EPA/540/G-90/004
                                                  OSWER Directive 9360.4-01
                                                             April 1990
QUALITY ASSURANCE/QUALITY CONTROL GUIDANCE
                 FOR REMOVAL ACTIVITIES
                      SAMPLING QA/QC PLAN
                                and
                 DATA VALIDATION PROCEDURES
                            Interim Final
                       Environmental Response Team
                       Emergency Response Division
                   Office of Emergency and Remedial Response
                     U.S. Environmental Protection Agency
                         Washington, DC 20460

-------
                                            Disclaimer
This document has been reviewed in accordance with U.S. Environmental Protection Agency policy and.
approved for publication.  Mention of trade names or commercial products does not constitute endorsement
or recommendation for use.

-------
              Preface


the QA/QC Guidance
                                                       Removal
      Mr. William A. Coakley
 Removal Program QA Coordinator
         U.S. EPA - ERT
Raritan Depot - Building 18, MS-101
     2890 Woodbridge Avenue
      Edison, NJ 08837-3679
             111

-------
                    Quality Assurance/Quality Control Guidance for
                                    Removal Activities
                                    EPA Work Group
                                    EPA Headquarters

Office of Emergency and Remedial Response
                                       EPA Regional
Region 6:

Region 8:



Region 10:




ICF Kaiser Engineers




Roy F. Weston



Ecology and Environment
Other Organizations
                                   William Coakley
                                   Duane'Geuder
                                   Tom Kady
                                   Bob Forrest

                                   Rick Edmonds
                                   John Geidt
                                   Peter jstevenson

                                   Jim Everts
                                   Peter Mavraganis
                                   Regina Prevosto
                                   Joannj Sabak
                                   Joseph Tang
                                   Christine Andreas
                                   Kenda Layne
                                   John Matep

                                   Michael Bray
                                   Bill CJarberry
                                   Bob Marguccio
                                   Glenn Scherb
 In addition to the work group members, helpful suggestions and comments on the draft
 provided by the following as well as other EPA and contractor staff.

        Carla Dempsey (Hazardous Response Support Division, OERR)
        Anibal Diaz (Region 2 TAT, Roy F. Weston)
        Owen Douglass (Zone 1 ZPMO, Roy F. Weston)
        Donnissa Duvic (Region 4 TAT, Roy F. Weston)
        David Friedman (Office of Solid Waste)
        John Geidt (Region 8, Environmental Services Division)
        Jerry McKenna (Region 2, Environmental Services Division)
        Tun Ott (Region 1 TAT, Roy F. Weston)
        Diane Terry (ZPMO, Roy F. Weston)
        Donald Zelazny (Zone II ZPMO, Ecology and Environment)
                                       document were
                                              IV

-------
                                TABLE OF CONTENTS

                          PARTI:  SAMPLING QA/QC PLAN
  1.0
Introduction
         1.1
         1.2
         1.3
       Purpose
       Background
       Analytical Methods and Data Quality
 2.0     Elements of a Sampling QA/QC Plan
         2.1
         2.2
         2.3
         2.4

         2.5
         2.6
         2.7
         2.8
        2.9
        2.10
       Title Page
       Background
       Data Use Objectives
       Quality Assurance Objectives
       2.4.1   Methods
       Approach and Sampling Methodologies
       Project Organization and Responsibilities
       Quality Assurance Requirements
       Error Determination (Analytical and Total Error)
       2.8.1    Matrix Spike Samples
       2.8.2    Site Background Samples
       2.8.3    Site Action Level Samples (Total Error)
       Deliverables
       Data Validation
 1
 1
 3
 4
 4
 4
 4
 7
 7
9
9
9
10
10
10
                                 LIST OF TABLES
1      Example Proposed Schedule of Work

2      Field Sampling Summary

3      QA/QC Analysis and Objectives Summary
                                                          11

                                                          12

                                                          14

-------
                                      TABLE OF CONTENTS

                           PART II: DATA VALIDATION PROCEDURES
gection

1.0 Introduction

2.0 Data Validation Qualifiers

3.0 Metallic Inorganic Parameters

    3.1   Sample Holding Times                   .
    3.2   Initial and Continuing Calibration Verification
    3.3   Blanks
    3.4   ICP Interference Check Sample
     3.5   Error Determination
          3.5.1   Determination of Bias
                 3.5.1.1   Percent Recovery                _        .
                 3.5.1.2   Adjustment of Sample Value for Bias
          3.5.2   Determination of Precision
                 3.5.2.1   Replicate Analysis
                 3.5.2.2   Coefficient of Variation
     3.6  Performance Evaluation Samples
     3.7  Optional Additional Instrument QC (for elevated concentrations)
          3.7.1   ICP Serial Dilution
          3.7.2   Atomic Absorption Analysis Specific QC
     3.8  Overall Assessment of Data

 4.0 BNAs by GC/MS Analysis

     4.1  Sample Holding Times
     42   GC/MS Tuning Criteria                 .                 .
     4.3   Initial and Continuing Calibration Verification
           4.3.1   Internal Standards                            ,
      4.4   Error Determination
           4.4.1   Determination of Bias
                  4.4.1.1  Percent Recovery
                  4.4.1.2  Adjustment of Sample Value for Bias
           4.4.2   Determination of Precision
                  4.4.2.1  Replicate Analysis
                  4.4.2.2  Coefficient of Variation
       4.5  Blanks
       4.6  Compound Identification
           461   Tentatively Identified Compounds
       4.7   Compound Quantitation and Reported Detection Limits
       4.8   Performance Evaluation Samples
       4.9   Overall Assessment of Data
       4.10  Optional QC Checks
            4.10.1 Surrogate Recovery
 Page

   17

   17

   18

   18
   18
   18
   19
   19
   19
    19
    19
    19
    19
    20
.   '20
    21
    21
    21
    21

.',:,  21
•.,.;  21
' "  22
    23
    23
    23
    24
    24
     24
     24
     24
     24
     25
     26
     26
    .26
     27
     27
     27
     27
                                                   VI

-------
                                        TABLE OF CONTENTS (continued)
 Section

 5.0 VOAs
     5.1
     5.2
     5.3

     5.4
     5.5
     5.6

     5.7
     5.8
     5.9
     5.10
 Sample Holding Times
 GC/MS Tuning Criteria
 Initial and Continuing Calibration Verification
 5.3.1   Internal Standards
 Error Determination
 5.4.1   Determination of Bias
        5.4.1.1    Percent Recovery
        5.4.1.2   Adjustment of Sample Value for Bias
 5.4.2   Determination of Precision
        5.4.2.1   Replicate Analysis
        5.4.2.2   Coefficient of Variation
 Blanks
 Compound Identification
 5.6.1   Tentatively Identified Compounds
 Compound Quantitation and Reported Detection Limits
 Performance Evaluation Samples
 Overall Assessment of Data
 Optional QC Checks
 5.10.1 Surrogate Recovery
 6.0  Pesticides/PCBs
     6.1
     6.2
     6.3
     6.4
         6.4.2
    6.5
    6.6
    6.7
    6.8
    6.9
    6.10
 Sample Holding Times
 Instrument Performance
 Initial and Continuing Calibration Verification
 Error Determination
 6.4.1   Determination of Bias
       6.4.1.1   Percent Recovery
       6.4.1.2   Adjustment of Sample Value for Bias
       Determination of Precision
       6.4.2.1   Replicate Analysis
       6.4.2.2   Coefficient of Variation
 Blanks
 Compound Identification
 Compound Quantitation and Reported Detection Limits
Performance Evaluation Samples
Overall Assessment of Data
Optional QC Checks
6.10.1 Surrogate Recovery
7.0  PCBs
    7.1  Sample Holding Times
    7.2  Instrument Performance
    7.3  Initial and Continuing Calibration Verification
  27

