United States                 Office of Water (4303)               EPA-821-B-98-003
Environmental Protection            Washington, DC 20460                  March 1999
Agency



 Protocol for EPA Approval of New Methods for

 Organic  and Inorganic Analytes in Wastewater

 and Drinking Water

-------
Protocol for Approval of New Methods for
    Inorganic and Organic Analytes in
     Wastewater and Drinking Water
                March 22,1999
         U.S. Environmental Protection Agency
                Office of Water
           Engineering and Analysis Division
               401 M Street, SW
              Washington, DC 20490

-------
                                         Foreword

       Within the U.S. Environmental Protection Agency (EPA), the Office of Water (OW) publishes test
procedures (analytical methods) for analysis of wastewater and drinking water. Listed at parts 136 and
141 of Title 40 of the Code of Federal Regulations (CFR), these methods are authorized for use in data
gathering and environmental monitoring under the Clean Water Act (CWA) and the Safe Drinking Water
Act (SDWA). These methods have been developed by EPA, by consensus standards organizations, and by
others. Many of these methods, especially methods published before  1990, are prescriptive with limited
ability to modify procedures or change technologies to accommodate  specific situations. There has been a
growing awareness within EPA and the analytical community that the requirement to use prescriptive
measurement methods and technologies to comply with Agency regulations has unintentionally imposed a
significant regulatory burden and created a barrier to the use of innovative environmental monitoring
technology.

       This document gives specific instructions to external organizations regarding the validation,
submission, and EPA approval of applications for the approval of new methods to determine inorganic and
organic analytes.  EPA anticipates that the standardized procedures described herein should expedite the
approval of new methods, encourage the development of innovative technologies, and enhance the overall
utility of the EPA-approved methods for compliance monitoring under National Pollution Discharge
Elimination System (NPDES) permits and national primary drinking water regulations (NPDWRs).

       This document is not a legal instrument and does not establish or affect legal obligations under
Federal regulations. EPA reserves the right to change this protocol without prior notice.

       All questions regarding the guidelines presented in this document should be directed to:
               William Telliard
               Director, Analytical Methods Staff
               Engineering and Analysis Division (4303)
               U.S. EPA Office of Water
               401 M Street, S.W.
               Washington, DC 20460.

-------
1.0    INTRODUCTION	 . 1
       1.1    Background and Objectives	1
              1.1.1 Clean Water Act and Safe Drinking Water Act	1
              1.1.2 40 CFR Parts 136.4,  136.5 and 141.27	1
              1.1.3 Description of Document 	2
       1.2    Tiered System for Validation of New Methods  	2

2.0    APPLICATION REQUIREMENTS	4
       2.1    Submission Addresses	4
       2.2    Application Information	5
              2.2.1 Justification for New Method	5
              2.2.2 Standard EPA Method Format	6
              2.2.3 Validation Study Report  	6
              2.2.4 Method Information and Documentation to Facilitate EPA Preparation of Preamble
                   and Docket   	6
       2.3    Proprietary Information in Applications	7

3.0    METHOD VALIDATION	8
       3.1    Introduction	8
       3.2    Summary of Validation Requirements	8
       3.3    Tier 1, 2, and 3 Validation Studies	10
              3.3.1 Tier 1 Validation Studies for Wastewater and Drinking Water	10
              3.3.2 Tier 2 Validation Studies for Wastewater and Drinking Water	12
              3.3.3 Tier 3 Validation Studies (for Wastewater Only)  	13
       3.4    Development of a Validation Study Plan	14
              3.4.1 Background  	14
              3.4.2 Objectives	14
              3.4.3 Study Management	13
              3.4.4 Technical Approach	14
              3.4.5 Data Reporting and Evaluation	15
              3.4.6 Limitations	15
       3.5    Detailed Procedures for Conducting Validation Studies  	15
              3.5.1 Method Compilation	15
              3.5.2 Method Detection Limit Study	15
              3.5.3 Calibration  	15
              3.5.4 Initial Precision and Recovery  	16
              3.5.5 Field Sample Analyses	16
              3.5.6 Ongoing Precision  and Recovery 	17
              3.5.7 Calibration Verification 	17
              3.5.8 Contamination Level in Blanks	17
              3.5.9 Surrogate or Labeled Compound Recovery	18
              3.5.10  Absolute and Relative Retention Time	18
              3.5.11  Further Validation Studies  	18
       3.6    Validation Study Report	18
              3.6.1 Background  	19
              3.6.2 Study Design and Objectives  	19

                                              iii

-------
              3.6.3 Study Implementation	19
              3.6.4 Data Reporting and Validation	20
              3.6.5 Results	20
              3.6.6 Development of QC Acceptance Criteria	20
              3.6.7 Data Analysis/Discussion	20
              3.6.8 Conclusions	20
              3.6.9 Appendix A - The Method	20
              3.6.10 Appendix B - Validation Study Plan	20
              3.6.11 Appendix C - Supporting Data	 .  20

4.0    EPA REVIEW AND APPROVAL	22
       4.1     EPA Review of Applications	22
       4.2     Approval Recommendation	23
       4.3     Rulemaking Process	23

5.0    REFERENCES  	24

6.0    APPENDIX A - APPLICATION FORM	25

7.0    APPENDIX B - EPA HEADQUARTERS AND REGIONAL CONTACTS	26

8.0    APPENDIX C - STANDARD EPA METHOD FORMAT  	27

9.0    APPENDIX D - QUALITY CONTROL REQUIREMENTS	30
       9.1     Introduction	30
       9.2     Standardized Quality Control Tests	30
              9.2.1 Calibration Linearity	30
              9.2.2 Calibration Verification  	33
              9.2.3 Absolute and Relative Retention Time Precision 	34
              9.2.4 Initial Precision and Recovery  	34
              9.2.5 Ongoing Precision and Recovery  	35
              9.2.6 Analysis of Blanks  	35
              9.2.7 Surrogate or Labeled Compound Recovery	35
              9.2.8 Matrix Spike and Matrix Spike Duplicate	35
              9.2.9 Demonstration of Method Detection Limit	36
              9.2.10 Reference  Sample Analysis  	36
       9.3     Development of Quality Control Acceptance Criteria	36
              9.3.1  Quality Control Acceptance Criteria Development for New Methods at Tier 1  37
              9.3.2  Quality Control Acceptance Criteria Development for New Methods at Tier 2  42
              9.3.3 Quality Control Acceptance Criteria Development for New Methods at Tier 3
                    	48
                                            IV

-------
1.0     INTRODUCTION

1.1     Background and Objectives

1.1.1   Clean Water Act and Safe Drinking Water Act

       CWA section 304(h) requires the EPA Administrator to promulgate guidelines establishing test
procedures for data gathering and monitoring compliance with published guidelines. EPA's approval of
analytical methods is authorized under this section of CWA, as well as the general rulemaking authority in
CWA section 501 (a).  The section 304(h) test procedures (analytical methods) are specified at 40 CFR part
136.  They include Methods for Chemical Analysis of Water and Waste (MCAWW); the 600- and 1600-
series methods; methods published by consensus standards organizations; methods used by the U.S.
Geological Survey; methods developed by the environmental community; and other methods referenced in
CWA regulations. EPA uses these test procedures to support development of effluent limitations guidelines
approved at 40 CFR parts 400 - 499, to establish compliance with (NPDES) permits issued under CWA
section 402, for implementation of the pretreatment standards issued under CWA section 307, and for
CWA section 401 certifications.

       The SDWA requires the EPA Administrator to promulgate National  Primary Drinking Water
Regulations (NPDWRs) that specify maximum contaminant levels (MCLs) or treatment techniques for
listed drinking water contaminants (section 1412). In addition, section 1445(a) of SDWA authorizes the
Administrator to  establish regulations for monitoring to assist in determining whether persons are acting in
compliance with  the requirements of SDWA. EPA's approval of analytical test procedures is authorized
under these sections of SDWA, as well as the general rulemaking authority in SDWA section 1450(a).

       SDWA section 1401 (1)(D) specifies that NPDWRs contain criteria and procedures to ensure a
supply of drinking water that dependably complies with MCLs, including quality control (QC) and testing
procedures to ensure compliance with such levels and to ensure proper operation and maintenance of
drinking water supply and distribution systems.  These test procedures (analytical methods) are approved at
40 CFR part 141. They include MCAWW methods; the 200-, 300-, and 500- series methods; and other
methods referenced in SDWA regulations.  EPA uses these test procedures to establish MCLs under
SDWA section 1412 and to establish monitoring requirements under SDWA Section 1445(a).

1.1.2   40 CFR 136.4, 136.5 and 141.27

       Requirements for approval of alternate analytical techniques (methods) are specified at 40 CFR
136.4 and 136.5 for wastewater methods and at 40 CFR 141.27 for drinking water methods. These
requirements  are the basis for the Agency's alternate test procedure (ATP) program for water methods.
Under the ATP program, an organization may submit an application for approval of a modified version of
an approved method or for approval of a new method to be used as an alternate to an approved method.
The submitting organization is responsible for validating the new or modified method.  The Agency reviews
the ATP validation package and, if required, promulgates successful applications in the CFR. Rulemaking
is required when a new or revised method is added to the list of approved methods in the CFR.  The ATP
and rulemaking processes make heavy demands on stakeholder, contractor, EPA, and Federal Register
resources. These processes can require several months to approve a minor method modification and a year
or more to promulgate a major modification or a new  technology.  Because advances in analytical
technology continue to outpace the capacity of OW's method approval program, the program has been
                                                                           March 22, 1999

-------
under-utilized and slow to respond to emerging technologies.  This protocol is intended to specify a more
rapid and less resource intensive process for approval of new technologies.

1.1.3   Description of Document

       This protocol details the requirements for approval of new methods to be included at 40 CFR part
136 or 141.  A new method is a set of procedures that has been written in the seventeen section standard
EPA format as detailed in the Guidelines and Format for Methods to be Proposed at 40 CFR Part 136 or
Part 141; contains standardized QC elements with associated QC acceptance criteria; employs a
determinative technique for an analyte of concern that differs from determinative techniques employed for
that analyte in methods previously approved at 40 CFR part 136 or 141 and employs a determinative
technique that is as sensitive and/or selective as the determinative techniques in all methods previously
approved for the analyte.

       The new methods approval program provides chemists with the opportunity to utilize best
professional judgement to enhance compliance monitoring. Approval for a new  method may be sought
when the new method reduces analytical costs, overcomes matrix interferences problems, improves
laboratory productivity, or reduces the amount of hazardous materials used and/or produced in the
laboratory.  The new methods approval program thus can serve as a mechanism for gaining approval of
innovative technologies for use in compliance monitoring programs.  The protocol described in this
document is designed to reduce the barriers to gaining acceptance of new methods, to spur the development
and use of new  technologies, and to expedite the review and approval process for gaming acceptance of a
new method. A method developer may apply to gain approval for the use of a new method for
determination of an analyte of interest to the NPDES or NPDWR monitoring programs by developing and
validating the new method using either the procedures described in this document or the classical
interlaboratory validation procedures provided by organizations such  as ASTM1  and AOAC-
International.2'3  While EPA can be contacted at any point for assistance,  EPA's main role will be to review
the application for completeness and to determine acceptability. Consequently, EPA will be able to
approve new methods for use more quickly and efficiently.

1.2     Tiered System for Validation of New Methods

       EPA recognizes that a formal interlaboratory method validation may not be suitable for all
situations and may be prohibitively costly to implement, especially for small laboratories and regulated
entities. Therefore, EPA has developed a three-tiered, cost-effective  approach to method validation that
classifies the intended use of a new method and requires a method validation study that reflects the level of
use associated with each tier. An applicant would have to determine  the most appropriate tier for the new
method and develop QC acceptance criteria using the procedures specified in Appendix D of this protocol.
The three method validation tiers are listed below.

Tier 1  methods may only be used by a single laboratory (limited-use) for one or  more matrix type(s). A
matrix type is defined as a sample medium (e.g., air, soil, water, sludge)  with common characteristics
across  a given industrial subcategory.  For example, C-stage effluents from chlorine bleach mills, effluent
from the continuous casting subcategory of the iron and steel industrial category, POTW sludge, and in-
process streams in the Atlantic and Gulf Coast Hand-shucked Oyster  Processing subcategory are each a
matrix type. Tier 1 validation requires a single laboratory validation study in the matrix type(s) of interest.
                                                                            March 22, 1999

-------
Tier 2 methods may be used by all laboratories (nationwide use) for only one matrix type.  Validation
requires a three-laboratory validation study.

Tier 3 methods may be used by all laboratories (nationwide use) for all matrix types.  Validation requires
a nine-laboratory validation study.
                                                                               March 22, 1999

-------
2.0     APPLICATION REQUIREMENTS

       Every new method application shall be made in triplicate and include a completed new method
approval application form (provided in Appendix A) with required attachments.

2.1     Submission Addresses

       A summary of where to submit new method applications and the approval authorities for each tier
level is provided in Table 1.

                  Table 1: Submission of New Method Applications
TIER
Tier 1
Tier 2
Tier3
LEVEL OF
USE
Limited Use
for
Wastewater
Nationwide
Use
Nationwide
Use
APPLICANT
EPA Regional laboratories
States, commercial laboratories,
individual dischargers, or permitees
in States that do not have authority
States, commercial laboratories,
individual dischargers, or permitees
in States that have authority
All applicants
All applicants
SUBMIT
APPLICATION TO1
EPA Regional
Administrator2
(Regional ATP
coordinator)
EPA Regional
Administrator2
(Regional ATP
coordinator)
Director of State
Agency issuing the
NPDES permit2
Director, Analytical
Methods Staff, EPA
Headquarters
Director, Analytical
Methods Staff, EPA
Headquarters
APPROVAL
AUTHORITY
EPA Regional
Administrator
EPA
Administrator
EPA
Administrator
1 See Appendix B for EPA addresses.

2 The Regional ATP coordinator may choose to forward Tier 1 (LU) applications to the Director of the
Analytical Methods Staff (AMS) for an approval recommendation.
Upon receipt of the application, AMS staff will assign an identification number to the application. The
applicant should use the identification number in all future communications concerning the application.
                                                                       March 22, 1999

-------
2.2     Application Information
       Information required on the new method application form includes: the name and address of the
applicant; the date of submission of the application; the method number and title of the proposed new
method; the analytes(s) for which the new method is proposed; the type of application (i.e., wastewater,
drinking water, or a combined wastewater/drinking water application); the level of use desired (i.e., limited
use or nationwide use); the tier level at which the proposed new method will be validated; and, for limited-
useapplications, the applicant's NPDES permit number, the issuing agency, the type of permit and the
discharge serial number if applicable.

