Determine the Relative Error of the TFE
*The TFE error must meet the +/- 5% performance standard
Data
Out
In
Best Est.
V
(slm)
0.425
0.443
ov
(slm)
0.005
0.006
Calculations: A, ORE, and True Fraction Emitted
av/V
0.0110448
0.0127318
Av
O-A
1 -Ay
Sum of (av/V)2
Relative Error A
ORE, %
Relative Error ORE
Relative Error TFE
(av/V)2
0.00012198847956824
0.0001620981711280
0.958843303
0.016194732
0.041156697
0.0002841
1 .7%
4%
2%
2%
Units
NA
NA
dimensionless
dimensionless
dimensionless
dimensionless
dimensionless
dimensionless
dimensionless
dimensionless
34
------- |