  27
  28
  28
  29
  29
  29
  29
  29
  30
  30
  30'
  31
  31
  32
  32
  32
 32
 33
 33

 33

 33
 33
 34
 35
 35
 35
 35
 35
 35
 35
 36
 37
 37
 38
 38
 38
 38

 38

38
38
39
                                                vn

-------
                                      TABLE OF CONTENTS (continued)
Section

7.0 PCBs (continued)

    7.4  Error Determination
         7.4.1   Determination of Bias
                7.4.1.1  Percent Recovery
                7.4.1.2  Adjustment of Sample Value for Bias
         7.4.2   Determination of Precision
                7.4.2.1  Replicate Analysis
                7.4.2.2  Coefficient of Variation
    7.5  Blanks                             .
    7.6  Compound Identification                        .  .
    7.7  Compound Quantitation and Reported Detection Limits
    7.8  Performance Evaluation Samples
    7.9  Overall Assessment of Data
    7.10  Optional QC Checks
         7.10.1 Surrogate Recovery

 8.0  2,3,7,8-TCDD

     8.1   Sample Holding Times
     8.2   Instrument Performance
     8.3   Initial Calibration
     8.4   Continuing Calibration
     8.5   Error Determination
          8.5.1   Determination of Bias
                 8.5.1.1   Percent Recovery
                 8.5.1.2   Adjustment of Sample Value for Bias
          85.2   Determination of Precision
                 8.5.2.1   Replicate Analysis
                 8.5.2.2   Coefficient of Variation
     8.6  Blanks
     8.7  Internal Standard Requirements
     8.8  Identification of 2,3,7,8-TCDD
     8.9  Performance Evaluation Samples
     8.10 Overall Assessment of Data
     8.11 Optional QC Checks
          8.11.1 Surrogate Recovery

  9.0  Generic Data Validation Procedures

      9.1   GC Analyses (i.e., Herbicides, Organophosphate,
           Pesticides)
           9.1.1   Sample Holding Times
           9.1.2   Instrument Performance
           9.1.3   Initial and Continuing Calibration Verification
           9.1.4   Error  Determination
                  9.1.4.1   Determination of Bias
                          9.1.4.1.1   Percent Recovery
                          9.1.4.1.2   Adjustment of Sample Value for Bias
39
39
39
40
40
40
40
41
41
42
42
42
42
42

43

43
43
43
44
 44
 44
 44
 44
 45
 45
 45
 46
 46
 46
 47
 47
 47
 47

 47
  47
  47
  47
  48
  48
  48
  48
  49
                                                  vm

-------
                                   TABLE OF CONTENTS (continued)
9.2
9.0  Generic Data Validation Procedures (continued)

                9.1.4.2  Determination of Precision
                        9.1.42.1    Replicate Analysis
                        9.1.4.2.2    Coefficient of Variation
                Blanks
                Compound Identification
                Compound Quantitation and Reported Detection Limits
                Performance Evaluation Samples
                Overall Assessment of Data
         Non-Metallic Inorganic Parameters (i.e., anions, pH, TOC, nutrients)
         9.2.1   Sample Holding Times
         9.2.2   Initial and Continuing Calibration Verification
                Error Determination
                9.2.3.1   Determination of Bias
                        9.2.3.1.1   Percent Recovery
                        9.2.3.1.2   Adjustment of Sample Value for Bias
                        Determination of Precision
                        9.2.3.2.1   Replicate Analysis
                        9.2.3.2.2   Coefficient of Variation
                Blanks
                Compound Quantitation and Reported Detection Limits
                Performance Evaluation Samples
                Overall Assessment of Data
     9.1.5
     9.1.6
     9.1.7
     9.1.8
     9.1.9
     9.2.3
            9.2.3.2
     9.2.4
     9.2.5
     9.2.6
     9.2.7
                                                                                        Page
 49
 49
 49
 50
 50
 51
 51
 51
 51
 51
 52
 52
 52
 52
 52
 53
 53
53
54
54
54
54
                                           IX

-------
       PARTI
SAMPLING QA/QC PLAN

-------
  1.0 Introduction

  Part I provides a detailed description of each section to be
  contained in a "Sampling QA/QC Plan." The development
  of the Sampling QA/QC Plan is the responsibility of the
  On-Scene  Coordinator (OSC).  The  OSC reviews and
  approves the site-specific plan and may obtain assistance
  from the Regional QA Officer.  This  guidance will help
  ensure that reliable,  accurate, and   quality  data  are
  obtained through field sampling efforts  as well as field and
  laboratory  analytical  services.   The   document to  be
  produced from this  guidance is  neither  intended  to
  supersede nor replace the QA Project Plan; however, it is
  intended to augment  the project plan by detailing site-
  specific information regarding sampling, analysis, and QA
  protocols.

 Note:  QA/QC and QA are interchangeable terms used
 throughout the guidance document.
                             '      •'-  ?       ' •  '••" •
 1.1  Purpose

 The purpose of this document is to provide guidance in
 establishing,   implementing,   and   using   quality
 assurance/quality control  (QA/QC) protocols for data
 collection  activities  performed  under  the  Removal
 Program.

 1.2  Background

Agency policy requires that all EPA organizational units,
including program offices, EPA regional  offices, and EPA
laboratories,  that  perform   environmentally  related
measurements, participate hi a centrally managed quality
assurance (QA) program, as stated in the Administrator's
Memorandum of May 30, 1979. This requirement applies
to all environmental monitoring and measurement efforts
  mandated or supported by EPA through regulations, grants,
  contracts, or other formal means not currently covered by
  regulation. The responsibility for developing, coordinating,
  and directing the implementation of this program has been
  delegated to the  Office of  Research and Development
  (ORD),  which  has  established the  Quality Assurance
  Management Staff (QAMS) for this purpose.  As stated in
  EPA  Executive  Order  5360.1,  "Policy  and  Program
  Requirements  to   Implement   the   Mandatory  Quality
  Assurance Program," the primary goal of the QA program is
  to  ensure that  all  environmentally related measurements
  performed or supported by EPA produce data of known
  quality.   The  quality of  data  is  known  when  all
  components  associated  with  its  derivation  are
  thoroughly documented,  with,such documentation being
  verifiable and defensible^

 As  part of their participation  in  the Agency-wide  QA
 program, program offices are required to establish their own
 "QA ^Program Plan." ? This .plan is  to be prepared  and
 annually updated based on guidelines established  by QAMS.
 It specifies the quality of data required from  environmentally
 related measurements and provides sufficient  resources to
 assure  that an adequate level  of QA is performed.  The
 program plan is established at the Headquarters EPA level.
 For the Removal Program, the responsibility for the program
 plan lies with the  Office of  Emergency  and  Remedial
 Response  (OERR).   In addition to program  plans, plans
 need to be developed for each regional office.  These plans
 are  similar to the program plans, but are  tailored  to the
 specific operational  needs of  the regional office.   The
 program and regional plans are both broad in scope and
 merely provide the objectives and resources for undertaking
 environmentally-related measurements.

The  most specific element of QA documentation is the QA
Project Plan (see  Figure I).  A QA Project Plan specifies

-------
the policies, organization (where applicable), objectives,
functional activities, and  specific QA and QC activities
designed to achieve the data quality goals of a specific
projects) or continuing operation(s).  The QA Project
Plan is required for each specific  project or continuing
operation (or group  of  similar projects  or continuing
operation(s)).  Guidance for preparing  such  plans is
contained in "Guidelines and Specifications for Preparing
Quality Assurance Project Plans" (also known as QAMS-
005), which was developed by QAMS. This document
describes sixteen elements that must be  considered for
inclusion in all QA Project Plans.