       The following items must be submitted with the application:  the justification for proposing the new
method; the proposed new method prepared in standard EPA format; the method validation study report,
including supporting data; and, for nationwide applications that will undergo rulemaking, method
development information and documentation that EPA can use in preparing the preamble and docket for the
proposed rule.

       Before proceeding with the new method validation, the Agency strongly encourages an applicant to
submit its validation study plan for EPA review and comment.

       The elements required for a complete application at each tier are presented in Table 2. EPA must
receive all required application information and attachments before the application is considered complete.

                            Table  2. Application Requirements
          Tier
         Tierl
         Tier 2
         Tier 3
    Level of Use
Limited Use
                   Nationwide Use
    Application Requirements
Completed application form
Justification for new method
Method in EPA format
Validation study report
                            Completed application form
                            Justification for new method
                            Method in EPA format
                            Validation study report
                            Method development information and
                            documentation
2.2.1   Justification for New Method

       The entity that proposes a new method should provide a brief justification for why the new method
is being proposed. Examples include but are not limited to: the new method successfully overcomes some
or all of the interferences associated with the approved method; the new method significantly reduces the
amount of hazardous wastes generated by the laboratory; or the cost of analyses are significantly reduced
when using the new method.
                                                                            March 22, 1999

-------
2.2.2   Standard EPA Method Format

       In accordance with the standard EPA format advocated by EPA's Environmental Monitoring
Management Council (EMMC), methods must contain 17 specific topical sections in a designated order.
The 17 sections listed in Appendix C to this document are mandatory for all methods. Additional
numbered sections, may be inserted after Section 11.0, Procedure, as appropriate for a particular method.
For more detailed information on the EPA format for proposed methods, see the Guidelines and Format
document.4

2.2.3   Validation Study Report

       The applicant must conduct a validation study and provide a comprehensive validation study report
with the new method application.  The validation study report must include the following elements:

•      Background
•      Study Design and Objectives
•      Study Implementation
•      Data Reporting and Validation
•      Results
•      Development of QC Acceptance Criteria
•      Data Analysis/Discussion
•      Conclusions
•      Appendix A - The Method
•      Appendix B - Validation Study Plan (optional)
•      Appendix C - Supporting Data (Raw Data and Example Calculations)

       These elements are described in Section 3.6.

2.2.4  Method Information and Documentation to Facilitate EPA Preparation of Preamble and
       Docket

       For Tier 2 and 3 applications, the new method will be approved by the EPA Administrator through
rulemaking.  In these cases, the applicant shall provide to EPA information and documentation that will aid
EPA in preparing the preamble and docket for the proposed rule that will be published in the Federal
Register. Information to be provided includes: a detailed background and summary of the method,  a
discussion of QC acceptance criteria development, and a description and discussion of the interlaboratory
method validation study and any other method studies conducted during method development and
validation. Specifically, the applicant shall  submit information that:

•      Defines the purpose  and intended use of the method
•      States what the method is based upon, noting any relationship of the method to other existing
       analytical methods and indicates whether the method is associated with a sampling method
•      Identifies the matrix(ces) for which the method has been found satisfactory
•      Describes method limitations and indicates any means of recognizing cases where the method may
       not be applicable to the specific matrix types
•      Outlines the basic steps involved in performing the test and data analysis
•      Describes the QC acceptance criteria development process and gives example calculations
                                                                           March 22, 1999

-------
•      Lists options to the method, if applicable
•      Discusses in a summary fashion the acceptability criteria for the method
•      Describes and discusses the validation study report, including study design and objectives, study
       limitations, study management, technical approach, data reporting and validation, results, data
       analysis discussion, and conclusions

       Previous method rules that may serve as examples of the type of information and the appropriate
level of detail necessary include 49 FR 43234, October 26, 1984; 56 FR 5090, February 7, 1991; 60 FR
53988, October 18, 1995; and 61 FR 1730, January 23, 1996.  In addition to method information, the
applicant must provide copies of all relevant supporting documents used in developing the method, for
EPA's inclusion in the rule docket.

2.3     Proprietary Information in Applications

       All information provided to the Federal government is subject to the requirements of the Freedom
of Information Act.  Therefore, any proprietary information submitted with the proposed new method
application should be marked as confidential. EPA staff will handle such information according to the
regulations in subparts A and B of 40 CFR Part 2.

       In accordance with 40 CFR ง2.203, a business that submits information to EPA may assert a
business confidentiality claim covering the information by placing on (or attaching to) the information at
the time it is submitted to EPA, a cover sheet, stamped or typed legend, or other suitable  form of notice
employing language such as trade secret, proprietary, or company confidential. Allegedly confidential
portions of otherwise non-confidential documents should be clearly identified by the business, and may be
submitted separately to facilitate identification and handling by EPA. If the business desires confidential
treatment only until a certain date or until the occurrence of a certain event, the notice should so state.

       If a claim of business confidentiality is not made at the time  of submission, EPA will make such
efforts as are administratively practicable to associate a late claim with copies of previously submitted
information in EPA files. However, EPA cannot assure that such efforts will be effective in light of the
possibility of prior disclosure or widespread prior dissemination of the information. Methods to be
proposed in the Federal Register cannot be claimed as confidential.
                                                                            March 22, 1999

-------
3.0     METHOD VALIDATION

3.1     Introduction

       Method validation is the process by which a method developer substantiates the performance of a
new method.  New methods must be validated to prove that they accurately measure the concentration of an
analyte in an environmental sample. If, during a compliance inspection or audit, it is determined that a
regulated party is using an unvalidated new method, the data generated by the unvalidated method will be
considered unacceptable for compliance monitoring or reporting.  The validation requirements listed below
were developed to reflect the level of intended use of the method. This is accomplished through a three-
tiered approach, as shown in Table 3.


                         Table 3: Tiered Validation Strategy
Tier Level
Tier 1
Tier 2
Tier3
Laboratory Use
Single Laboratory
(Limited-use or LU)
All Laboratories
(Nationwide use or
NW)
All Laboratories
(Nationwide use or
NW)
Applicable to ...
One or more matrix types from any industry; (Excluding
One matrix type within one industrial subcategory; or all
PWSs)
PWSs
All matrix types from all industrial subcategories
       Under Tier 1, single laboratories will be allowed to validate and use new test methods without the
burden of conducting an interlaboratory validation study, whereas new methods intended for multi-
laboratory use in a given industrial subcategory (Tier 2) or for multi-laboratory use for all industrial
subcategories (Tier 3) require interlaboratory testing.

3.2     Summary of Validation  Requirements

       EPA has developed a tiered validation approach that coordinates validation requirements with the
level of intended use of the new method. Tier 1 (LU) represents validation in a single laboratory, Tier 2
(NW) represents interlaboratory validation in one industrial subcategory, and Tier 3 (NW) represents
interlaboratory validation in multiple matrix types. New methods may be used after validation at the
appropriate level is performed and formal approval is granted by the appropriate authority. Tier 1 (LU)
contains two levels of validation, depending on whether the individual laboratory will be applying the new
method to a single matrix type or to multiple matrix types. The Tier 1- Single Matrix Type category
allows the laboratory to apply the new method to a single matrix type. The Tier 1- Multiple-Matrix Type
category allows a single laboratory to apply the new method to an unlimited number of matrix types after
the method has been validated on a minimum of nine matrix types.

       Table 4 summarizes the validation requirements for wastewater new methods. Table 5 summarizes
the validation requirements for drinking water new methods. Only Tier 2 (NW) validations are applicable
                                                                            March 22, 1999

-------
to drinking water because the Office of Ground Water and Drinking Water (OGWDW) no longer approves
Tier 1 (LU) new methods and the drinking water program regulates a limited number of matrix types.
               Table 4.  Summary of Validation Requirements for New Methods
                                 for the Analysis of Wastewater(1)
Method Application
Number of
Matrix
Labs types
Number of Analyses Required
IPR- reagent IPR- sample
waterฎ matrix(3) MS/MSD
MDL(4)
Tier 1-Single-lab
First matrix type
Each additional matrix type
(8 max.)
Tier 2-Multi-lab, single
matrix type
Tier 3-Multi-lab, multiple
matrix types
All matrix types
4

0<5)


12


36
18<6)
 7

 0<5)


21

63
Notes:
(1)     Numbers of analyses in this table do not include background analyses or additional QC tests such as
        calibration, blanks, etc. Validation requirements are based on the intended application of the method.
        Nine would be the maximum number of matrix types (or facilities) that would be required to validate a
        new wastewater method at Tier 1 or 3.
(2)     IPR reagent water analyses would be used to validate method performance and to establish QC acceptance
        criteria for initial precision and recovery (IPR) and ongoing precision and recovery (OPR) for a new
        method. The required number of IPR analyses, except as noted under footnote 6, would be four times the
        number of laboratories required to validate a new method because each laboratory would perform a 4-
        replicate IPR test.
(3)     IPR sample matrix analyses would be used to establish QC acceptance criteria for matrix spike/matrix
        spike duplicate (MS/MSD) recovery and precision for a Tier 1 new method only.  IPR sample matrix
        analyses would not be required for validation of Tier 2 or 3 new methods because this variability data
        would be obtained from MS/MSD tests.
(4)     A method detection limit (MDL) test would be performed in each laboratory using the new method. 40
        CFR part 136, Appendix B, requires a minimum of seven analyses per laboratory to determine an MDL.
        Each lab involved in validation of a new wastewater method would demonstrate that the new method
        would achieve the detection limits  specified in the regulations at 40 CFR parts 136 and/or in another EPA
        specified documents.
(5)     The MDL, reagent water IPR, and sample matrix IPR tests would not have to be repeated after the first
        matrix type or facility was validated.
(6)     The MS/MSD analyses would establish MS/MSD recovery and precision for the new method. The
        required number of MS/MSD analyses would be two times the number of facilities or matrix types tested.
(7)	The number of laboratories and samples would vary if a conventional interlaboratory study is used.	
                                                                                  March 22, 1999

-------
    Table 5.  Summary of Validation Requirements for New Methods for the Analysis of
                                      Drinking Water (1)
Method Application
Number of
Labs PWSs
Number of Analyses Required
IPR- reagent IPR- sample
water'2' matrix*3' MS/MSD
MDL(4)
Tier 2-Multilab               3          3           12            0           6(4)          21

Notes:
(1)      Numbers of analyses in this table do not include background analyses or additional QC tests such as
        calibration, blanks, etc.
(2)      IPR reagent water analyses would be used to validate method performance and to establish QC acceptance
        criteria for initial precision and recovery (IPR) and ongoing precision and recovery (OPR) for a new
        method.
(3)      IPR sample matrix analyses would not be required for validation of Tier 2 new methods because this
        variability data would be obtained from MS/MSD tests.
(4)      A method detection limit (MDL) test would be performed in each laboratory using the new method. 40
        CFR part 136, Appendix B, requires a minimum of seven analyses per laboratory to determine an MDL.
       All new methods must be validated to demonstrate that the method is capable of yielding reliable
data for compliance monitoring purposes. All validation study results must be documented in accordance
with the requirements outlined below.

3.3    Tier 1, 2, and 3 Validation Studies

       The tiered approach to validation encourages laboratories to take advantage of new technologies,
overcome matrix interference problems, lower detection limits, improve the reliability of results, lower the
costs of measurements, and improve overall laboratory productivity without undertaking costly and time-
consuming interlaboratory studies. Tier 1 is expected to be used by commercial laboratories, dischargers,
and state and municipal laboratories repetitively testing samples from the same site(s) on a routine basis.
Tier 2 studies are expected to be used by water supply laboratories, dischargers, and state and municipal
laboratories repetitively testing samples from multiple sites within the same industrial subcategory on a
routine basis.  Tier 3 studies are  expected to be used by vendors, commercial laboratories, dischargers, and
state and municipal laboratories testing a wide variety of sample matrices from diverse sites.  Tier 3 also is
expected to be used by vendors seeking nationwide approval of a new technology.

3.3.1    Tier 1 Validation Studies (for Wastewater Only)

       The primary intent of Tier 1 is to allow use of a new method by a single laboratory.  Tier 1 can be
applied to one or more matrix types.

Tier 1 - Single Matrix Type
                                                                              March 22, 1999
                                               10

-------
       Tier 1- single matrix type validation studies are performed in a single laboratory on a single matrix
type.  Results of the validation study and the method are applicable in this laboratory to this matrix type
only and cannot be used by another laboratory or for another matrix type.

Tier 1 - Multiple Matrix Types

       If a laboratory intends to apply the method to more than one matrix type, the laboratory must
validate the method on each matrix type.  Nine matrix types for new methods for the analysis of wastewater
is the maximum number of matrix  types to which the new method must be applied to demonstrate that it
will likely be successful for all other matrix types. EPA chose this upper limit of matrix tests for Tier 1-
multiple matrix types validation, because the maximum number of matrices tested should not be greater
than the number required for Tier 3 validation of a wastewater method (nine). Therefore, nine different
wastewater matrix types is the number after which a test on each subsequent matrix type is not required.
The specific tests to be conducted on the first wastewater matrix type and those for each additional matrix
type are enumerated in Table 4.  In all cases, the laboratory must try to determine if the measurement result
for the target analyte using a new matrix type differs from the result obtained in a reagent water matrix or
in a previously validated matrix type.

       Matrices that must be tested for Tier 1- multiple matrix type validation of a new method are given
in Table 6. As with a Tier 1-  single matrix type validation study, Tier 1- multiple  matrix type validation
studies are performed in a single laboratory and, therefore, cannot be  transferred to another laboratory. If a
method is validated by a single laboratory in two  to eight discrete matrix types, the validation is applicable
to those matrix types only. However, once a laboratory has validated the method  on nine matrix types, and
those matrix types possess the characteristics required in Table 6, the  validation is applicable to all other
matrix types.

       If results of Tier 1 - multiple matrix type validation studies are to be applied to a different medium
(e.g., air, water, soil, sludge), each medium must be represented in the samples tested in the validation
study.
                                                                               March 22, 1999
                                               11

-------
                           Table 6.  Matrix Types Required for
                         Multiple Matrix Type Validation Studies
  1.       Effluent from a POTW
  2.       ASTM D 5905 - 96, Standard Specification for Substitute Wastewater
  3.       Sewage sludge, if sludge will be in the permit
  4.       ASTM D 1141 - 90 (Reapproved 1992), Standard Specification for Substitute Ocean Water, if ocean
         water will be in the permit
  5.       Drinking water, if the method will be applied to drinking water samples
  6.       Untreated and treated wastewaters to a total of nine matrix types


  At least one of the above wastewater matrix types must have at least one of the following characteristics:
  •       Total suspended solids (TSS) greater than 40 mg/L
  •       Total dissolved solids (TDS) greater than 100 mg/L
  ซ       Oil and grease greater than 20 mg/L
         NaCl greater than 120 mg/L
         CaCO3 greater than 140 mg/L
3.3.2    Tier 2 Validation Studies for Wastewater and Drinking Water

        The primary intent of Tier 2 is to allow all regulated entities and laboratories to apply a new
method to a single sample matrix type in a single industry. Since drinking water is considered a single
matrix type and PWSs represent a single industry, Tier 2 facilitates nationwide use of a new method for the
analysis of drinking water.