To meet the requirement for a QA Project Plan in the
removal program, the Emergency Response Division of
OERR established a QA Workgroup to provide guidance.
The workgroup decided  that the QA Project Plan would
be divided into  two functional documents:  a generic
"Branch QA Project Plan," and a  site-specific "Sampling
 QA/QC Plan." When combined, both documents address
 the sixteen elements described in QAMS-005. The Branch
 QA  Project  Plan will  be  prepared  by  each  regional
 removal branch  and  will address only  those elements
 generic to all  activities occurring  within the Region; the.:
 Sampling QA/QC Plan will be  prepared  for each site
 where sampling will  be performed and address  those
 Clements specific to the site, such as sample collection and
 analysis. The Branch Plan should  be updated periodically
 to reflect any operational changes  in the Region.  The
 Sampling QA/QC Plan should be prepared for each site
 and updated (amended) when the scope of work changes
 significantly from the scope of  work described in any
  previous plan.  Elements that are  not addressed in the
  Sampling QA/QC Plan are included in the Branch Plan.
  For  emergency responses, a Sampling QA/QC Plan is
  required to be submitted no later than 30 days after the
  response date for documentation purposes.
                                   i
The  intent of this  document is  to  jprovide  guidance on
developing a site  specific  "Sampling! QA/QC Plan"  and
assessing and substantiating data for various data users.  The
guidance  is not intended  to addres|s field  and lab  QC
practices.  It is assumed and expected that field samplers
and analytical labs will follow approved methods (with their
inherent QC checks) and adhere to generally accepted "good
laboratory practices."
QA Program Plan (HQ
AS per QAMS-004/80 G


Regional QA Progra
(Regional Leve
As per QAMS-004/80


Generic QA Projec
(Branch Level
As per QAMS -005/80


Sampling QA/QC
(Site-Specif:
As per OSWER Dire
9360.4-01 Guid<
Level)
uidance

m Plan
1)
Guidance

,t Plan
)
Guidance

.Plan
-cj
active
ince
  Figure 1: EPA Quality Assurance Documentation
  This guidance has been designed to allow for the greatest
  possible variation in monitoring  str itegies.  However,  it is

-------
   recognized that occasionally certain  quality assurance
   requirements cannot be met. In such cases, the reason for
   the deviation should be stated in the Sampling QA/QC
   Plan along with the expected or observed impact on the
   data.

   13  Analytical Methods and Data Qualify

   The quality of data is  determined  by its accuracy and
   precision against prescribed requirements or specifications,
   and by its usefulness in  assisting the user to make a
   decision or answer a question with confidence.  The use of
   any  one  particular  analytical method. or  instrument,
  therefore, cannot determine  the quality of data obtained
  without an evaluation of the analytical accuracy (qualita-
  tive and quantitative) of the  data and  of the relevance
  (representativeness)  of the data to user needs. Likewise,
  certain analytical methods may provide more information
  than other, methods, but not necessarily better quality data.

 To illustrate,  a gas  chromatograph/mass spectrometer
 method  provides  more  information  than  a  gas
 chromatograph method,  which in turn provides more
 information than a spectrophotometermethod. ^However
 more information's not synonymous with accurate or
 useful data. Analytical quality is dependent on analytical
 accuracy; that is to say that there is a degree of confidence
 associated with the data. The term accuracy refers to both
 the  correctness  of the  concentration  value  and  the
 qualitative certainty that an analyte is  present.

 This guidance is based on the idea that the use of any one
 particular analytical method  or instrument  does not
 determine the quality of data  obtained.  This guidance
 prompts the data collector to define the data quality within
a framework that also incorporates the intended use of the
data.
   The guidance is structured around three quality assurance
   objectives.  Each quality assurance objective is associated
   with  a list of minimum requirements.   Therefore,  any
   method or analytical instrument that  can meet the quality
   requirements can be used for any one of the objectives.

   For example, if a spot test method was able to meet  the
  requirements for  QA3 (i.e.,  identify  the specific  analyte,
  determine the true concentration, and determine the error),
  then the spot test would not only be a valid method but it
  would give the same quality of data of a mass spectrometer
  (assuming the mass spectrometer method met all the QA3
  requirements).  It is anticipated that  QA1 and QA2 will
  satisfy most data  quality requirements  for the Removal
  Program.   QA3  is  expected to be used only in those cases
  where  an error  determination is needed to identify  false
  negative or false positive  values for critical decision level
  concentrations.
 2.0 Elements of a Sampling QA/QC Plan

 The,Sampling QA/QC:Plan/should contain.the following
 sections:                      ,     ,               , .

 •    Title page                         ,
 •    Background
     Data Use Objectives
 •    Quality assurance objectives
     Approach and sampling methodologies
     Project organization and responsibilities
     Quality assurance requirements
•    Deliverables
     Data validation

-------
2.1 Title Page

The title page should include the name of the site/project,
the contract and work order numbers (if the plan is being
prepared by contractors), the contractor name, the date,
key project personnel, and the approval signatures  of the
OSC and other  appropriate  persons. (Although it is
recommended that the QA Sampling Plan be reviewed by
Regional QA staff,  it is not necessary that the plan be
approved by the Regional QA officer.)

22  Background

This section  should provide a brief description  of the
events  or occurrences that led to the initiation  of the
sampling activity. This section may list chemicals which
 possibly  contributed to the  suspected  contamination,
 including the suspected  range  of contamination,  the
 sampling area size and proximity to local residents, or any
 other information that may be useful in an assessment of
 the situation  and  determination of QA, sampling, or
 analytical needs, possible contacts and existence of access
 agreements.  Sources of such data include inventories,
 manifests, or other records; prior sampling data, such as
 that generated by  an RI/FS; geological surveys;  and
 incidents of exposure.
 23  Data Use Objectives

 Before any sampling activity is conducted, the intended use
 of the data must be determined.  Careful consideration of
 intended data use is critical because it will affect the QA
 objective chosen and thereby maximize the probability of
 making a correct decision based on the analytical results.
 The decisions to be made, questions to be addressed, or
  both, should be listed in this section.
                                                         2.4  Quality Assurance Objectives
For each data collection activity, the QJA/QC objective must
be specified to correspond to the data ^ise objectives. Three
equally important QA/QC objectives have been defined for
assessing and substantiating the collection of data to support
its intended use. The three QA/QG  objectives, hereafter
referred to as QA1, QA2, and QA3, are described below.
Evaluate the characteristics of the following QA objectives
to determine which one or combination fits your data usage.
All  three  objectives provide  useful
enforcement   purposes,   disposal
responsible party identification,  and
The  QA  characteristics  are based
 and  valid  data  for
 and/or   treatment,
 cleanup  verification.
on  thp Agency QA
 objectives for  precision,  accuracy (both quantitative and
 qualitative), representativeness, compl ;teness, comparability,
 and detection level.

 QA1: Rationale for OA1 objective:

       QA1  is a  screening objective to afford a  quick,
       preliminary assessment of site  contamination.  This
       objective  for  data  quality  is available  for data
       collection.activities that  involve rapid,  non-rigorous
       methods of analysis and quality assurance.   These
       methods  are  used  to   makje quick,  preliminary
       assessments of types and  levels of pollutants.  The
       primary reason for this objeclve is  to  allow for the
       collection of the greatest amount of data with  the
       least expenditure of time and money.   The  user
       should be aware that data collected for this objective
       have neither definitive identification of pollutants nor
       definitive quantitation of their  concentration level.