        EPA believes that implementation of Tier 2 will encourage the development and application of
techniques that overcome matrix interference problems, lower detection limits, improve the reliability of
results, lower the costs of measurements, and improve overall laboratory productivity when analyzing
samples from a given industry.

        Significant industries within Tier 2 are: PWSs, publicly-owned treatment works (POTWs), and
individual industrial subcategories that are defined in the regulations at 40 CFR parts 405 - 503. At
present, there are approximately 650 industrial subcategories defined in the Part 405 -  503 regulations,
each of which constitutes an individual industry under this protocol.

        Tier 2 validation studies are performed in a minimum of three laboratories. Samples of the same
matrix type (e.g., drinking water, final effluent, extraction-stage effluent,) are collected from one or more
facilities in the same industrial subcategory.  In all cases, the  laboratory must try to determine if the
measurement result for the target analyte using a new method differs from the result obtained in a reagent
water matrix or in a previously validated matrix type or PWS sample.
                                                                               March 22, 1999
                                               12

-------
       Drinking water sources tested for Tier 2 validation of a new method for the analysis of drinking
water must include samples collected from PWSs with water quality characteristics that are sufficiently
different so that sample matrix effects, if any, can be observed.  Selection of suitable PWSs requires a
knowledge of the chemistry in the method. Analysts may review an applicable approved or published
method for indications of matrix effects that are unique to the analyte separation and measurement
technologies used in the new method. Water quality characteristics that can affect analysis of drinking
water samples include, but are not limited to: pH, total organic  carbon content, turbidity, total organic
halogen content, ionic strength, sulfate contamination, metal contamination, and trihalomethane
contamination of the drinking water sample.

       For POTWs, if a new method is validated on final effluent only, that method would be applicable
to final effluent only, and the title of the method must reflect that the method is applicable to final effluent
only.  If influent to treatment, primary effluent, and sludges are  to be monitored, the method must be
validated separately on these sample matrix types.

       In contrast to Tier 1, once a new method has been validated, the validation study results can be
transferred to other laboratories, and the  other laboratories may  freely use the method, as long as the
method is applied to analysis of samples  of the validated matrix  type from within the industrial
subcategory, and as long as the other laboratories meet all of the method's QC acceptance criteria. If the
new method is to be applied to another matrix type the method must be validated on that matrix type
separately.

3.3.3    Tier 3  Validation Studies

       The primary intent of Tier 3 is to allow nationwide use of a new method by all regulated entities
and laboratories  for all matrix types. The increased flexibility at Tier 3  should allow vendors to establish
that new devices and reagents produce results that are acceptable for compliance monitoring purposes, and
should allow commercial laboratory chains to apply new technologies and techniques throughout their
chain of laboratories to all matrix types.

       Tier 3 validation studies are performed in a minimum of nine laboratories, each with a different
matrix type, for a total of nine samples.  The minimum requirements for sample matrix types that must be
used in the validation  study are given in Table 6.  If the method is to be applied to more than one sample
medium (e.g., air, water, soil, sludge), a  separate validation must be performed on each medium.

       When validating a method directed at overcoming a matrix interference problem in a specific
matrix type, a minimum of three samples representative of those matrix types must be included in the
matrix types required by Item 6  in Table 6.  For example, if a new method is intended to overcome matrix
interferences associated with effluents containing high concentrations of polymeric materials from indirect
industrial discharges in the Thermoplastic Resins subcategory of the Organic Chemicals, Plastics, and
Synthetic Fibers industrial category, the method must be tested on a minimum of three such discharges.
Where possible,  EPA will assist the method  developer in identifying sources for samples of such
discharges.
                                                                              March 22, 1999
                                              13

-------
3.4     Development of a Validation Study Plan

       Prior to conducting Tier 1, 2, or 3 validation studies, the organization responsible for conducting
the study should prepare a detailed study plan. The validation study plan should contain the elements
described in Sections 3.4.1 through 3.4.6 of this document.

3.4.1   Background

       The Background section of the validation study plan must do the following:

•      Identify the new method
•      Identify the program use of the new method (drinking water or waste water or both)
•      Include a summary of the new method
•      Describe the reasons for development of the new method, the logic behind the technical approach of
       the new method, and the result of the new method
•      Identify the matrices, matrix types, and/or media to  which the new method is believed to be
       applicable
•      List the analytes measured by the new method including corresponding CAS Registry or EMMI
       numbers
•      Indicate whether any, some, or all known metabolites, decomposition products, or known
       commercial formulations containing the analyte are included in the measurement.  For example, a
       method designed to measure acid herbicides should include the ability to measure the acids and
       salts of these  analytes; a total metals method must measure total metals.

3.4.2   Objectives

       The Objectives section of the validation study plan should describe overall objectives and data
quality objectives of the study.

3.4.3   Study Management

       The Study Management section of the validation study plan should do the following:

•      Identify the organization responsible for managing the study
•      Identify laboratories, facilities, and other organizations that will participate in the study
•      Delineate the study schedule

3.4.4   Technical Approach

       The Technical Approach section of the validation study plan  should do the following:

•      Indicate at which tier the study will be performed
•      Describe the approach that will be followed by each organization involved in the study
•      Describe how sample matrices and participating laboratories will be  selected
•      Explain how  samples will be collected and distributed
•      Specify the numbers and types of analyses to be performed by the participating laboratories
•      Describe how analyses are to be performed
                                                                            March 22, 1999
                                             14

-------
3.4.5   Data Reporting and Evaluation

       This section of the validation study plan should explain the procedures that will be followed for
reporting and validating study data, and should address statistical analysis of study results.

3.4.6   Limita tions

       The Limitations section of the validation study plan should explain any limiting factors related to
the scope of the study.

3.5     Detailed Procedures for Conducting Validation Studies

       When validating new methods, laboratories must adhere to the standardized QC elements detailed
in the proposed new method.  Laboratories must use a reference matrix (usually, reagent water) and field
samples for the validation study.

3.5.1   Method Compilation

       Prior to conducting a validation study, the organization responsible for modifying the method
should detail  the full method in accordance with EPA's Guidelines and Format document.4 If the
organization that develops a new method is a consensus standards organization or government organization
with a standardized format, that format may be used.  The documented method should be distributed to
each laboratory participating in the validation study to ensure that each laboratory is validating the same
set of procedures.

3.5.2   Method Detection Limit Study

       Each laboratory participating in the Tier 1, 2, or 3 validation study shall use the procedures
specified in the new method and perform an MDL study in accordance with the procedure given at 40 CFR
part 136, Appendix B.

       For validation studies of a new method, each laboratory participating in the study must use the
results of the  MDL study to determine a minimum level (ML) of quantitation. Determination of an ML for
new drinking water methods is encouraged but not required, because the regulations at 40 CFR part 141
specify detection and sometimes quantitation limits for all regulated analytes.

3.5.3   Calibration

       Following completion of the MDL study, each laboratory participating in the study must perform a
multi-point calibration in accordance with the procedures specified in the new method. However, a single-
point calibration is allowed if the < 2% relative standard deviation (RSD) criteria given in Table D-1 (see
Appendix D) are met. The method developer shall use the data from the laboratories participating in the
study to develop linearity criterion.
                                                                            March 22, 1999
                                             15

-------
3.5.4   Initial Precision and Recovery

       After successfully calibrating the instrument, each laboratory participating in the study shall
perform initial precision and recovery (IPR) analyses using the procedures specified in the method. The
IPR consists of analyses of four replicates of reagent water spiked with the analytes of interest.  The
method developer shall use the results of these IPR analyses to develop precision and recovery QC
acceptance criteria.  The concentration of the IPR samples must be stated in the method.  This
concentration should be between one and five times the ML.

3.5.5   Field Sample Analyses

       After laboratories participating in the Tier 1, 2, or 3 validation study have successfully completed
the IPR analyses, the new method is validated on the matrix type(s) chosen for the validation study. The
numbers of analyses required are described below.

3.5.5.1  Tier 1 - Single (First) Matrix Type

       In a Tier 1-  single matrix type study performed to validate a new method, the laboratory must
analyze four spiked replicates of the matrix type to which the new method will be applied. The replicate
samples must be spiked with the analyte(s) of interest at a concentration one to five times the background
concentration of the analyte(s) in the sample or at one to five times the ML, whichever is greater.  In other
words, the laboratory will perform an IPR test in the matrix type of interest.  Prior to spiking the replicate
samples, the laboratory must determine the background concentration of an unspiked aliquot. In all, Tier
1 - single matrix type validation studies of new methods will require analysis of five field samples (one
background and  four matrix). The organization responsible for developing the method must use the results
of these sample analyses to develop MS/MSD precision and recovery QC acceptance criteria.

3.5.5.2  Tier 1 - Multiple (Additional) Matrix Types

       In Tier 1 - multiple matrix type studies performed to validate new methods, the laboratory must
determine QC acceptance criteria using a single matrix of interest as outlined in Section 3.5.5.1, and
determine the background concentration and analyze an MS/MSD pair for each additional matrix type
being tested, up to a total of eight additional matrix types.  For a method to be validated for each additional
matrix type, the results of the background/MS/MSD samples must fall within the QC acceptance criteria
determined in the single matrix. Because three  field sample analyses are required for each matrix type (one
background, one MS, and one MSD), and between two and nine matrix types may be  tested, a Tier 1-
multiple matrix type validation study will require analysis of 8 - 29 samples.

3.5.5.3  Tier 2  Validation Studies

       In a Tier 2 validation study, each of the three laboratories will determine the background
concentration and analyze an MS/MSD pair on the sample it receives.  Because there are three
laboratories, each of which performs three analyses (one background, one MS, and one MSD), Tier 2
validation studies will require analysis of 9 samples. The laboratory responsible for developing the new
method must use the results of these samples analyses to develop MS/MSD precision  and recovery QC
acceptance criteria.
                                                                             March 22, 1999
                                              16

-------
3.5.5.4  Tier 3 Validation Studies

       In a Tier 3 validation study, each of the nine laboratories participating in the study will determine
the background concentration and analyze an MS/MSD pair on the sample it receives. Because there are a
total of nine laboratories, each performing three field sample analyses (one background, one MS, and one
MSD), a Tier 3 validation study will require analysis of 27 samples.  The laboratory responsible for
developing the new method must use the results of these samples analyses to develop MS/MSD precision
and recovery  QC acceptance criteria.

3.5.6   Ongoing Precision and Recovery

       If the field samples discussed in Section 3.5.5 are analyzed as a batch with the IPR samples,
analysis of an OPR sample is unnecessary in the validation study. If, however, field samples are analyzed
in a different  batch or batches, then each laboratory participating in the Tier 1, 2, or 3 validation study
must analyze  an OPR sample with each batch. The concentration of the OPR sample must be as stated in
the method being validated.

       The organization responsible for developing the method must use the results of the IPR tests
described above in Section 3.5.4 to develop OPR recovery criteria as described in Appendix D (Section
9.0).

3.5.7   Calibration Verification

       If the field samples discussed in Section 3.5.5 are analyzed on the same shift or in the same set of
instrumental determinations as the initial calibration sequence, calibration verification is unnecessary.
However, if field samples are analyzed on a different shift or in a different instrument batch, each
laboratory participating in the Tier 1, 2, or 3 validation study must verify calibration as described in the
method.

       The organization responsible for developing the method must use the results of the calibration
sequence described above in Section 3.5.3 to develop QC acceptance criteria for the calibration verification
analyses as described in Appendix D (Section 9.0).

 3.5.8  Contamination Level in Blanks

       Each  laboratory that participates in a Tier 1, 2, or 3 validation study must prepare and analyze at
least one method blank with the sample batch during which the matrix samples are prepared and analyzed.
The actual number of blank samples analyzed by each laboratory must meet or exceed the frequency
specified in the method.

       For validation of a new method, the laboratory responsible for the development of the new method
must use the results of these sample analyses to develop QC acceptance criteria for allowable blank
contamination.
                                                                             March 22, 1999
                                              17

-------
3.5.9    Surrogate or Labeled Compound Recovery

       For methods that use surrogates or labeled compounds, each laboratory participating in the Tier 1,
2, or 3 validation study must spike all field and QC samples with the surrogates/labeled compounds at the
concentrations specified in the method.

       The laboratory responsible for the development of the new method must use the results of these
sample analyses to develop surrogate or labeled compound recovery QC acceptance criteria.

3.5.10  Absolute and Relative Retention Time

       Each laboratory participating in a Tier I, 2, or 3 validation study of a chromatographic method
must determine the absolute and relative retention times of the analytes of interest.

       For validation of a new method, each laboratory participating in the study must use the results of
these sample analyses to develop absolute and relative retention time QC  acceptance criteria.

3.5.11   Further Validation Studies

       After completing the Tier 1,  2, or 3 validation studies of new methods, the organization responsible
for  developing the method must document the study results and submit them to EPA.  If, based on its
review of the method, EPA concludes that the method is not sufficiently rugged or reliable for its intended
use, EPA may require further method development and further testing to define the stability  and reliability
of the method.  The tests and studies that must be performed in this case are dependent upon the analyte(s)
and the analytical system, and will be determined on a case-by-case basis  as these situations  arise.

3.6     Validation Study Report

       Laboratories or other organizations responsible for developing new methods at Tier 1, 2, or 3 must
document the results of the validation study in a formal validation study report that is organized and
contains the elements described in this section. In all cases, a copy of all required validation data should be
maintained at the laboratory or other organization responsible for developing the new method.

       The information and supporting data required in the validation study report must be  sufficient to
enable EPA to evaluate the performance of a new method. If data are collected by a contract laboratory,
the  organization responsible for using the method (e.g., permittee, POTW, PWS, or other regulated entity)
is responsible for ensuring that  all method-specified requirements are met by the contract laboratory and
that the validation study report contains  all required data.

       Like the validation study plan, the validation study report contains background information and
describes the study design.  In addition,  the validation study report details the process and results of the
study, provides an analysis and discussion of the results, and presents study conclusions. The validation
study plan should be appended  to and referenced in the validation study report. The validation study report
should identify  and discuss any deviations from the study plan (if developed) that were made in
implementing the study.

       The validation study report must contain the elements described in Sections 3.6.1 through 3.6.10.
                                                                             March 22, 1999
                                              18

-------
3.6.1   Background

       The Background section of the validation study report must describe the new method that was
validated and identify the organization responsible for developing the method.  This background section of
the validation study report must:

•      Include a method summary
•      Describe the reasons for and developing the new method, the logic behind the technical approach to
       the new method, and the result of the new method
•      Identify the matrices, matrix types, and/or media to which the method is believed to be applicable
•      List the analytes measured by the method including corresponding CAS Registry or EMMI
       numbers. (Alternatively, this information may be provided on the data reporting forms in the
       Supporting Data appendix to the validation study report)
•      Indicate whether any, some, or all known metabolites, decomposition products, or known
       commercial formulations containing the analyte are included in the measurement. (For example, a
       method designed to measure acid herbicides should include the ability to measure the acids and
       salts of these  analytes)
•      State the purpose of the study

3.6.2   Study Design and Objectives

       The Study Design and Objectives section of the validation study report must describe the study
design and identify overall objectives and data quality objectives of the study. Any study limitations must
be identified. The validation study plan may be appended to the validation study report to provide the
description of the study design. If no validation study plan was prepared, the study design must be
described  in this section (see Section 3.4 for required elements of the study design).

3.6.3   Study Implementation

       The Study Implementation section of the validation study report must describe the methodology
and approach undertaken in the study.  This section must:

•      Identify the organization that was responsible for managing the study
•      Identify the laboratories, facilities, and other organizations that participated in the study; describe
       how participating laboratories were selected; and explain the role of each organization involved in
       the study
•      Indicate at which Tier level the study was performed
•      Delineate the study schedule that was followed
•      Describe how sample matrices were chosen, including a statement of compliance with Tier
       requirements for matrix type selection
•      Explain how  samples were collected and distributed
•      Specify the numbers and types of analyses performed by the participating laboratories
•      Describe how analyses were performed
•      Identify any problems encountered or deviations from the study plan and their resolution/impact on
       study performance and/or results
                                                                             March 22, 1999
                                              19

-------
3.6.4   Data Reporting and Validation

       This section of the validation study report must describe the procedures that were used to report
and validate study data. EPA will not establish a standard format for analytical data submission because
of the large variety of formats currently in use.

3.6.5   Results

       This section of the validation study report presents the study results. Raw data and example
calculations are required as part of the results and shall be included in an appendix to the validation study
report (see  Section 3.6.11).

3.6.6   Development of QC Acceptance Criteria

       The validation study report must contain a section that describes the basis for development of QC
acceptance  criteria for all  of the required QC tests. The requirements for developing QC acceptance
criteria are  detailed in Appendix D of this protocol (Section 9).

3.6.7   Data AnalysislDiscussion

       This section of the validation study report must provide a statistical analysis and discussion of the
study results.

3.6.8   Con elusions

       The Conclusions section of the  validation study report must describe the conclusions drawn from
the study based on the data analysis discussion.  The Conclusions section must contain a statement(s)
regarding achievement of the study objective(s).

3.6.9   Appendix A - The Method

       A detailed step-by-step analytical method prepared in accordance with EPA's Guidelines and
Format document4, must be appended to the validation study report.

3.6.10  Appendix B - Validation Study Plan

       If a validation study plan was prepared, it should be appended to the validation study report.

3.6.11  Appendix C - Supporting Data

       The validation study report must be accompanied by raw data and example calculations that
support the results presented in the report.

3.6.11.1        Raw Data

       The Results  section of the validation study report must include raw data that will allow an
independent reviewer to verify each determination and calculation performed by the laboratory.
                                                                            March 22, 1999
                                             20

-------
This verification consists of tracing the instrument output (peak height, area, or other signal intensity) to
the final result reported. The raw data are method specific and may include any of the following:

•      Sample numbers or other identifiers used by the both the regulated entity and the laboratory
•      Sample preparation (extraction/digestion) dates
•      Analysis dates and times
•      Sequence of analyses or run logs
•      Sample volume
•      Extract volume prior to each cleanup step
•      Extract volume after each cleanup step
•      Final extract volume prior to injection
•      Digestion volume
•      Titration volume
•      Percent solids or percent moisture
•      Dilution data, differentiating between dilution of a sample and dilution of an extract or digestate
•      Instrument(s) and operating conditions
•      GC and/or GC/MS operating conditions, including detailed information on
           Columns used for determination and confirmation (column length and diameter, stationary
           phase, solid support, film thickness, etc.)
           Analysis conditions (temperature programs, flow rates, etc.)
           Detectors (type, operating conditions, etc.)
•      Chromatograms, ion current profiles, bar graph spectra, library search results
•      Quantitation reports, data system outputs, and other data to link the raw data to the results
       reported.  (Where these data are edited manually, explanations of why manual intervention was
       necessary must be included)
•      Direct instrument readouts; i.e., strip charts, printer tapes, etc., and other data to support the  final
       results
•      Laboratory bench sheets and copies of all pertinent logbook pages for all sample preparation and
       cleanup steps, and for all other parts of the determination

       Raw data are required for all samples, calibrations, verifications, blanks, matrix spikes and
duplicates, and other QC analyses required by the new method. Data must be organized so that an
analytical chemist can clearly understand how the analyses were performed. The names, titles, addresses,
and telephone numbers of the analysts who performed the analyses and of the quality assurance officer who
will verify the analyses must be provided.  For instruments  involving data systems (e.g., GC/MS), raw
data on magnetic tape or disk must be made available on request.

3.6.11.2     Example Calculations

       The validation study report must provide example calculations that will allow the data reviewer to
determine how the laboratory used the raw data to arrive at the final results. Useful examples include both
detected compounds and undetected compounds. If the laboratory or the method employs a standardized
reporting level for undetected compounds, this should be made clear in the example, as should adjustments
for sample volume, dry weight (solids only), etc.
                                                                             March 22, 1999
                                              21

-------
4.0     EPA REVIEW AND APPROVAL

4.1     EPA Review of Applications

       All requests for approval of proposed new methods will undergo review by EPA. Limited-use new
methods (Tier 1) will be approved through an EPA letter of approval.  New methods proposed for
nationwide-use (Tiers 2 and 3) will be approved through rulemaking.  Proposed test procedures prepared
under this protocol should demonstrate an improvement over current EPA- approved methods that offers
one or more of the following advantages: better method sensitivity or selectivity, lower analytical costs,
fewer matrix interference problems, improvement in laboratory productivity, or reduction in the amount of
hazardous materials used and/or produced in the laboratory.

       EPA's Analytical  Methods Staff (AMS) at EPA Headquarters will review all nationwide-use new
methods and will review limited-use applications if requested by the EPA Regional Office or State Agency.
AMS may be assisted in its technical review by contractor personnel. When a formal new method
application is received, AMS will first check the documentation for completeness.  If the documentation is
incomplete, AMS will contact the applicant and request missing documentation before proceeding with its
review.

       At a minimum, an application must include a completed new method application form, the method
in EPA standard format (or other standard format - see Section 3.5.1), and a Validation Study Report with
supporting data, before AMS will review the package. If these elements are present, AMS will begin an
internal review of the new method for scientific merit, consistency, and appropriateness. The internal
review at EPA may involve multiple programs and workgroups.  Should any problems or questions arise
during the review, EPA or its technical support contractor will communicate with the applicant to resolve
outstanding issues. Depending on the circumstances, EPA may return the application to the applicant for
revision.  Internal review of proposed new methods will involve the three steps briefly described below.

       The first step of EPA's technical review will evaluate the description of the proposed method and
assess the new methods applicability for approval at 40 CFR parts  136 or 141. If the proposed method is
not applicable to 40 CFR parts 136 or 141 and/or the method description is not acceptable, EPA will
recommend rejection of the application. If this information is acceptable, the evaluation will proceed.

       In the second step of EPA's review, the performance of the new method will be evaluated.  The
performance (sensitivity, precision and recovery) of the method is based on  data provided by the applicant
and the development  of QC  acceptance criteria. If method performance is acceptable, the review will
continue.

       As the third and final step, EPA will perform a detailed audit of the  proposed method test data. The
evaluation of test data in applications can be accomplished more quickly if machine-readable files of test
data (and analysis software where different from EPA software) are provided on floppy disks with the
application.  Data files should be in IBM-PC compatible format, suitable for input directly into statistical
analysis software, such as the Trimmed Spearman-Karber, Probit, Dunnett,  and ICP programs.
                                                                           March 22, 1999
                                             22

-------
4.2    Approval  Recommendation

        EPA will complete its review and notify the applicant of EPA's recommendation. For limited-use
applications, the Regional Administrator will issue the formal approval for limited-use of the new method.
For all nationwide use applications (Tiers 2 or 3), AMS will notify the applicant of EPA's
recommendation, and if the new method is recommended for approval, will initiate the ralemaking process
through which the new method is formally approved by the EPA Administrator.

4.3     Rulemaking  Process

        Using the information provided with the new method application to develop the preamble, EPA will
prepare the proposed rule  for approval, complete the rule docket, pass the draft rule through internal review
at EPA, and submit it to the Office of the Federal Register (OFR) for publication.  Preparation, approval,
and publication of a proposed rule generally requires a minimum of four months, and may take longer
depending on the nature of the method.  When published, the proposed rule requests public comment and
allows a specified comment period, generally 30 to 60 days. At the end of the comment period, EPA will
forward any significant comments to the method applicant for technical  assistance to EPA in drafting
responses to comments. All comments that have scientific or legal merit, or raise substantive issues with
the proposed rule, must be answered to complete the ralemaking process.

        EPA will review the comment responses provided by the applicant and complete the response-to-
comments document for the final rule. EPA will then prepare the final rale, compile the rule docket, and
submit the final rale to the OFR for publication. The final rale will state the date that the rale becomes
effective, typically 30 days after rale publication.  As of this effective date, the method is approved by  EPA
and will be included in the appropriate table(s) at 40 CFR part 136 and/or 141 in the next CFR update. It
generally requires a minimum of eight months after the proposed rule is published to receive and
respond to comments, prepare and process the final rule through internal EPA review, and publish the
final rule in the Federal Register.
                                                                           March 22, 1999
                                             23

-------
5.0     REFERENCES

1.     ASTM, 1994. Standard Practice for Determination of Precision and Bias of Applicable Methods
       of Committee D-19 on Water. Designation D-2777-86 (Reapproved 1994). Annual Book of'ASTM
       Standards,. Vol. 11.04.

2.     Youden, W.J. and E.H.  Stiener, 1975. Statistical Manual of the AOAC. AOAC-International.
       1111 N. 19th Street; Suite 210, Arlington, VA 22209.

3.     Wernimont, G.T., 1985. Use of Statistics to Develop and Evaluate Analytical Methods. AOAC-
       International.

4.      USEPA 1996. Guidelines and Format for Methods to Be Proposed at 40 CFR Part 136 or Part
       141 (Guidelines and Format document). U.S. Environmental Protection Agency. Office of Water,
       Engineering and Analysis Division. Washington, D.C. EPA- 821-B-96-003.
                                                                         March 22, 1999
                                           24

-------
6.0     APPENDIX A - APPLICATION FORM
                                  EPA Office of Water
                             New Method Application Form
                                 for Chemical Analytes
     Applicant Name and Address:
       Date Application Submitted:
         Method Number & Title:
                    Analyte(s):
     Type (WW, DW, or WW/DW):
                   Level of Use:
                   (LUorNW)
Validation Tier:
(1,2 or 3)
 FOR LIMITED-USE APPLICATIONS ONLY:
 ID number of existing or pending permit:
 Issuing agency:
 Type of permit:
 Discharge serial number:
 ATTACHMENTS:
 |^     Justification for New Method

 |~j     Method in standard EPA format

 n     Validation Study Plan (optional)

 Q     Validation Study Report

 [j     Method Information for Preamble

 ||     Method Documentation for Docket

        Other	
                         Submit Application and Attachments in Triplicate
                                                                          March 22, 1999
                                            25

-------
7.0     APPENDIX B - EPA HEADQUARTERS AND REGIONAL CONTACTS
Headquarters
William Telliard
Director, Analytical Methods Staff (AMS)
Mail Code 4303
Waterside Mall
401 M. Street, S.W.
Washington, D.C. 20460

Region 1
Arthur Clark
QA Chemist
USEPA Region 1
EQA
60 Westview Street
Lexington, MA 02173

Region 2
Linda M. Mauel
USEPA Region 2
Division of Science and Monitoring
2890 Woodbridge Avenue (MS-220)
Building 10
Edison, NJ  08837-3679

Region 3
Charles Jones
Regional QA Officer
USEPA Region 3
Environmental Assessment and Protection Division
841 Chesnut Building
MC-3EP10
Philadelphia, PA 19107-4431

Region 4
Wayne Turnbull
Chemist
USEPA Region 4
Room: SESD
960 College Station Road
Athens, GA 30677-2700
Region 5
Kenneth Gunter
USEPA Region 5
77 W. Jackson Blvd., WT-15J
Chicago, IL 60604

Region 6
David Stockton
USEPA Region 6 Laboratory
Houston Branch
10625 Fallstone Road (6MD-HI)
Houston, TX 77099

Region 7
Doug Brune
USEPA Region 7
726 Minnesota Avenue
Kansas City, KS 66101

Region 8
Rick Edmonds
Regional Quality Assurance Officer
USEPA Region 8
999 18th Street - Suite 500 (8TMS-L)
Denver, CO 80202-2405

Region 9
Roseanne Sakamoto
USEPA Region 9
75 Hawthorne Street / P-3-2
San Francisco, CA 94105

Region 10
Bruce Woods
QAO
USEPA Region 10
1200 Sixth Avenue, OEA-095
Seattle, WA 98101
                                             26
                                                                           March 22, 1999

-------
8.0     APPENDIX C - STANDARD EPA METHOD FORMAT

The following is a listing of the 17 EMMC-required method sections. Applicants should consult the
Method Guidelines and Format document4 for a detailed description of the required content for each
section and other formatting guidelines and conventions.

1.0    Scope and application

       This section outlines the purpose, range, limitations, and intended use of the method, and identifies
       target analytes.

2.0    Summary of Method

       This section provides an overview  of the method procedure and quality assurance.

3.0    Definitions

       This section includes definitions of terms, acronyms, and abbreviations used in the method.  If
       preferred, definitions may be provided in a glossary at the end of the method or manual. In this
       case, the definitions section must still appear in the method, with  a notation that definitions are
       provided in a glossary at the end of the method.  Refer to the specific section number of the
       glossary.