       Although there is no quality assurance data collected
       with the data at this  objective, a  calibration or
        performance check of the method is required along

-------
 with verification of the detection level. Methods will
 be applied as per standard operating procedures and
 squipment manufacturer's specifications.
'••••'       ,         );

 The QA1 objective does not preclude the adherence
 to prescribed  quality control  checks given in EPA
 methods   and   SOPs   or   the   manufacturer's
 recommendations.  The QA1 objective is generally
 applied to but not limited to the following activities:
 physical  and/or  chemical  properties of samples;
 extent  and degree  of  contamination relative  to
 concentration  differences;  delineation of pollutant
 plume  in ground  water  (head space  or soil gas
 analysis techniques);  monitor well placement;  waste
 compatibility;   preliminary   health   and  safety
 assessment; hazardous categorization; and preliminary
 identification   and   quantitation   of   pollutants
 (determination   of  PH,   flammability,   chlorine
 presence, etc.).

  QA1 .Characteristics!

      Non-analyte  or analyte specific (may also
      be specific for a chemical class, i.e., PCBs,
      total hydrocarbons, total organic halides,
      total ionizable organics,  radiation).
      Non-definitive   (i.e.,   unconfirmed)
      identification;  non-qualitative  to semi-
      qualitative.
      Non-definitive  quantitation;  no  error
      determination   (no   precision   and
      accuracy determination).
      Representative, comparable, complete1.
 •    QA requirements  for  objective "QA1"
   .   are  specified in Section  2.7,  "Quality
      Assurance Requirements."
QA2: Rationale for OA2 obiecrivp.-

     QA2 is a verification obiective used to verify analytical
     (field or lab)  results.  A minimum of 10% verification
     of results is required.  This objective for data quality is
     available  for  data  collection activities  that  require
     qualitative and/or quantitative verification of a "select
     portion of sample findings" (10% or more) that were
     acquired using non-rigorous methods of analysis and
     quality assurance. This quality objective is intended to
    give the decision-maker (OSC) a level of confidence for
    a select portion of preliminary data.   This objective
    allows the OSC  to  focus on specific  pollutants and
    specific levels  of concentration quickly, by using field
    screening methods and verifying at least 10% by more
    rigorous analytical methods and quality assurance, The
    results of  the 10% of substantiated data gives  an
    associated sense of confidence for the remaining 90%.
    However, QA2 is not limited to only verifying screened
    data.  The QA2 objective is also applicable to data that
    are generated by any method which satisfies all the QA2
    requirements and thereby incorporates any one or a
    combination of the three verification requirements.

    Generally the methods used for verification are more
   rigorous,  as to  analytical  methodology and quality
   assurance.   Only  those verification methods that are
   analyte specific can  be  considered for this quality
   objective.   When  required,  the analytical  error  is
   determined  for all analytes that are of interest to the
   decision-maker (OSC) on at least 10% of samples.
                                                      Representative:  The degree to which sample data accurately and
                                                      precisely represent the characteristic of'the population. 'Comparable-
                                                      An evaluation of the similarity of conditions under which different set
                                                      of data are produced.  Complete:  The percentage of measurements
                                                      made which are judged to be valid.

-------
The QA2 objective is generally applied, but not

limited to the following activities: physical and/or

chemical properties of samples; extent and degree
of contamination; verification of pollutant plume

definition in ground water; verification of health
and safety  assessment;  verification  of  pollutant

identification; and verification of cleanup.


OA2 Characteristics:

•    Analyte specific (i.e., benzene, cyanide,
     2,3,7,8-TCDD, chromium).

     VERIFICATION  of  analyte  identity
     and/or concentration. Choose any one or
     any combination of the following three:

     1.  Definitive identification (choose one):
     Note:
        Except   for  X-ray  fluorescence
        (XRF),  confirmation  of  identity
        applies  to  organic analytes only.
        Confirm XRF determined analytes
        by an EPA-approved method.

    a.  Screened data  -  confirm   analyte
        identification  by an EPA-approved
        method, different from the screening
        method,  on   at  least  10%   of
        preliminary screened samples.
    b.  Unscreened data -  confirm  analyte
        identification  by an EPA-approved
        method    on   all  unscreened
        environmental samples (field or lab).

2.  Non-definitive quantitation (choose one):

    a.  Screened   data   -  verify   analyte
         concentration on  at  least  10% of
         preliminary screened samples (field
         or  lab)  using  an EPA-approved
         method, different from the screening
         method.
    b.   Unscreened data - determine analyte
         concentration  on  all  unscreened
         environmental samples (field or lab)
         using an EPA-approved method.

 3.   Definitive quantitation/analytical error
     (choose one):  Also, see Section 2.8 -
     Part I and Error Determination  - Part
     II.
                                                   Note:    Error determination is advised if data are
                                                            fr^ng pvalnafp.H against k critical action level.

                                                            a.   Screened data - determine the analytical
                                                                error  by  calculating  the   precision,
                                                                accuracy, and coefficient of variation for
                                                                a subset (at least [lO%) of the verified
                                                                data using an EPA-approved method.
                                                            b.   Unscreened   data  -  determine  the
                                                                analytical  error  I by calculating  the
                                                                precision, accuracy,  and coefficient  of
                                                                variation for allj.of the  quantitative
                                                                results using an EPA-approved method.

                                                   Note:    If  definitive  quantitation is chosen along
                                                            with definitive identification for all the data,
                                                            then your data meet tjie  QA3 objective.
                                                        Representative, comparable, complete.

                                                        QA  requirements  for  objective "QA2
                                                   are
           »•*:- —  — — T.—	
           specified in Section  2.7,
           Requirements."


QA3: Rationale for OA3  objective:


      QA3 is  a definitive objective

      accuracy of  the concentration
                                                                                  "Quality Assurance
                                                                                        used to assess the
                                                                                       level  as well as the
                                                          identity of the analyte(s) of interest. This objective
                                                          for  data quality is available  for  data collection

                                                          activities that require a high degree of qualitative and
                                                          quantitative accuracy of all findings using  rigorous
                                                          methods of analysis and quality assurance for "critical

                                                          samples" (i.e., those samples for which the  data are
                                                          considered essential in makirg a  decision).  This

                                                          quality  objective is intended to give the  decision
                                                          maker (OSC) a level of confidence for a select group

                                                          of "critical samples" so  he/she can  make a decision
                                                          based on an action level with regard to: treatment;

                                                          disposal;   site   remediation j and/or   removal  of
                                                          pollutants;  health  risk or  environmental  impact;
                                                           cleanup  verification; pollutant

                                                           delineation of contaminants;
                                    source identification;

                                    and other significant
                                                           decisions where an action level is of concern.  Only
                                                           those methods that are analyte specific can be used

-------
        for this quality objective.   Error determinations
        are made for. all analytes that are of interest to
        the  decision  maker  (OSC)  for  each  critical
        sample that is of interest.

        OA3 Characteristics:

        •    Analyte specific.
        •    Definitive   identification  -   confirm
            analyte  identification   by a  second
            method, such as mass spectroscopy, on
            100% of the "critical samples" collected;
            applies only to organic analytes.
            Note:    Except   for   X-ray  fluorescence
                     (XRF),   confirmation  of  identity
                     applies  to  organic analytes  only.
                     Confirm  XRF determined analytes
                     by an EPA-approved method.
            Non-definitive quantitation (choose one):
            a.   Screened   data   -  verify   analyte
                concentration   on   at  least   10%  of
                preliminary  screened samples  (field  or
                lab)  using  an EPA-approved method,
                different from the screening method.
            b.   Unscreened data  -  determine  analyte
                concentration   on   all - unscreened
                environmental samples (field or lab) using
                an EPA-approved method.
       •    Definitive quantitation/analytical error
            - (determine the analytical  error  by
            calculating the precision,  accuracy, and
            coefficient of variation) on 100% of the
            "critical samples"  collected using  an
            EPA-approved method.
           Representative, comparable, complete.
           QA requirements for objective "QA3" are
           specified   in   Section   2.7,   "Quality
           Assurance Requirements."
  provides.  Quality is a matter of degree and can only be
  assessed against specific criteria. Therefore, one can choose
  any analytical method to use for any one of the three quality
  assurance objectives in Section 2.4, provided all of the quality
  assurance  requirements are  met for  that objective  as
  specified in Section 2.7. The methods that can be used for
  any of these three objectives include, but are not limited to,
  spot  tests;  paper strip  tests; indicator tubes;  chemical
  reactions producing colors, gases, or precipitates; electronic
 meters such  as  Geiger counters, pH meters, conductivity
 meters; electronic detectors such as photoionization, electron
 capture, flame ionization, flame photometric, electrolytic, and
 infrared;  gas chromatography;  mass spectroscopy; atomic
 absorption; inductively coupled plasma (ICP),  and X-ray
 fluorescence.  These methods may respond to  either groups
 of analytes or specific analytes or  both.

 2.5  Approach and Sampling Methodologies

 This section should provide a description of the possible
 sample matrices, required  equipment  and  fabrication,
 sampling design (reference SOPs and EPA procedures used
 for collecting samples), sample documentation, corrective
 action, sample analyses, and a schedule of work  (see Table
 1).  Procedures  for  decontamination  of equipment  and
 materials should be outlined in this section. In addition, a
 field sampling summary table  (see  Table 2) should be
 completed.  In this table, specify the number of samples
 required  per parameter per matrix,  the number of QA
samples,  the  required preservatives,  appropriate sample
containers and sample volumes.
2.4.1 Methods
It  should not  be assumed that an  analytical  method
imparts a  certain degree  of quality  to  the results it

-------
2.6 Project Organization and Responsibilities


This section should list the managers, coordinators, and
field sampling personnel, along with their project duties
and responsibilities. The name and type of the laboratory
performing the analysis, if appropriate,  should also be
included in this section.  In addition, the parameters of
interest (BNAs, VOAs, metals) should be detailed.
2.7  Qualify Assurance Requirements


This section should describe the appropriate data quality
indicators and QA/QC protocols, based on the QA/QC
objective determined in Section 3.0, which will be followed
in  the evaluation  of lab data packages.  A QA/QC
Analysis and Objectives Summary, including references to
analytical methods (see Table 3), should be completed.
The data quality indicators of concern for each QA/QC

 objective are listed below.


     OA1


       The following requirements apply:

       A.   Sample documentation.
       B.   Instrument calibration data or a performance
            check of a test method (i.e.,  Draeger tubes,
            test strips, spot tests).
       C.   Detection limit" should be determined, unless
            inappropriate.

       Note:   QC  procedures  prescribed in SOPs and
       methods must be followed.

      OA2

       The following requirements apply:

       A.   Sample documentation.
        B.   Chain of custody (optional for field screening
            locations).
        C.   Sample  holding times  (document  sample
            collection and analysis  dates).
D.    Initial and continuing instrument calibration data.
E.    Method blank, rinsate  blank, trip  blank data
      (refer to Table 2, footnotes 2 and 3).

F.    Choose any one "or  any  combination of the
      following three:

  1.   Definitive identification (chaose one):

      a.   Screened data -  confirm the identification
           of analytes via an EPA-approved method
           different from the  screening method (field
           or lab) on at least 10^> of the preliminary
           screened   samples   collected;  provide
           documentation such as gas chromatograms,
           mass spectra, etc.    |
      b.   Unscreened data - confirm the identification
           of analytes via an EPAJapproved method on
           all  unscreened   environmental  samples;
           provide  documentation  such  as  gas
           chromatograms,  mass spectra,  etc.
  2.   Non-definitive auantitaticn
(choose one):
       a.   Screened data - provi ie documentation of
           quantitative results from both the screening
           method and the EPA-approved verification
           method.
       b.   Unscreened data -  provide documentation
           of quantitative results}

   (Documentation includes informjation and/or evidence
   on calculation procedures, calibration data,  sample
   weight  or volume, dilution factor, etc.)

   3.   Definitive quantitation/analytical error
       (choose one):

       a.   Screened data - determine  the  analytical
            error by calculating trie precision, accuracy,
            and coefficient of  variation* by  preparing
            and analyzing eight (8) QA replicates from
            the subset  of  samples  used  to verify
            screening  results usiiig an EPA-approved
            method. (See error determination  Section
            2.8.)
       b.  Unscreened data -  determine the analytical
            error by calculating the precision, accuracy,
            and coefficient of  vajriation* by  preparing
            and analyzing eight (8) QA replicates from
            all of the samples  anjalyzed using an EPA-
            approved method.
    G.  Performance Evaluation Sample (optional) and
        where available.
    H.  Detection  limit should be determined, unless
        inappropriate.

-------
OA3:
 Note:
          The following requirements apply:

          A.  Sample documentation.
          B.  Chain of custody.
          C.  Sample holding times (document sample
              collection and analysis dates).
          D.  Initial   and   continuing  instrument
              calibration data.

          E.  Definitive identification:

              Confirm the identification of analytes by
              an EPA-apprqVed method on 100% of the
              "critical" samples collected; and provide
              documentation   such   as    gas
              chromatograms, mass spectra, etc.

          F.  Non-definitive quantitation (choose one):

              a.  Screened   data   -   provide
                  documentation of quantitative results
                  from both the screening method and
                  the  EPA-approved   verification
                  method.
              b.  Unscreened   data   -   provide
                  documentation of quantitative results.

          (Documentation includes information and/or
          evidence on calculation procedures, calibration
          data, sample weight or volume, dilution factor,
          etc.)

          G.  Definitive quantitation/analytical error

              Determine  the analytical error by  an
              EPA-approved method on 100% of the
              "critical" samples collected. Calculate the
              precision, accuracy,  and  coefficient of
              variation* by  preparing  and  analyzing
              eight (8) QA  replicates from the critical
              samples   collected.   (See  error
              determination Section 2.8.)

          H.  Method blank, rinsate blank,  and  trip
              blank  data (refer to Table 2, Footnotes 2
              and 3).
See data validation protocols for determining
precision, accuracy, and coefficient of variation.
                                                    I.    Performance   Evaluation  Samples,   where
                                                         available.
                                                    J.    Detection  limit should be determined, unless
                                                         inappropriate.


                                              Reference must be made to standard QA/QC protocols (i.e.,

                                              SOPs, EPA reference procedures) for generating the above
                                              data quality indicator information.
                                              2.8  Error Determination (Analytical and Total Error)


                                              Any  one of  the following  options  can be  used  when
                                              determing error for QA2 or QA3:


                                              2.8.1   Matrix Spike Samples


                                              Spike and analyze at least eight (8) replicate samples with

                                              a concentration level equal to  the level of  interest.  Use

                                              samples  whose  unspiked concentrations are less than or

                                              equal  to the  level  of interest.    Samples  should  be
                                              homogeneous.   Determine bias (percent  recovery)  and
                                              precision (coefficient of variation) according to Section 3.5
                                              of Part II - Data Validation Procedures.


                                              2.8.2   Site Background Samples


                                              Spike and analyze at least eight (8) replicate samples with
                                              a concentration level equal to the level of interest. These

                                              samples are  from the site of interest (or nearby proximity).
                                             The analyte  of interest is not detectable in the sample for

                                              the method  used.  Samples should be made homogeneous.

                                             Determine bias (percent recovery) and precision (coefficient
                                             of variation) according to Section 3.5  of Part  II -  Data
                                             Validation Procedures.

-------
2.83  Site Action Level Samples (Total Error)

Collect  and analyze at least 8 replicate samples whose
analyte  concentrations are equal to the level of interest.
(Do this by collecting one sample with sufficient material
to divide into the required number of replicates.  Except
for VOA samples, homogenize the  sample thoroughly
before dividing into replicates.) These samples are from
the selected site and contain the target analyte at or near
the level of interest.  Determine bias (percent recovery)
and precision (coefficient of variation)  according  to
Section 3.5 of Part II - Data Validation Procedures. Bias
can not be determined unless these samples are spiked
first and percent recovery is calculated.