4.0    Interferences

       This section identifies  known or potential interferences that may occur during use of the method,
       and describes ways to  reduce or eliminate interferences.

5.0    Safety

       This section describes  special precautions needed to ensure personnel safety during the
       performance of the method.  Procedures described here should be limited to those which are above
       and beyond good laboratory practices. The section must contain  information regarding specific
       toxicity of analytes or reagents.

6.0    Equipment and Supplies

       This section lists and describes all non-consumable supplies and  equipment needed to perform the
       method.

7.0    Reagents and Standards

       This section lists and describes all reagents and standards required to perform the method, and
       provides preparation instructions and/or suggested suppliers as appropriate.

8.0    Sample  Collection, Preservation, and Storage
                                                                            March 22, 1999
                                             27

-------
       This section provides requirements and instructions for collecting, preserving, and storing samples.

9.0    Quality Control

       This section cites the procedures and analyses required to fully document the quality of data
       generated by the method.  The required components of the laboratory's quality assurance (QA)
       program and specific quality control (QC) analyses are described in this section. For each QC
       analysis, the complete analytical procedure, the frequency of required analyses, and interpretation
       of results are specified.

       Note:  To ensure data quality, water methods must specify a comprehensive laboratory QA
       program. The method must contain the standard QC elements and specify QC acceptance criteria
       for each of those elements in accordance with Appendix D of this protocol

10.0   Calibration and Standardization

       This section describes the method/instrument calibration and standardization process, and required
       calibration verification. Corrective actions are described for cases when performance
       specifications are not met.

11.0   Procedure

       This section describes the sample processing and instrumental analysis steps of the method, and
       provides detailed instructions to analysts.

12.0   Data Analysis and Calculations

       This section provides instructions for analyzing data, and equations and definitions of constants
       used to calculate final sample analysis results.

13.0   Method Performance

       This section provides method performance criteria for the method, including precision/bias
       statements regarding detection limits and source/limitations of data produced using the method.

14.0   Pollution Prevention

       This section describes aspects of the method that minimize or prevent pollution known to be or
       potentially attributable to the method.

15.0   Waste Management

       This section describes minimization and proper disposal of waste and samples.
                                                                              March 22, 1999
                                               28

-------
16.0   References

       This section lists references for source documents and publications that contain ancillary
       information.  Note:  Each method should be a free-standing document, providing all information
       necessary for the method user to perform the method may be found. References within a method
       should be restricted to associated or source material. Procedural steps or instructions should not
       be referenced as being found elsewhere, but should be included in total within the method.

1 7.0   Tables, Diagrams, Flowcharts, and Validation Data

       This section contains all method tables and figures (diagrams and flowcharts), and may contain
       validation data referenced in the body of the method.
                                                                             March 22, 1999
                                              29

-------
9.0     APPENDIX D - QUALITY CONTROL REQUIREMENTS

9.1     Introduction

       All new methods must contain standardized QC tests and specify QC acceptance criteria for each
test.  The person or organization that develops a new method for a particular combination of analyte and
determinative technique will be responsible for validating the method and for developing the QC acceptance
criteria.  QC acceptance criteria will be based on data generated during the method validation study.
Under this protocol, EPA requires a method validation study that reflects the level of intended use for a
method. Because QC acceptance criteria will be developed from validation studies and because the
validation requirements vary with each tier, the statistical procedures used to develop the criteria will vary
by tier.

       This appendix lists and describes the standardized QC tests required in all new methods, and
outlines procedures for developing QC acceptance criteria  for new methods at Tiers 1,  2, and 3.

9.2     Standardized Quality Control Tests

       Under this protocol, standardized QC tests are a mandatory component of all new methods. The
following standardized QC tests must be included in new methods as appropriate to the technology:

•      calibration linearity
•      calibration verification
•      absolute and relative retention time precision (for chromatographic analyses)
•      initial precision and recovery
•      ongoing precision and recovery
•      analysis of blanks
•      surrogate or labeled compound recovery
•      matrix spike and matrix spike duplicate precision and recovery (for non-isotope dilution analyses)
•      method detection limit demonstration
•      analysis of a reference sample

These tests are described in Sections 9.2.1 - 9.2.10 below.

9.2.1   Calibration

       Calibration is the process of establishing the relationship between the concentration or amount of
an analyte and the response of an analytical instrument or system to the analyte.   The process begins by
measuring instrument responses to known concentrations or amounts of the analyte.  The calibration
equation is then established by fitting a line or curve through the calibration data.  Concentration is the
independent variable and the corresponding instrument response, which will include some random variation,
is the dependent variable. The most common calibration model is a straight line through the origin (zero
response at zero concentration).
                                                                            March 22, 1999
                                             30

-------
       Analyte concentrations in future samples are estimated by measuring instrument response and
applying the inverse of the calibration equation. To achieve the overall goal of obtaining the most accurate
estimates of concentrations in future samples, the most effective calibration procedures involve:

•      selecting the proper response relationship

•      calculating the most precise estimates of the parameters of the calibration equation through
       regression.

9.2.1.1  Unweighted and weighted regression

       Simple linear regression is based on the assumption that the standard deviation of the dependent
variable is constant over the range of the regression. This simple regression is also termed "unweighted
regression." For nearly all analytical  instruments and systems, the standard deviation of the response is not
constant  over the analytical range but increases with increasing concentration.  The most accurate
statistical method for estimating the best calibration line or curve with non-constant standard deviation is
"weighted regression."

       Current analytical instrument data reduction packages often include several different statistical
options for analysts, including both weighted and unweighted regression1. The unweighted or simple
regression is often the default choice. However, it leads to tremendously inefficient use of the calibration
data and  totally ignores the common observation that  instrument responses have non-constant variances.  In
a weighted regression,  it is possible to allow the more precise calibration points (typically the lower level
standards) to more heavily influence the resulting calibration curve. Therefore, weighted regression is the
most appropriate choice for nearly all analytical systems and instruments.  Application of weighted
regression to the a straight line through the origin is discussed below.  Application of weighted regression
to straight lines not through the origin and to second and higher order quadratic equations is beyond the
scope of this protocol.  EPA plans to  provide a supplement to this protocol that discusses the details of
these regressions.

       In the simplest case of calibration using a proportional model ( y  = mx ), the weighted regression
estimate  for the proportionality coefficient m is as follows:
                                                1=1  W.
                                                 1=1 W.
where yt and x, are corresponding pairs of instrument response and analyte amount, n is the number of
calibration points, and w, is the variance or squared error of response yf. With the assumption that the
1 An unweighted regression weights all points equally (hence the term unweighted) but the weighted
regression weights the more precise points more heavily.
                                                                              March 22, 1999
                                               31

-------
response error is proportional to concentration (i.e., a (>>,, x,) = fcc,; w, = x,2), the weighted regression
estimate of the proportionality coefficient becomes:

                                                 n  y
                                                y   i
                                               ; — 1 x-
                                               I — L  I
                                       m    =	-.
                                         est      n


The proportionality coefficient is estimated as the average ratio between instrument response and standard
concentration. For external standard calibration, this ratio is often termed the "calibration factor" (CF).
For internal standard calibration, terms for the response and concentration or amount of the internal
standard are included and the average ratio is termed the "response factor" (RF).  For isotope dilution, the
internal standard is a labeled compound and EPA has designed the average ratio as "relative response"
(RR). For details of use of "calibration factor," "response factor," and "response ratio," see the respective
methods in which they are used; e.g., the organic methods at 40 CFR 136, Appendix A.

9.2.1.2 Calibration linearity

        The calibration linearity specification establishes a break point between a straight line through the
origin and other forms of calibration (described below). The break point is specified as a maximum
relative standard deviation (RSD) of the:

•       relative response (RR) for isotope dilution calibration,
•       response factor (RF) for internal standard calibration, or
•       calibration factor (CF) for external standard calibration,

below which an averaged RR, RF, or CF may be used.  If the RSD is greater than the limit specified,
another form of calibration must be used.

        The number of calibration points required for calibration is dependent on the error of the measuring
technique. During method development, measurement technique error is  determined by:

•       calibrating the instrument at the minimum level of quantitation (ML) and a minimum of two
        additional points

        determining the RSD of the RR, RF, or CF.

Depending on the resulting RSD, calibration during the subsequent validation study must be performed at
the minimum number of points shown in Table D-l.  Additional criteria for RSDs are obtained  from the
results of the validation study following procedures specific to Tiers 1, 2 and 3 (see subsections on
calibration linearity under Section 9.3).
                                                                              March 22, 1999
                                              32

-------
                 Table D-l:  Minimum Number of Points Required for Calibration1
                 Percent RSD2                       Minimum Number of Calibration Points
                    0- <2                                            I3
                   2-<10                                            3
                   10-<25                                            5
                     >25                                              7
 1 Based on Rushneck et al., 1987.  Effect of number of calibration points on precision and accuracy of GC/MS,
 in Proceedings of Tenth Annual Analytical Symposium, USEPA: Washington, DC.
 2 Percent RSD shall be determined from the calibration linearity test.
 3 Assumes linearity through the origin (0,0).  For analytes for which there is no origin (such as pH), a two-point
 calibration shall be performed.

Calibrations other than a straight line through the origin are required when the linearity criterion cannot be
met. For most instruments  and analytical systems, these calibrations are straight line not through the origin
(y = mx + b) or a second-order quadratic equation (y = ax2 + bx + c). A second or higher order calibration
may be justified when an analyte can only be determined with a method that uses a determinative technique
with a nonlinear response over the calibration range. For example, enzyme-linked immunoassay methods
typically use log-logistic or similarly shaped curves for calibration. A second or higher order calibration
may be used provided that the response function increases or decreases monotonically with concentration.
A monotonic calibration function  ensures that a unique  analyte amount or concentration corresponds to a
given instrument response.

        Most instruments and analytical systems are linear over a range large  enough to preclude the need
for second order or higher calibration. If the linear range of any of these systems is limited, sample dilution
and reanalysis should be performed to bring the concentration within the linear range, rather than extend
the calibration into a nonlinear region of the response.  EPA discourages use of higher order calibrations,
where possible, because responses in the nonlinear region can mask curvature that may be attributable to
preparation of an  inaccurate standard. EPA requires that all calculations of concentrations of analytes in
blanks, field samples, QC samples, and samples prepared for other purposes be based on an averaged RR,
RF, or CF, on a straight line not through the origin, or on a calibration curve.

9.2.2   Calibration  Verification

        This test is used to periodically verify that instrument performance has not changed significantly
from calibration.  Verification is based on time (e.g., working day, 12-hour shift) or on the number of
samples analyzed in a batch (e.g.,  after every 10th sample). The terms "shift" and "batch" should be
specified in  the method. If not, the general rule has been that calibration verification is performed every 12-
hour shift on instruments used for determination of organic analytes and every 1 Oth sample on instruments
used for determination of metals.  However, the over-riding rule should be that  verification  is performed
frequently enough to ensure that the response of the instrument or analytical system has not drifted
significantly from calibration.

        Calibration verification tests are typically performed by analyzing a single standard in the
concentration range of interest for the target analyte(s).  In most methods, this standard is in the range of 1
                                                                              March 22, 1999
                                               33

-------
- 5 times the minimum level (ML) of quantitation and is at the same level as one of the standards used for
calibration. The calibration verification standard concentration should be within 1-5 times the ML rather
than at a "midpoint" concentration because specifying the midpoint can be interpreted as one-half (1/2) the
highest calibration point. Using a concentration this high when the calibration covers orders of magnitude
may lead to erroneous results, because this midpoint standard may be far removed from the range where
most measurements will be made.

       If the calibration is linear through the origin (as defined by linearity criteria in Table D-l),
specifications for calibration verification are developed to define the allowable deviation of the RR, RF, or
CF of the calibration verification standard from the averaged RR, RF, or CF of the initial calibration. If a
line with a non-zero intercept or a higher-order curve is used for calibration, specifications for calibration
verification must be defined as a maximum allowable deviation of the verification standard from the
calibration line or curve.

       For calculation of analyte concentrations in field samples, the averaged RR, RF, or CF, or the
calibration curve is always used; i.e., the calibration is not updated to the RR, RF, CF of the single point
verification standard.  Updating the calibration to a single point after establishing an averaged RR, RF, or
CF, or a calibration curve  is equivalent to performing a single-point calibration. This updating procedure,
which is sometimes termed "continuing calibration," is unacceptable  and shall not be used because it
nullifies the statistical power of the full calibration.

9.2.3    Absolute and Relative Retention Time Precision

       Absolute retention time (RT) and relative retention time (RRT) are the QC criteria used  in
chromatographic analyses to aid in the identification of each detected analyte and to confirm that sufficient
time was allowed for the chromatographic separation of the analytes  in complex mixtures. These criteria
also  prevent laboratories from accelerating the analysis in an effort to reduce costs, only to find that
complex mixtures cannot be adequately resolved.

       A minimum RT specification is developed for those methods in which a minimum analysis time
must be established to ensure separation of the analytes in complex mixtures including known or expected
interferences.  An RT precision specification is developed for identification of an analyte by external
standard measurements, and an RRT precision specification is developed for (1) each analyte relative to its
labeled analog by isotope  dilution measurements, (2) each labeled compound relative to its internal
standard for isotope dilution measurements, and (3) each analyte relative to an internal standard for internal
standard measurements.

9.2.4    Initial Precision and Recovery

       The initial precision and recovery (IPR) test, also termed a "startup test," is used for initial
demonstration of a laboratory's capability to produce results that are at  least as precise and accurate as
results from practice of the method by other laboratories.  The IPR test also is used to demonstrate that a
method modification will produce results that are as precise and accurate as the reference method. The IPR
test consists of analyzing at least four replicate aliquots of a reference matrix spiked with the analytes of
interest and with either surrogate compounds or, for isotope dilution analysis,  labeled compounds. The
concentration of the target analytes in the spike solution may vary between one and five times the
concentration used to establish the lowest calibration point (e.g., one to five times the ML). The spiked
                                                                              March 22, 1999
                                               34

-------
aliquots are carried through the entire analytical process.  The IPR test is performed by the laboratory
before it utilizes a method for analysis of actual field samples. Specifications are developed for the
permissible range of recovery for each analyte and for an upper limit on the standard deviation or RSD of
recovery.

9.2.5  Ongoing Precision and Recovery

       The ongoing precision and recovery (OPR) test, sometimes termed a "laboratory control sample,"
"quality control check sample," or "laboratory-fortified blank," is used to ensure that the laboratory remains
in control during the period that samples are analyzed, and it separates laboratory performance from
method performance in the sample matrix. The test consists of a single aliquot of reference matrix spiked
with the analyte(s) of interest and carried through the entire analytical process with each batch of samples.
Typically, the concentration of the target analyte(s) in the same as the concentration used in the IPR test.
Specifications are developed for the permissible range of recovery for each analyte.