Note:  This procedure (2.8.3) is useful in determining the
total (sampling  and  analytical)   error  as well as  the
analytical error since it evaluates the sample collection,
sample preparation,  and  the analysis.  Sampling  error
determination is  being  addressed  hi  representative
sampling guidance documents for each media.  These
 documents are under development for removal activities.
 2.9  Deliverables

 This section should provide a description of the reports
 and other deliverables (e.g., field activities, trip reports,
 status  reports,  maps/figures,  analysis,  data review,
 analytical reports, and draft final reports) to be generated
 as a result of the sampling activity.

 2.10 Data Validation

 This section details the criteria used to ensure that the
 analytical results received from a laboratory are valid and
 accurate for the QA objective chosen. Consult the "Data
Validation Procedures" in this guidance document for the
appropriate evaluation criteria. These procedures have been
developed mainly  from the  "Laboratory  Data Validation
Functional Guidelines for Evaluation ofj Organic, Inorganic,
and  Dioxm  Analyses" used in  the
Laboratory Program.
QA1
Agency's  Contract
     QA1 data need only be evaluated for calibration and
     detection limits criterion.
QA2
     The results of 10% of the samples in the analytical data
     packages should be evaluated forj all of the elements
     listed  in  Section 2.7,  "QA  Requirements"  of the
     Sampling  QA/QC  Plan.   The  Holding  times, blank
     contamination, and detection capability will be reviewed
     for all remaining samples.
 OA3
     This objective, the most stringent
of all the objectives,
     requires that at least 10% of the Camples in a lab data
     package be evaluated for all of the listed elements in
     Section 2.7  "QA  Requirements™'  of  the Sampling
     QA/QC Plan.  Of the remaining  samples,  holding
     times, blank contamination, precision, accuracy, error
     determination,  detection   limitjs,   and  confirmed
     identification data will be reviewed/This objective also
     requires review of all elements for all samples in each
                                                              analyte category (i.e., VOAs and
     data package received from an individual lab.
                                     PCBs) in every 10th
                                                         10

-------
 Table 1:  Example Proposed Schedule of Work
                   Item
        (time period)
J.	        /
 1.  Laboratory Procurement

 2.  Phase 1 Site Work

 3.  Drilling Subcontract Procurement

 4.  Phase 2 Site Work

 5.  Laboratory Analysis

 6.  Data  Review

7.  Draft Report

8.  Final  Report
                                                 11

-------
Table 2:  Field Sampling Summary
Analytical
'arameter
VOA
VGA
BHA
BNA
PESTICIDE
PESTICIDE
PCB
PCB
P.P.
HETALS
P.P.
HETAUS
CYANIDE
CYANIDE
Level
of
ensiti-
vity'












*
Matrix
S
W
S
W
S
W
S
W
S
W
S
W
Container Type
and Volume
# container rq'd)
40ml vial
(1)
40ml vial
(3)
8oz glass
(1)
32oz amber glass
(2)
8oz glass
(1)
32oz amber glass
(2)
8oz glass
(1)
32oz amber glass
(2)
8oz glass
(1)
1 liter glass or
polyethylene
(1)
8oz glass
CD
1 liter
polyethylene
(1)
reserv-
ative
4°C
**
4°C
4°C
4°C
4°C
**
4°C
4°C
**
4°C
4°C
N03 ph<2
4«C
4°C
NaOH to
pH > 12
4°C
iolding
Times
7 day
7 day
7/40 d
7/40 d
7/40 d
7/40 d
7/40 d
7/40 d
6 mos
6 mos
14 day
14 day

ubtotal
amples












QC Extras
insate
lanks"












rip ,
tanks'1
VOAs)


•
•
•
•
•
•
H
•
•
H
SH
oc
ositiv*













s4












Matrix,-
Spikes3












Total
Field
Samples












 *
 **
    Matrix:  S-Soil, W-Water, 0-Oil, DS-Drum Solid,  DL-Drum Liquid,  TS-Tank  Solid,' TL-Tank Liquid, X-Other, A-Air
      If  residual chlorine is present, preserve with 0.008%
  1.   The concentration  level, specific or generic, that is ne
in order to make an evaluation.
      provide a basis  for  determining  the analytical method to be used.
      onlv reouired if dedicated  sampling tools are not used.  For QA2 and QA3, one blank required
2.  Only required if dedicated sampling
                                                                                                 This level will

                                                                                                per parameter per 20
  3    ForPGA2"and°QA3,  one"trip blank  required per cooler used  to ship VOA samples.  Each trip blank consists  of  two
      40ml  vials filled with  distilled/deionized water.  For QA1, enter "N/A".                    I
  4.   Performance check samples; optional  for QA2, mandatory for QA3 at one per parameter per matrix.   For  QA1, enter

  5    For QA2 (optional) and  for QA3 (mandatory):  Determine bias (% recovery) using a minimum of 2 matrix  spikes.
   "   Determine precision using a minimum  of 8 matrix spikes.   Ensure that sufficient environmenta  sample  is  collected
      for lab spiking.   For QA1, enter "N/A".
                                                           12

-------
 Table 2:  Field Sampling Summary (continued)
Analytical
Parameter
PHENOLS
PHENOLS










Level
of
Sensiti-
vity'












Matrix
S
W










Container Type
and Volume
(# container rq'd)
8oz glass
(1)
1 liter amber
glass
(1)










Preserv-
ative ;
4°C
H2SO, to
pH 5 2
4«C










Holding
Times
28 day
28 day










Subtota
Samples












QC Extras
Rinsate
Blanks












Trip ,
Blanks
(VOAs)
•
•










QC
Positives












Matrix-
Spikes













Total
Field
Samples












3.
**  tf nXL S',=<• r.,,tt;~:—»  	.-	,	•   -r   .'
                                                                                                        is
                                                        13

-------
TableS:  OA/OC Analysis and Objectives Summary
Analytical
Parameter
VGA
VOA
BHA
BHA
PESTICIDE
PESTICIDE
PCS
PCB
P.P.
METALS
P.P.
METALS
CYANIDE
CYAHIDE
Matrix
S
W
S
W
S
W
S
W
S
W
S
W

Analytical
Method Ref.
8240/SW-846
624/CLP
8250 or 8270/
SW-846
625/CLP
8080/SW-846
608
8080/SW-846
608
SW-846
EPA-600/CFR 40
SW-846
SW-846
Spil
Matrix1













ces
Surrogate






• *

•
•
B^S8PSpii|Biffi$S||iji|


election
Limits3












A Objective4












  *   Matrix:  S-Soil,  W-Water, 0-0! I, DS-Drum Solid, DL-Drun Liquid, TS-Tank Solid, TL-Tank Liquid, X-Other
  '•      ef 'SSSS!

  2.  iF^U«58SSSt£rS;«f ^K^lvlr.  "VSTrunT  each sa^ie; therefore, enter -y...  For


  3.  ?o"b4 detlrminedby the  person arranging the  analysis.  Should be equal to or less than the | level of

      sensitivity.
  4.  Enter the QA Objective desired:  QA1, QAZ, or QA3.
                                                        14

-------
 Table 3:  QA/QC Analysis and  Objectives Summary  (continued)
Analytical
Parameter
PHENOLS
PHENOLS











Matrix
S
W











Analytical
Method Ref .
8040/SW-846
604/CFR 40











Spikes
Matrix1













Surrogate

in











QA/QC

Limits'5













QA Objective













    Matrix:  S-Soil, W-Water,  0-Oil,  DS-Drum Solid,  DL-Drum  Liquid, TS-Tank Solid, TL-Tank Liquid, X-Other,
    A-Air
1.  For QA2, (optional) and for QA3 (mandatory):   Determine bias  (% recovery) using a minimum of 2 matrix
    spikes.  Determine precision using a minimum of  8 matrix spikes.  Ensure that sufficient environmental
    sample is collected for lab spiking.  For QA1, enter "N/A".
2.  For QA2 and QA3, surrogate spike  analysis is to  be  run for each sample; therefore  enter "yes"   For
    QA-1:  enter "N/A".
3.  To be determined by the person arranging the analysis.   Should be equal to or less than the level of
    sensitivity.
4.  Enter the QA Objective desired:   QA1,  QA2, or QA3.
                                                        15

-------
          PARTH
DATA VALIDATION PROCEDURES
             16

-------
 1.0   Introduction


 Part II provides guidance in the validation of laboratory
 data packages, according to the guidelines established by
 the Sampling QA/QC Plan.  It is a compilation of those
 procedures  used hi the Contract Laboratory Program
 (CLP) and those found in the "Laboratory Data Validation
 Functional Guidelines for Evaluating Organic, Inorganics,
 Pesticides, and Dioxin  Analysis."    This guidance was
 developed for the Emergency Response Divisions' (ERD)
 use  and  is not intended to supercede the  guidance
 documents developed for  CLP data validation used for
 Remedial activities.

 Items reviewed during  the  data validation  process are
 dependent upon the QA objectives previously established
by the data user in the Sampling QA/QC Plan. According
to the tiered  approach implemented  in  the Sampling
QA/QC Plan each QA  objective requires the following
review:
       QA3 - This objective, the most stringent of all the
       objectives, requires 'that at  least 10%  of the
       samples in a lab data package be reviewed for all
       of  the  elements.    Of  the  remaining samples,
       holding times, blank contamination, precision,
       accuracy, error determination, detection limits, and
       confirmed identification data will be reviewed.
       This  level also  requires the review of  all the
       elements for all samples in each analyte category
       in every  10th data package  received  from an
       individual  lab.

       QA2  - This objective requires that the results of
       10% of the samples reported in the analytical data
      package should be evaluated for all of the elements
      listed in Section  7, QA Requirements,  of the
      Sampling QA/QC Plan.  The holding times, blank
      contamination,  and  detection  limits  will  be
      reviewed for the remaining.
         QA1 - Thjs objective requires review of only the
         calibration and detection limits for all data.

 Included  hi the  section on Matrix  Spike/Matrix Spike
 Duplicates are formulas for calculating confidence limits and
 the coefficient of variation.  Confidence limits should be
 determined for all data generated under QA3 and may be
 calculated for QA2 if, a sufficient number of spiked samples
 are collected.  Although not stated in the following data
 validation procedures,  the reviewer must examine  the data
 packages for transcription/calculation errors that may have
 been overlooked by the lab.
 2.0    Data Validation Qualifiers
 J      The associated  numerical value is an  estimated
        quantity because the reported concentrations were
        less  than  the required  detection limits or quality
      - control criteria were not met.

 N      Presumptive evidence  of presence of material.
 NJ     Presumptive evidence of the presence of the material
        at an estimated quantity.

 PND   Precision Not Determined.   -
R
The sample results arc rejected (analyte may or may
not be present) due to gross deficiencies is quality
control criteria.  Any reported value is  unusable.
Resampling  and/or  reanalysis  is  necessary  for
verification.
RND  Recovery Not Determined.
U
UJ
The material  was analyzed for, but not  detected.
The associated  numerical  value  is  the sample
detection limit or adjusted sample detection limit.

The material  was analyzed for, but not  detected.
The reported  detection limit is estimated because
Quality Control criteria were not met.
                                                   17

-------
3.0    Metallic Inorganic Parameters

3.1     Sample Holding Times

       1. Were  any  of  the  sample  holding  tunes
         exceeded?*

         Sample Holding Times:
                        Metals - 6 months
                        Cyanide - 14 days
                        Mercury - 28 days
                        Chromium4"6 - 24 hours

       ACTION: If yes, flag as estimated (J) those values
       above the  Instrument Detection Limit (IDL).
       Values .that are less than the IDL can be flagged
       as estimated (UJ) or rejected (R) based on the
       reviewers professional judgement and the nature of
       the sample and analyte.

       *Because of their long shelf lives, performance
       evaluation samples  do not have any associated
       holding times.
3.2    Initial and Continuing Calibration Verification

       1. Are values outside the range of 90% to 110% of
          the mean value, except for tin and mercury, for
          which the range is 80% to 120%, and cyanide,
          for which the range is 85% to 115%?

       ACTION: If values are between 75-89% or 112-
       125% (65-79% and 121-135% for Hg and Sn, 70-
       84% and 116-130% for cyanide), flag as estimated
       (J).

       If values are outside of the above windows, reject
       (R)  as unacceptable data  between  calibration
       standard outside  of above windows and nearest
       adjacent acceptable calibration standard(s).
2. Was a calibration standard and blank analyzed at
  the beginning of the analysis and after every  10
  samples?

ACTION: If no, flag as estimated (J) all values not
analyzed within 5 samples of a calibration standard
or blank.
3. Were any sample results greater than 110% of the
  highest calibration standard?
ACTION:  If yes, flag result reported as estimated
(J).
 1. Do the concentrations oi; all blanks fall below the
   IDL for all parameters?

 ACTION:  If no, flag as undetected (U) all reported
 positive data that has a concentration less than 5
 times the blank value.     |
                         I
 NOTE:  In instances where more than one blank is
 associated with a given sample,  qualification should
 be based upon  a comparison with the  associated
 blank  having  the  highest  concentration  of  a
 contaminant. The results must  not be corrected by
 subtracting any blank value.

 2. Was one  method blank analyzed for  each 20
   samples?

 ACTION: If no, flag as estimated (J) all data for
 which a method blank was not analyzed. If only one
 blank was analyzed for more than 20 samples,  the
 first 20 samples analyzed do not have to be flagged
 as estimated (J).
                                                    18

-------
 3.4     ICP Interference Check Sample

        1. If all ICP  Interference Check Sample (ICS)
          results are not inside of control limits (± 20%
          of mean value), are concentrations of Al, Ca,
          Fe, or Mg lower in the sample than in the ICS?

       ACTION: If no, flag as estimated (J) those  sample
       results for which ICS recovery is between  +. 50%
       of mean value.  For those sample results in which
       ICS recovery is above 150% or 50%, reject (R) all
       results.

       2. Was  ICS analyzed at the beginning and end of
          each run or  at least  twice  every  8  hours,
          whichever is more frequent?

       ACTION: If no, flag as estimated (J) all samples
       for which Al, Ca, Fe, or Mg concentration is higher
       than in ICS.
3.5    Error Determination
       See Part I - Section 2.8 for QA Samples to be used
       for error determination.

3.5.1   Determination of Bias (% Recovery - Optional for
       QA-2; Mandatory for QA-3)

       3.5.1.1   Percent Recovery

               1.  Were at least  eight spiked  sample
                  replicates for the matrix of interest
                  analyzed at the required frequency?

               ACTION:  If  no,  flag as recovery not
               determined fRND) all data for  which
               spiked samples were not analyzed.

               2.  Determine the average recovery of the
                  eight spiked replicates. Is  the average
                  recovery within the applicable control
                  limits (80% to 120%)?
          % recovery for a single spiked sample =

               Spiked sample cone. - Sample cone. x IQQ
                       Spike cone,  added

               ACTION:  If recoveries are within applicable
               control limits, no bias is considered.  If %
               Recovery is less than 80% or greater than
               120%, the sample data should be flagged with
               a (J) estimate and a corresponding (-) or ( + )
               sign  to   show   direction   of  the  bias.
               Adjustment of sample values should  be
               considered  whenever  there, is  consistent
               evidence of bias.