9.2.6  Analysis of Blanks

       Blanks are analyzed either periodically or with each sample batch to demonstrate that no
contamination is present that would affect the analysis of standards and samples for the analytes of interest.
The period or batch size is defined in each method. Typical periods and batch sizes are one per shift or one
for every 10 or 20 samples, but more or fewer may be required, depending upon the likelihood of
contamination.

       For most methods, QC acceptance criteria  for blanks are given in each method and are specified as
the concentration or amount of the  analyte allowed in each type of blank. The source of contamination in a
blank must be identified and eliminated before the  analysis  of standards and samples may begin. Samples
analyzed with an associated contaminated blank must be reanalyzed.  Methods for which blank
contamination cannot be eliminated should use a "y = mx + b" calibration model.

9.2.7  Surrogate or Labeled Compound Recovery

       The surrogate or labeled compound recovery is used to assess the performance of the method on
each sample. For this test, surrogates or stable, isotopically labeled analogs of the analytes of interest are
spiked into the sample and the recovery is calculated.  Specifications are developed for the permissible
range of recovery for each surrogate and/or labeled compound from each sample.

9.2.8  Matrix Spike and Matrix Spike Duplicate

       The matrix spike and matrix spike duplicate (MS/MSD) test is used in non-isotope dilution
methods to assess method performance in the sample matrix.  In most cases, analytes of interest are added
to a field sample aliquot that is then thoroughly homogenized and split into two  spiked replicate aliquots.2
2 For analytes, such as oil and grease, that adhere to container walls and cannot be adequately
homogenized, it is not possible to divide a spiked aliquot into two replicate aliquots. In these cases, two
field samples are collected and each field sample is spiked with identical concentrations of the analytes of
interest to produce an MS and MSB sample.


                                                                            March 22, 1999
                                             35

-------
One of these replicates is identified as the matrix spike sample and the other is identified as the matrix spike
duplicate sample. The recoveries of the analytes, relative to the spike, are determined in each sample. The
precision of the determinations also is assessed by measuring the relative percent difference (RPD) between
the analyte concentrations measured in the MS and MSB. The MS and MSB should each be spiked at a
level that results in the concentration of the target analyte(s) being

•      At the regulatory compliance limit
•      One to five times the background concentration of unspiked field sample, or
•      At the level specified in the method, whichever is greater.

If the background concentration in the field sample is so high that the spike will cause the calibration range
of the analytical system to be exceeded, the sample is spiked after the field sample is diluted by the minimal
amount necessary for this exceedance not to occur. This dilution of the sample to stay within the
calibration range of the analytical system for the target analyte is necessary to verify that the sample matrix
has not prevented reliable determination of the analyte. Specifications are developed for the permissible
range of recovery and RPB for each analyte.

9.2.9   Demonstration of Method Detection Limit

       Nearly all of the 40 CFR part  136, Appendix A, methods contain method detection limits (MBLs),
although few of the methods explicitly require laboratories to demonstrate their ability to achieve these
MBLs. Each laboratory that intends to practice a new method will be required to demonstrate achievement
of an MBL that meets acceptable criteria.  The MBL must be determined according to the procedures
specified at 40 CFR part 136, Appendix B.  The Appendix B, MBL calculation and analytical procedure is
described in Section 9.3.1.1.

9.2.10 Reference Sample Analysis

       The most common reference sample is a Standard Reference Material (SRM) from the National
Institute of Standards  and Technology (NIST). The reference sample and the period for its use are
specified in each method. EPA is considering setting acceptance criteria for standard reference materials to
be within  some percentage of the stated value based on the variability of measurement for that analyte.
One possible  indicator of that variability is the relative standard deviation calculation for the initial
precision and recovery samples. Corrective action to be taken when the acceptance criteria are not met
should involve identifying the samples affected, determining the amount of the effect, and if the effect is
significant, determining the impact of the effect on the environmental samples analyzed.

9.3    Development of Quality Control Acceptance Criteria

       The procedures for developing QC acceptance criteria for Tier 1, Tier 2, and Tier 3 methods are
described in Sections 9.3.1, 9.3.2, and 9.3.3, respectively.  Interlaboratory study data are required to
develop QC acceptance criteria for Tier 2 and Tier 3 methods. Although these studies are not necessary for
Tier 1 methods, interlaboratory study data may be available.  If interlaboratory data are  available for a Tier
1 method, these data should be used to develop QC acceptance criteria for Tier 1 methods by following the
Tier 2 or Tier 3 procedures described in Section 9.3.2 or 9.3.3, respectively. Where possible,
interlaboratory study data used for development of QC acceptance criteria should be derived from study
designs that follow the basic principles outlined in this protocol, Guidelines for Collaborative Study
                                                                             March 22, 1999
                                              36

-------
Procedures to Validate Characteristics of a Method of Analysis, JAOAC 72 No. 4, 1989, Use of
Statistics to Develop and Evaluate Analytical Methods (published by AOAC-International), ASTM
Standard D-2777 (published by ASTM), or other well-established and documented principles.

       The statistical procedures described in Sections 9.3.2 and 9.3.3 for Tier 2 and Tier 3 are based on
the use of interlaboratory multipliers.  These multipliers were derived from a comparison of intralaboratory
versus interlaboratory variability in the development of EPA Method 1625.3  The variation in the
interlaboratory multiplier used is directly related to the number of laboratories used at each of the two tiers.
The general relationship follows the concept that an increase in the number of laboratories used results in a
decrease in the interlaboratory multiplier.

       If the method being developed is applicable to a large number of compounds, the organization
responsible for developing QC acceptance criteria for the method may wish to consider the use of statistical
allowances for simultaneous compound testing.  Procedures associated with simultaneous compound testing
and the development of applicable QC acceptance criteria can be found at 49 FR 43242 and in the Method
1625 validation study report.4

9.3.1  Quality Control Acceptance Criteria Development for New Methods at Tier 1

       Method validation at Tier 1 consists of (1) using the new method to perform an MDL study in
accordance with the procedure described at 40 CFR part 136, Appendix B, (2) using the results of this
MDL study to establish an ML, and (3) running, in a single laboratory, a test of four spiked reference
matrix samples and four spiked samples of the matrix type(s) to which the method is to be applied. The
spike level  of these reference matrix and real-world matrix IPR samples must be in the range of one to five
times the ML, or at the regulatory compliance level, whichever is higher.

9.3.1.1  Method detection limit and minimum level

       An MDL must be determined for each target analyte using the procedure detailed at 40 CFR part
136, Appendix B.  This procedure involves spiking seven replicate aliquots of reference matrix or the
sample matrix with the analytes of interest at a concentration within one to five times the estimated MDL.
The seven aliquots are then carried through the entire analytical process, and the standard deviation of the
seven replicate determinations is calculated.  The standard deviation is multiplied by 3.14 (the Student's t
value at 6 degrees of freedom) to form the MDL. If the spike level is greater than five times the determined
MDL, the spike level must be reduced and the test repeated until the MDL is within a factor of five of the
spike level. The precautions concerning blanks and the effect of the matrix, and the detailed steps in 40
CFR part 136, Appendix B must be observed to  arrive  at a reliable  MDL.  In addition, if the analytical
system or instrument fails to produce a positive response for any of the seven replicates (i.e., produces a
zero or negative result), the MDL procedure must be repeated at a higher spike level. To assure that the
MDL is reliable, the optional interative procedure in Step 7 of the MDL procedure must be performed, the
3 Appendix I, "Estimation of Variance Components", of the Interlaboratory Validation of U.S.
Environmental Protection Agency Method 1625A, available from the EPA Sample Control Center
operated by DynCorp, Alexandria, VA 22314, 703/519-1140.

4Interlaboratory Validation of U.S. Environmental Protection Agency Method 1625A. See above.
                                                                             March 22, 1999
                                              37

-------
F-ratio test criteria met, and the pooled MDL from the two levels must be the MDL specified in the
method.

       The ML is established by multiplying the MDL by 3.18 and rounding to the number nearest to  (1,
2, or 5) x 10", where n is positive or negative integer.  The purpose of rounding is to allow instrument
calibration at a concentration equivalent to the ML without the use of unwieldy numbers. The use of 3.18
results in an overall standard deviation multiplier of 10, which is consistent with the American Chemical
Society's (ACS) limit of quantitation (LOQ) (P.S. Porter et al., Environ. Sci. TechnoL, 22, 1988).

9.3.1.2 Calibration linearity

       Once the ML is established, the instrument or analytical system is then calibrated at the ML and a
minimum of two additional points to calculate an initial RSDCAL for the response factor and to determine
the number of points required for subsequent calibrations. The highest point should be at, but not exceed,
the upper end of the analytical range for the instrument and the remaining point should be mid way between
the ML and highest point on a logarithmic scale.  For example, if the ML is 1.0 and the highest point is
100, the mid-point is 10. If the initial RSDCAL is < 2%, a one- or two-point calibration can be used (see
Section 9.2.1.2) and it is unnecessary to establish a limit for calibration linearity.

       If three or more calibration points are required, the maximum allowable RSDCAL (RSDCAL max) for
the RFs, CFs or RRs is determined as follows:

(1)     Determine the average response factor (RF), calibration factor (CF), or relative response (RR)
       for each analyte from the initial calibration:

                                RF = (RF, + RF2 4- ... + RFJ/rc

       where n is the number of calibration points.

(2)     Determine the RSDCAL using RF, CF,or RR and the standard deviation (s) of the RF, CF, or RR
       for each analyte from the initial calibration.  The RSDCAL is determined by:

                                     RSDCAL = 100s/(RF)

(3)     Develop RSDCALmax as follows:

                            RSDCAL,max = Minimum (35%,  *RSDCAL)

       where k is the square root of the 95th percentile of an F distribution with degrees of freedom
       corresponding to the number of points in the initial calibration minus 1 in both numerator and
       denominator. For a three point calibration, the value of & is 4.4, and for a five-point calibration,
       the value of k is 2.5. The maximum allowable specification for RSDCALmax is 35%.

       Note: In the above equations, the RF and RF terms should be replaced by CF  and CF or  RR and
       RR terms where appropriate.

9.3.1.3 Calibration verification
                                                                            March 22, 1999
                                             38

-------
Calibration verification criteria are specified as allowable percentage deviations from the response factor
(RF), calibration factor (CF), or relative response (RR) obtained from the initial calibration.  The upper
and lower QC acceptance criteria for the calibration verification as follows:

(1)    Calculate a multiplier, k, as the 97.5th percentile of a Student's / distribution with n - 1 degrees of
       freedom times the square root of (1 + \in), where there are n points in the calibration. For a three
       point calibration, the n - 1 Student's t value is 4.3, and for a five point calibration, the Student's t
       value is 2.8, resulting in values for k of 5.0 for a three point and 3.0 for a five point calibration.

(2)    Calculate the upper and lower QC acceptance criteria for the response or calibration factors for
       each analyte by developing a window around the average response factor found in the initial
       calibration by:
                                                    RF-A:s
                                  Lower limit (%) =  —=—
                                                      RF
                                                     RF+&S
                                  Upper limit (%) =
                                                       RF
       where k is the multiplier determined in Step 1 and s is the standard deviation determined in 9.3.1.2,
       Step 2.

       Note:  In the above equations, the RF terms should be replaced by CF or RR terms where
       appropriate.

9.3.1.4 Initial and ongoing precision and recovery

       For Tier 1 methods, an IPR test must be performed in both a reference matrix (usually, reagent
water) and the sample matrix of interest. Results of the reference matrix IPR tests are used to generate QC
acceptance criteria for IPR and OPR tests as described in this subsection. Results of the sample matrix
IPR test are used to develop QC acceptance criteria for the MS/MSD tests (see Section 9.4.1.5 below).
The reference matrix IPR test is performed by analyzing four aliquots of the reference  matrix spiked with
the target analyte(s) at the concentration determined in Section 9.2.4.

       Calculate the QC acceptance criteria for the IPR and OPR tests using results of the test of the
reference matrix per the following steps:

(1)    Calculate the average percent recovery (X), the standard deviation of recovery (s), and the relative
       standard deviation (RSDIPR=100s/X) of the four IPR results.

(2)    IPR QC acceptance criterion for precision - To approximate a 95% confidence interval for
       precision, the RSD1PR is multiplied by the square root of the 95th percentile of an F distribution
       with n - 1 degrees of freedom in the numerator and the denominator, where n is the number of IPR
       data points. The resulting multiplier on RSD,PR for four data points will then be 3.0, and the QC
       acceptance criterion for precision in the IPR test (RSDIPRjn)ax)  is calculated as follows:

                                    RSD1PR,max = 3.0RSD1PR.
                                                                              March 22, 1999
                                              39

-------
(3)     IPR QC acceptance criteria for recovery - Calculate the QC acceptance criteria for recovery in the
       IPR test by constructing a window around the average percent recovery (X). This factor comes
       from the 97.5th percentile of the t-distribution for n - 1 degrees of freedom, multiplied by
       ^1.15(1+1)+(1/4 + II n), where n is the number of IPR data points, to account for
       interlaboratory variability and the estimation of the mean. For four data points, this simplifies to
       5.3s, and the limits are as follows:
                                  Lower limit (%) = X - 5.3s
                                  Upper limit (%)  =X + 5.3s

       (Based on EPA's interlaboratory validation study of Method 1625, the additional variance due to
interlaboratory variability is estimated as  1.15s2.)

(4)     OPR QC acceptance criteria for recovery - A similar multiplier is used as for the IPR test but the
       second factor is ^1.15(1+1)  +  (1 + 1/w), so the multiplier is 6.0 for 4 IPR data points.
       Calculate the QC acceptance criteria for recovery in the OPR test by constructing a window
       around the average percent recovery X.  For 4 IPR data points, the limits would be:

                                  Lower limit (%) = X - 6.0s
                                  Upper limit (%)  = X + 6.0s

       Note: For highly variable methods, it is possible that the lower limit for recovery for both the IPR
       and OPR analyses will be a negative  number. In these instances, the data should either be log-
       transformed and the recovery window recalculated, or the lower limit established as "detected," as
       was done with some of the 40 CFR part 136, Appendix A, methods (49 FR 43234).

9.3.1.5 Matrix spike and matrix spike  duplicate

       As noted above, an IPR test must be performed in both an appropriate reference matrix and the
sample matrix of interest for Tier 1 new methods. The results of the sample matrix IPR test are used to
develop acceptance criteria MS/MSD analyses.  Sample matrix IPR tests are performed by:  (1)
determining the background concentration of the sample matrix, (2) spiking four replicate aliquots of the
sample matrix at a concentration equal to the regulatory compliance limit, one to five times the ML
determined in Section 9.3.1.1, or one to five times the background concentration of the sample, whichever
is greater, and (3) analyzing each of these spiked replicate samples.

       Calculate the QC acceptance criteria for the recovery of MS and MSB samples as follows:

(1)     Calculate the average percent recovery (X) and the standard deviation of recovery (s) of each target
       analyte in the sample matrix IPR aliquots.

(2)     Calculate the QC acceptance criteria for recovery in the MS and MSB tests by constructing a ฑ
       6.0s window (assuming 4 IPR aliquots) around the average percent recovery (X) (derived the same
       as for the OPR test above):
                                  Upper limit (%)  = X + 6.0s
                                                                              March 22, 1999
                                              40

-------
       Note: For highly variable methods, it is possible that the lower limit for recovery for both IPR and
       OPR analysis will be a negative number. In these instances, the data should either be log-
       transformed and the recovery window recalculated, or the lower limit established as "detected," as
       was done with some of the 40 CFR part 136, Appendix A, methods (49 FR 43234).

       Calculate the QC acceptance criteria  for the relative percent difference between the MS and MSB
as follows:

(1)     Calculate the relative standard deviation (RSD) of the recoveries of each target analyte in the
       sample matrix IPR aliquots as follows:

                                       RSD]PR= lOOs/X

(2)     Calculate the maximum allowable relative percent difference (RPDmax) by multiplying the RSDIPR
       by ^2 times the square root of the 95th percentile of an F distribution with 1  and n - 1 degrees of
       freedom, where n is the number of IPR data points. For 4 IPR data points, the calculation
       simplifies to:

                                      RPDmax = 4.5RSD

9.3.1.6 Absolute and relative retention time
       Determine the average retention time, RT (and/or average relative retention time,  RRT), and the
standard deviation (s) for each analyte and standard. Determine the upper and lower retention time (or
relative retention time) limits using the following:

                                 Lower limit = RT-tsJ 1 + —
                                 Upper limit = RT + ts
       The relative retention time upper and lower limits are determined by replacing RT with RRT in
the equations above.  The t value is the 97.5th percentile of a t distribution with n - 1 degrees of freedom,
where n is the number of retention time or relative retention time values used.

9.3.17 Blanks

       Establish the QC acceptance criteria for blanks.  The usual requirement is that the concentration of
an analyte in a blank must be below the ML or below one-third (1/3) the  regulatory compliance level,
whichever is higher.  In instances where the level of the blank is close to  the regulatory compliance level or
the level at which measurements are to be made, it may be necessary to require multiple blank
measurements and establish the QC acceptance criteria based on the average  of the blank measurements
plus two standard deviations of the blank measurements.
                                                                            March 22, 1999
                                             41

-------
9.3.1.8 Reference sample

       Establish the QC acceptance criteria for the reference sample based on the error provided with the
reference sample.
9.3.2  Quality Control Acceptance Criteria Development for New Methods at Tier 2

       Method validation at Tier 2 consists of running tests on a single matrix type collected from three
different facilities in the same industrial subcategory, with the sample being analyzed in three separate
laboratories (see 40 CFR parts 405 - 503 for industrial categories and subcategories). If the matrix type
being validated is drinking water, then tests shall be run on a drinking water matrix collected from three
different sources or on three drinking water samples that each have different characteristics.

       Each of the three laboratories will need to run a full suite of tests, beginning  with an MDL study to
determine the appropriate ML, followed by calibration, IPR, OPR, and blank analyses, along with a pair of
MS/MSD  analyses for each sample matrix.  Results from each laboratory will be submitted to the
organization responsible for developing the method. That organization will use the laboratory results to
develop QC acceptance criteria as described in the following subsections.

9.3.2.7 Method detection limit and minimum level

       Each laboratory participating in the MDL study must perform an MDL test as described in Section
9.2.9  The organization responsible for developing  the new method must establish an MDL for the method,
using a pooled MDL from the three laboratories. The precautions concerning blanks and the effect of the
matrix, and the detailed steps in 40 CFR part 136, Appendix B  must be observed to arrive at a reliable
MDL.

       A  pooled MDL is calculated from m individual laboratory MDLs by comparing the square root of
the mean of the squares of the individual MDLs and multiplying the result by a ratio  of t-values to adjust
for the increased degrees of freedom.
MDT
iviui.pooled
d (MDL(Labl)o d (MDL(Lab2))2 , d (MDL(Labm))2
(0.99,d,) (0.99,d2) (0.99,dm)
d1+d2 + ...dm
t(0.99,d,+d2 + ...dm
where m = the number of laboratories, and d, = the number of replicates used by lab I to derive the MDL.
In the case of 3 laboratories with 7 replicates per laboratory, the equation simplifies to:
MDL
pooled \
MDL(2Labl)-

FMDL(2Lab2)
3
+ MDL(2Lab3)

2.55
3.14
                                                                             March 22, 1999
                                              42

-------
       The organization responsible for developing the method also must use this pooled MDL to develop
an ML.  Procedures for determining the ML are given in Section 9.3.1.1.

9.3.2.2 Calibration linearity

       Once the ML is established, the instrument or analytical system is then calibrated at the ML and a
minimum of two additional points to calculate an initial RSD for the response factor and to determine the
number of points required for subsequent calibrations. (Section 9.2.1.2). If the initial RSD is < 2%, a one-
or two-point calibration can be used (see Section 9.2.1.2) and it is unnecessary to establish a limit for
calibration linearity.

       If three or more calibration points are required, the upper limit on the RSD of the RFs ,CFs, or
RRs is determined as follows:

(1)    Calculate the mean and standard deviation of the RFs, CFs or RRs for each laboratory and analyte.

(2)    Calculate the relative standard deviation of the RF, CF, or RR of each laboratory and analyte as:

                                                100s
                                        RSD. = -
                                                RF.

          where s, and RF. are the standard deviation and mean of the RFs for laboratory ;'.

(3)    Calculate the pooled RSD of the RF, CF or RR for each analyte from all laboratories. The pooled
       RSD is calculated as the square root of the mean of the squares of the sample RSDs from each
       individual laboratory. For example, for three laboratories, the pooled RSD is calculated as:
                                       RSD,2  + RSD,2  +  RSD 2

(4)    Calculate the maximum RSDCAL of the RF, CF, or RR for each analyte as follows:

                            RSDCAUmax = Minimum(35%, ฃRSDpool)

       where k is the square root of the 95th percentile of an F distribution with n - 1 degrees of freedom
       in the numerator and m(n - 1) degrees of freedom in the denominator, where m is the number of
       laboratories and n is the number of calibration points. For three laboratories using a three point
       calibration, (m = 3, n = 3), the value of k is 2.3, and for three laboratories using a five point
       calibration (m = 3, n = 5), the value of A: is 1.8.  The maximum allowable specification for
       RSDCAUmax is 35%.

       Note: In the above equations, the RFand RF terms should be replaced by CF and CF or RR and
       RR terms where appropriate.
                                                                            March 22, 1999
                                             43

-------
9.3.2.3 Calibration verification

       The calibration verification criterion is stated as a maximum relative distance between the average
RF obtained by a future laboratory's initial calibration (RF,NIT) and the RF obtained from its calibration
verification standard (RFVER).The maximum allowable deviation is based on the pooled relative RSDpool
obtained from Section 9.3.2.2.

(1)    Determine kVER by multiplying the 97.5th percentile of a Student's t distribution with
        m(n — 1) degrees of freedom times the square root of (1+1/n), where there are n points in the
       calibration and m laboratories:
        For a three point calibration with three laboratories, the m (n - 1) Student's t value is 2.4, and for a
        five point calibration, the Student's / value is 2.2, resulting in combined multipliers of 2.8 for a
        three point calibration, and 2.4 for a five point calibration.

(2)     The calibration verification criterion for the new method would then be stated as the maximum
        relative distance as follows:
                              100
                              1UU*
        For example, if the calibration verification criterion, calculated as kVer * RSDpoob equals 1 7%,
        then the difference between the mean RF from the initial calibration and the RF from the cal ver
        sample must be less than or equal to 17% of the initial mean RF.

        Note:  In the above equations, the RF,N|T and RFVER terms should be replaced by CF,NIT and
        CFVER or RRVER  and  RRVER terms where appropriate.

9.3.2.4 Initial and ongoing precision and recovery

        For the IPR and OPR tests, QC acceptance criteria are calculated using the average percent
recovery and the standard deviation of recovery from the IPR tests on four aliquots of the reference matrix
and the OPR test of one aliquot of the reference matrix (for a total of five samples) in the three
laboratories, as follows:

( 1 )    Calculate the average percent recovery (X ) for each analyte based on all data points from all
       laboratories, the between-laboratory standard deviationjsb) of the_mean results for each of the three
       laboratories (standard deviation of the three lab means X(|ab „,  X,,ab 2), Xflab 3)), and the pooled
       within-laboratory standard deviation (sw). sw  is calculated as the square root of the mean of all
       within-laboratory variances. For example for 3 laboratories:
                                                                              March 22, 1999
                                               44

-------
                                                   2       2
                                           '(lab 1) +S(lab 2)   S(lab 3)
      Note:  the organization responsible for developing the method must ensure that all laboratories are
      spiking IPR and OPR samples at the same concentration.

(2)    IPR QC acceptance criterion for precision - To calculate a 95% confidence interval for precision,
      the RSD (computed as sw divided by X) is multiplied by the square root of a 95th percentile F
      value with 3 degrees of freedom in the numerator and m(n - 1 ) degrees of freedom in the
      denominator, where m = the number of laboratories, and n is the number of data points per
      laboratory. For example, the  resulting multiplier on the RSD for three laboratories and five data
      points per laboratory will then be 1.9, and the QC acceptance criterion for precision in the IPR test
      (RSDraax) is calculated as follows:
(3)    IPR QC acceptance criteria for recovery - Calculate the combined standard deviation for
       interlaboratory variability and estimation of the mean (sc) as:
                                ^
      where m = the number of laboratories, and n = the number of data points per laboratory.  For 3
      laboratories and 5 data points per laboratory,
(4)     Calculate the QC acceptance criteria for_ recovery in the IPR test by constructing a ฑ 3.2 sc window
       around the average percent recovery (X, where 3.2 is the 97.5th percentile Student's t value for 3
       degrees of freedom (an estimated degrees of freedom based on the variance ratios observed with
       EPA Method 1625):


                                  Lower limit(%) =X -3.2sc
                                  Upper limit(%)=X + 3.2s
                                                                             March 22, 1999
                                              45

-------
      If more than 3 laboratories are used, the degrees of freedom for t will increase, but a complete
      calculation is beyond the scope of this document. An approximation of degrees of freedom equal to
      the number of laboratories will serve for most situations.

(5)    OPR QC acceptance criteria for recovery - Calculate the combined standard deviation for
      interlaboratory variability and estimation of the mean (sc) as:
       where m = the number of laboratories, and n = the number of data points per laboratory.  For 3
       laboratories and 5 data points per laboratory,
(6)    Calculate the QC acceptance criteria for_ recovery in the OPR test by constructing a ฑ 2.6 sc window
       around the average percent recovery (X, where 2.6 is the 97.5th percentile Student's t value for 5
       degrees of freedom (an estimated degrees of freedom based on the variance ratios observed with
       EPA Method 1625):


                                  Lower limit(%) =X-2.6sc
                                  Upper limit(%)=X+2.6sc


       If more than 3 laboratories are used, the degrees of freedom for t will increase, but a complete
       calculation is beyond the scope of this document. An approximation of degrees of freedom equal to
       twice the number of laboratories will serve for most situations.

9.3.2.5     Matrix spike and matrix spike duplicate

       Results of the MS/MSD analyses performed in  the validation study are used to develop the
MS/MSD QC acceptance criteria for Tier 2.  Each laboratory will measure MS and MSD in one sample.
Calculate the MS and MSD performance criteria as follows.

(1)    Calculate the mean and sample standard deviation of the recoveries of each MS/MSD pair, and then
       compute the overall mean recovery (X), the between-laboratory standard deviation of the  3
       pairwise means (sb), and the pooled within-laboratory standard deviation (sw) for each  target analyte
       (see Section 9.3.2.4).

(2)    In order to allow for interlaboratory variability, calculate the combined standard deviation  (sc) for
       interlaboratory variability and estimation of the mean as:
                                                                             March 22, 1999
                                              46

-------
                                  HKK4-
      where m = the number of laboratories.  For three labs,
                                           4  2
                                     ••M"    ,


(3)    MS/MSD QC acceptance criteria for recovery - Calculate the QC acceptance criteria for reco_very
      in the MS/MSD test by constructing a ฑ 2.6sc window around the average percent recovery (X)
      using the combined standard deviation. This factor comes from a t value for an estimated 5 degrees
      of freedom (based on this experimental design and variance ratios observed in Method 1625):


                                 Lower limit(%) =X-2.6sc
                                 Upper limit(%)=X+2.6sc


      If more than 3 laboratories are used, the degrees of freedom for t will increase, but a complete
calculation is beyond the scope of this document. An approximation of degrees of freedom equal to the
number of laboratories plus 2 will serve for most situations.

      Note: For highly variable methods, it is possible that the lower limit for recovery will be a negative
      number.  In these instances, the data should either be log-transformed and the recovery window
      recalculated, or the lower limit established as "detected," as was done with some of the 40 CFR part
      136, Appendix A methods.

(4)    MS/MSD QC acceptance criteria for relative percent difference (RPD) - To evaluate a 95%
      confidence interval for precision, the RSD (computed using the pooled within laboratory standard
      deviation sw of the MS/MSD samples divided by X) is multiplied by the square root of the 95th
      percentile F value with 1 degrees of freedom in the numerator and m degrees of freedom in the
      denominator multiplied by y2, where m is the number laboratories. The resulting multiplier on the
      RSD for 3 laboratories will then be 4.5.  The QC acceptance criterion for precision in the
      MS/MSD test (RPDmax) is calculated as follows:


                                     RPD   =4.5RSD.
                                         max
9.3.2.6     Absolute and relative retention time

       Establishing QC acceptance criteria for RT and RRT precision is problematic when multiple
laboratories are involved because laboratories have a tendency to establish the chromatographic conditions
that suit their needs.  Calculating average RTs and RRTs based on different operating conditions will result
                                                                          March 22, 1999
                                            47

-------
in the establishment of erroneously wide windows.  It is advised, therefore, that the organization developing
the method specify to the participating laboratories the chromatographic conditions and columns to be used.
Any future laboratories operating under different conditions will need to develop new acceptance criteria
for RT and RRT precision.
       Determine the average retention time, RT, (or average relative retention time, RRT), and the
corresponding standard deviation (s) for each analyte and standard. Determine the upper and lower
retention time (or relative retention time) limits using the following:

                                Lower limit = RT -tsav .   1 + —


                                Upper limit = RT+ts   I  1 + —
                                                     t>\\     n
where the t value is the 97.5th percentile of a t distribution with n - 1 degrees of freedom and where n is the
number of retention time or relative retention time data values to be used.

9.3.2.7     Blanks

       Establish the QC acceptance criteria for blanks. The usual requirement is that the concentration of
an analyte in a blank must be below the ML or below one-third (1/3) the regulatory compliance level,
whichever is higher. In instances where the level of the blank is close to the regulatory compliance level or
the level at which measurements are to be made, it may be necessary to require multiple blank
measurements and establish the QC acceptance criteria based on the average of the blank measurements
plus two standard deviations of the blank measurements.

9.3.2.8     Reference sample

       Establish the QC acceptance criteria for the reference sample based on the error provided with the
reference sample.

9.3.3   Quality Control Acceptance Criteria Development for New Methods at Tier  3

       In Tier 3, a single sample collected from each of a minimum of nine industrial categories is
analyzed in nine separate laboratories (one sample analyzed by each laboratory). Because data gathered
from nine laboratories lends itself to the statistical procedures used for interlaboratory method validation
studies, the procedures suggested by ASTM and AOAC-International are particularly applicable and those
procedures are preferred for development of QC acceptance criteria.  However, QC acceptance criteria may
also be developed for the Tier 3 methods in ways that are analogous to development of these criteria at
Tiers 1 and 2, with minor modifications described below.
                                                                            March 22, 1999
                                             48

-------
9.3.3.1     Method detection limits and minimum levels

       Each laboratory participating in the validation must perform an MDL study as described in Section
9.3.1.1. The organization responsible for developing the new method must establish an MDL for the
method, using a pooled MDL from the nine laboratories. A pooled MDL is calculated from m individual
laboratory MDLs by computing the square root of the mean of the squares of the individual MDLs and
multiplying the result by a ratio of ^-values to adjust for the increased degrees of freedom.
I                      MDL,. M,V    ( MDL,, ,9,]2       [ MDL,, .  .
                           (Labl)   + j        
-------
       where s, and RF (. are the standard deviation and mean of the RFs for laboratory /.

(3)    Calculate the pooled RSD of the RF, CF or RR for each analyte from all laboratories. The pooled
       RSD is calculated as the square root of the mean of the squares of the sample RSDs from each
       individual laboratory. For example, for nine laboratories, the pooled RSD is calculated as:

RSD," + RSD/ + .

. + RSD/
9
                           RSDPOO|


(4)    Calculate the maximum RSD for each analyte by the following:

                            RSDCAUmax = Minimum (35%,kRSDpool),

       where k is the square root of the 95th percentile of an F distribution with n - 1 degrees of freedom in
       the numerator and m(n - 1) degrees of freedom in the denominator, where in is the number of
       laboratories and n is the number of calibration points.  For nine laboratories using a three-point
       calibration (n = 3), the value of A; is 1.9, and for nine laboratories using a five-point calibration (n =
       5), the value of k is 1.6.  The maximum allowable specification for RSDCALmax is 35%.

       Note: In the above equations,  the RF and RF terms should be replaced by CF and CF or RR and
       RR terms where appropriate.

9.3.3.3     Calibration Verification

       The calibration verification criterion is stated as a maximum relative distance between the average
RF obtained by a  future laboratory's initial calibration (RF,NIT) and the RF obtained from its calibration
verification standard (RFVER).The maximum allowable deviation is based on the pooled relative RSDpo0|
obtained from Section 9.3.3.2.

(1)    Determine kVER by multiplying the 97.5th percentile of a Student's t distribution with
       m(n — 1) degrees of freedom times the square root of (!+!/ซ), where there are n points in the
       calibration and m laboratories:
                                       k    =
                                        VER
For a three-point calibration with nine laboratories, the m(n - 1) Student's t value is 2.1 and for a five-point
calibration, the Student's t value is 2.0, resulting in combined multipliers of 2.4 for a three-point
calibration and 2.2 for a five-point calibration.

(2)    The calibration verification criterion for the new method would then be stated as the maximum
       relative distance as follows:
                                                                             March 22, 1999
                                              50

-------
       For example, if the calibration verification criterion, calculated as AVer * RSDpool, equals 17%, then
       the difference between the mean RF from the initial calibration and the RF from the cal ver sample
       must be less than or equal to 17% of the initial mean RF.

       Note: In the above equations, the RF]N1T and RFVER terms should be replaced by CF]N|T and CFVER
       or RRVER and RRVER terms where appropriate.

9.3.3.4     Initial and ongoing precision and recovery

       For the IPR and OPR tests, QC acceptance criteria are calculated using the average percent
recovery and the standard deviation of recovery from the IPR tests of four aliquots of the reference matrix
and the OPR test of one aliquot of the reference matrix (for a total of five samples) in nine laboratories.
The QC acceptance criteria are developed using the following steps:

(1)    Calculate the average percent recovery (X) for each analyte based  on all data points from all
       laboratories, the between-laboratory standard deviation (sb) of the_mean results  for each of the m
       laboratories (the standard deviation of the m laboratory averages X|ab p Xlab2,  ...., X|abm), and
       the pooled within-laboratory standard deviation (sw) (For calculation of sw, see Section 9.3.2.4).
       Note: the organization responsible for developing the method must ensure that  all laboratories are
       spiking IPR and OPR samples at the same concentration.

(2)    IPR QC acceptance criteria for precision - To calculate a 95% confidence interval for precision, the
       RSD (computed as sw divided byX) is multiplied by the square root of the 95th percentile F value
       with 3 degrees of freedom in the numerator and m(n - 1) degrees of freedom in  the denominator, and
       n  is the number of data points per laboratory. The resulting multiplier for nine laboratories will be
       1.7.  The QC acceptance criterion for precision in the IPR test (RSD1PRmax) for 9 laboratories and 5
       data points  per laboratory is calculated as follows:
(3)    IPR QC acceptance criteria for recovery -Calculate the combined standard deviation for
       interlaboratory variability and estimation of the mean (sc) as:
       where m = the number of laboratories, and n = the number of data points per laboratory.  For 9
       laboratories and 5 data points per laboratory,
                                                                              March 22, 1999
                                              51

-------
(4)    Calculate the QC acceptance criteria for_recovery in the IPR test by constructing a ฑ 2.3 sc window
      around the average percent recovery (X, where 2.3 is the 97.5th percentile Student's t value for 10
      degrees of freedom (an estimated degrees of freedom based on the variance ratios observed with
      EPA Method 1625):


                                  Lower  limit(%) =X-2.3sc
                                  Upper  limit(%)=X+2.3s
      If more than 9 laboratories are used, the degrees of freedom for t will increase, but a complete
      calculation is beyond the scope of this document. An approximation of degrees of freedom equal to
      the number of laboratories will serve for most situations.

(5)    OPR QC acceptance criteria for recovery - Calculate the combined standard deviation for
      interlaboratory variability and estimation of the mean (sc) as:
 M  2  (,   l\
— K  +  l —
m)     {    n
                                                          2
      where m = the number of laboratories, and n = the number of data points per laboratory. For 9
      laboratories and 5 data points per laboratory,
(6)    Calculate the QC acceptance criteria for_ recovery in the OPR test by constructing a ฑ2.1 sc window
      around the average percent recovery (X, where 2.1 is the 97.5th percentile Student's t value for 19
      degrees of freedom (an estimated degrees of freedom based on the variance ratios observed with
      EPA Method 1625):


                                  Lower  limit (%) = X - 2.1 sc
                                  Upper  limit(%)=X + 2.1s
      If more than 9 laboratories are used, the degrees of freedom for t will increase, but a complete
      calculation is beyond the scope of this document.  An approximation of degrees of freedom equal to
      twice the number of laboratories will serve for most situations.
                                                                            March 22, 1999
                                             52

-------
   9.3.3.5     Matrix spike and matrix spike duplicate

         Results of the MS/MSD analyses performed in the Tier 3 validation study are used to develop the
   MS/MSD QC acceptance criteria for Tier 3. Calculate the MS and MSD performance criteria as follows.

   (1)    Calculate the mean percent recovery over all labs (X) and the between-laboratory standard
         deviation (sb) of the mean results for each of the nine laboratories and also the pooled within-
         laboratory standard deviation (sw) for each target analyte using the MS and MSD analyses (see
         Section 9.3.3.4)

   (2)    In order to allow for interlaboratory variability, calculate the combined standard deviation (sc) for
         interlaboratory variability and estimation of the mean as:
         where m = the number of laboratories. For nine labs,
                                          o
                                             N
                                                10 2   1 2
                                         S =      -  •   -
   (3)    MS/MSD QC acceptance criteria for recovery - Calculate the QC acceptance criteria for reco_very
         in the MS/MSD test by constructing a ฑ 2.2 sc window around the average percent recovery X
         using the combined standard deviation. This factor comes from a t value for an estimated 11
         degrees of freedom (based on this experimental design and variance ratios observed in Method
         1625):


                                     Lower limit(%) =X-2.2sc
                                     Upper limit(%)=X + 2.2sc

         If more than 9 laboratories are used, the degrees of freedom for t will increase, but a complete
         calculation is beyond the scope of this document. An approximation of degrees of freedom equal to
         the number  of laboratories plus 2 will serve for most situations.

         Note: For highly variable methods, it is possible that the lower limit for recovery will be a negative
         number. In these instances, the data should either be log-transformed and the recovery window
         recalculated, or the lower limit established as "detected," as was done with some of the 40 CFR part
         136, Appendix A methods.

(4)    MS/MSD QC acceptance criterion for relative percent difference (RPD) - To calculate a 95% confidence
      interval for precision, the RSDJcomputed using the pooled within-laboratory standard deviation, sw, of the
      MS/MSD samples divided by X) is multiplied by the square root of the 95% percentile  F value with 1
                                                                                March 22, 1999
                                                 53

-------
       degree of freedom in the numerator and m degrees of freedom in the denominator multiplied by y2.  The
       resulting multiplier on the RSD for nine laboratories will be 3.2.  The QC acceptance criterion for
       precision in the MS/MSD test (RPDmax) is calculated as follows:

                                         RPDmax = 3.2RSD.

9.3.3.6     Absolute and relative retention time

       Establishing QC acceptance criteria for RT and RRT precision is problematic when multiple laboratories
are involved because laboratories have a tendency to establish the chromatographic conditions that suit their
needs. Calculating average RTs and RRTs based on different operating conditions will result in the establishment
of erroneously wide windows.  It is advised, therefore, that the organization developing the method specify to the
participating laboratories the chromatagraphic conditions and columns to be used. Any future laboratories
operating under different conditions will need to develop new acceptance criteria for RT and RRT precision.

(1)    Using replicate RT and/or RRT data, calculate the upper and lower QC acceptance criteria for each
       analyte using the procedures in the calibration verification test in section 9.4.1.3.

(2)    Determine the  average retention time,RT (or average relative retention time, RRT), and the corresponding
       standard deviation (s) for each analyte and standard. Determine the upper and lower retention time (or
       relative retention time) limits using the following:
                                                          i	
                                    Lower limit = RT - ts. I 1 + —
                                    Upper limit = RT+ts
       where the t value is the 97.5th percentile of a t distribution with n - 1 degrees of freedom, where n is the
       number of retention time or relative retention time data values to be used.

9.3.3.7      Blanks

       Establish the QC acceptance criteria for blanks.  The usual requirement is that the concentration of an
analyte in a blank must be below the ML or below one-third (1/3) the regulatory compliance level, whichever is
higher. In instances where the level of the blank is close to the regulatory compliance level or the level at which
measurements are to be made, it may be necessary to require multiple blank measurements and establish the QC
acceptance criteria based on the average of the blank measurements plus two standard deviations of the blank
measurements.

9.3.3.8      Reference sample

       Establish the QC acceptance criteria for the reference sample based on the error provided with the
reference sample.
                                                                                March 22, 1999
                                                 54

-------
10.0  APPENDIX E - Data Reporting Form

       This appendix provides an example data reporting form. The form illustrates those aspects of data
reporting which are expected, regardless of the specific format used; specifically, data should be
presented in a clear and logical format, and should be labeled clearly.

       In addition to using an appropriate data reporting format, submitting electronic versions of data can
be very helpful in expediting the review of an ATP.  Data files should be in IBM-PC compatible format,
suitable for input directly into statistical analysis software, such as the Trimmed Spearman-Karber, Probit,
Dunnett, and ICP programs.
                                                                              March 22, 1999
                                              55

-------
                                    New Method Data Form1
I Method Title*  ||
                                (Revision Date   I     /   /_    I
                                             =^i===^=i
Include Method Number and Revision Number

Please record all data and quality control (QC) acceptance criteria from your validation study using this form. If
you have additional data, please attach it to this form in a tabular format, being sure to label all columns and
rows clearly.

For Tier 1 Studies (Single Laboratory Use): Complete 1 form for each matrix type.
For Tier 2 (Nationwide Use; Single Matrix) or Tier 3 (Nationwide Use; Multiple Matrices): Complete 1 form for each
participant laboratory.	__^	
Units of Concentration:
Linear Calibration Data

Units of Response:
Number of Points:
Analyte Cone.
Response
RF/CF/RR*





















'Response Factor/Calibration Factor/Relative Response
                                   Method Detection Limit (MDL) Data

Spiking Concentration used for MDL Study (include units):
| MDL Data |

'


	 1
                                  Initial Precision Recovery (IPR) Data

Spiking Concentration used for IPR Study (include units):.
| IPR Data
-1 	


	 1
                          Matrix Spike / Matrix Spike Duplicate (MS/MSD) Data

Spiking Concentration used for IPR Matrix Study (include units):
Tierl Tier 2 or 3
| IPR Matrix Data |

1

MS Concentration
MSD Concentration
Background Concentration



                                  New Method QC Acceptance Criteria
Calibration
Points

Lin

Spike
Cone

IPR RecoveryO and
Precision
Low

High

^Precision

OPR Recovery
Low

[high

MS/MSD Recovery and
RPD
Low || High

[_ RPD

MDL/ML
MDL || ML

T For multi-analyte methods, present additional Data and QC acceptance criteria for each analyte in a tabular format, making
 sure to include proper labels, and attach to this form.
                                                                                 March 22, 1999
                                                   56

-------