       3.5.1.2  Adjustment of Sample Value for Bias

               1.  Depending  on  bias  direction,  add  or
                 subtract  the  value   (%  Bias   x  spike
                 concentration)  to  or from the  sample
                 values.  % bias is the reciprocal  value of
                 %  recovery (i.e., for 70% recovery you
                 have a  negative 30% bias).   Use the
                 average % recovery from the total number
                 of matrix spikes analyzed. This adjustment
                 approach assumes a spiking concentration
                 .equal  to the  concentration found in the
                 sample.

3.5.2   Determination  of  Precision  (Optional for QA-2;
       Mandatory for QA-3)

       3.5.2.1 Replicate Analysis

              1.  Was  a  minimum  of  eight  replicates
                 analyzed? If yes, determine coefficient of
                 variation. If no, flag data with precision
                 not determined (PNDV for which replicate
                 samples were not analyzed.
                                                    19

-------
3.5.2.2   Coefficient of Variation (Percent Relative
         Standard Deviation)

         1. The coefficient of variation (CV) is
           used in determining the precision or
           standard deviation. The CV expresses
           the standard deviation as a percentage
           of the mean  (average)  value of the
           replicate values.  The  CV is used to
           determine a  false positive or  false
           negative value  for results that are
           respectively greater than or less than
           a decision level concentration.

           Determine the coefficient of variation
           using the following equation:

           CV =  s  x 100

           where:

           XDL  = the decision level concentration

           s =  the sample  standard deviation
           given by the equation:

           s*  =    [fc-xf/Oi-l)]*4
   *Note: When using a programmable calculator
   or computer statistics software,  be sure the
   above equation with (n - 1) is used and not (n)
   by itself.  The equation using (n) is to determine
   the population standard deviation (a)  rather
   than the sample standard deviation (s).

   Apply the CV to the decision level to determine
   the false negative  or  false positive value as
   follows:

   False positive value = Decision level value  +
   (CV x decision level)
3.6
         False negative value = Decision level value - (CV
         x decision level)

       Example:

         For an decision level = 50 ppm and CV = 20%

         False positive value = 50 ppm + (20%
         50 ppm)
                 = 50 ppm  + (10 ppm)
                 = 60 ppm       i

         False negative value = 50 ppm - (20% x
         50 ppm)
                 = 50 ppm  - (10 pbm)
                 = 40 ppm

       For the above false  positive  example, any value
       between 50 ppm and  60 ppm are considered suspect
       and should be reanalyzed. Values above 60 ppm are
       considered actionable. In many cases, false positives
       have been considered actionable by the Agency for
       safety reasons. However, depending on the action to
       be taken, this can  be costly and unjustifiable.
       Consult the QA plan for intended use  of data and
       data quality objectives.

       For the above false  negative  example, any values
       between 40 ppm and  50 ppm are considered suspect
       and should be reanalyzed. Values below 40 ppm are
       considered  non-actionable. .   In  most cases, the
       decision maker will be using the false negative value
       as his decision level and not
                          be concerned about the
       false positive value. Whenever sample values need
       to be corrected for both bias and precision,  first
       correct the value for bias, [then correct the biased
       value for precision.
Performance Evaluation Samples
                         I;
1. Were  recovery  limits  within those set by the
  EMSL lab?
                                              20

-------
        ACTION:   If outside  the limits, review on a
        compound by compound basis.  If 50% of the
        compounds  are outside  of confidence limits or
        were  misidentified, all sample results should be
        rejected (R).
3-7    Optional Additional Instrument QC (for elevafprf
       concentrations)

3-7.1   ICP  Serial  Dilution  (if  recovery  is  outside
       acceptable range)

       1. Was serial dilution performed on one of each 20
         samples of similar matrix where concentrations
         exceed 50 times IDL?

       ACTION:  If no, flag associated data as estimated
       (J).

       2. If analyte concentration after a five fold dilution
         is greater than 10 times IDL, did analysis of
         diluted sample agree to within 10% of original
         determination for each parameter?

      ACTION: If no, flag associated data as estimated
      (J).
3-7.2   Atomic Absorption Analysis Specific OC

       1. Is any furnace result flagged with an (E) by the
         laboratory to indicate interference?

         If yes, is any associated post-digestion spike
         recovery less than 10% for any result  flagged
         with an  (E).

      ACTION:  If yes, reject (R) affected data.
                                                        4.0
                                                        4.1
                                                                2. When  the method  of standard  addition  was
                                                                   required, is the coefficient of correlation less than
                                                                   0.995 for any sample?

                                                                ACTION:   If yes,  flag the  associated data  as
                                                                estimated (J).
                                                                 Overall Assessment of Data

                                                                 It is appropriate  for  the data  reviewer  to use
                                                                 professional  judgment  and express  concerns and
                                                                 comments on the validity of the overall data package
                                                                 for a case. This is particularly appropriate for cases
                                                                 in which there  are several  QC  criteria  out of
                                                                 specification.  The additive nature  of  QC factors
                                                                 which are out of specification is difficult to assess in
                                                                 an  objective  manner,  but  the  reviewer  has  a
                                                                responsibility to inform the user about data quality
                                                                and data limitations.  This helps the user to avoid
                                                                using data inappropriately, while  not  precluding
                                                                consideration of the data.  The data reviewer would
                                                                be greatly assisted in this endeavor if the data quality
                                                                objectives were provided.
                                                               BNAs by GC/MS Analysis
                                                               Sample Holding Times

                                                               1. Were any of the sample holding times exceeded?*

                                                                 Sample  Holding  Times  from date of  sample
                                                                 collection:

                                                                 Water - 7 days to  extract
                                                                 Soil, sediment, sludges - 14 days to extract

                                                                 Water/soil - analyze within 40 days after extraction
                                                   21

-------
      ACTION: If yes, flag as estimated (J) those values
      above  the Instrument Detection  Limit (IDL).
      Values that are less than the IDL can be flagged
      as estimated (UJ) or rejected (R) based on the
      reviewers professional judgement and the nature of
      the sample and analyte.

      *Because of their long shelf lives, performance
      evaluation samples do not have any associated
      holding times.
42     GC/MS Tuning Criteria

       1. Has  decafluorotriphenylphosphine  (DFTPP)
         been run for every 12 hours of sample analysis
         per instrument?

       ACTION:  If no, reject (R) all associated data for
       that instrument which fall outside an acceptable
       12-hour time interval.

       2. Have the DFTPP ion abundance criteria been
         met for each instrument used?

         m/z    Ion abundance criteria

         51      30-60% of mass 198
         68      Less than 2% of mass 69
         69      (reference only)
         70      Less than 2% of mass 69
          127     40-60% of mass 198
          197     Less than 1% of mass 198
          198     Base peak, 100% relative abundance
          199     5-9% of mass 198
          275     10-30% of mass  198
          365     Greater than 1% of mass 198
          441     Less than mass 443
          442     Greater than 40% of mass 198
          443     17-23% of mass  442

        ACTION:  If no,  evaluate against expanded  ion
        abundance criteria.

        3. Have the appropriate expanded ion abundance
          criteria been met for each instrument used?
  m/z   Expanded ion abundance criteria

  51     22-75% of mass 1R8
  68     Less than 2% of mass 69
  69     (reference only)   :
  70     Less than 2% of mass 69
  127    30-75% of mass 198
  197    Less than 1% of mass 198
  198    Base peak, 100% [relative abundance
  199    5-9% of mass 198J:
  275    7-37% of mass 198
  365    Greater than 0.75;% of mass 198
  441    Present, but lessjhan mass 443
  442    Greater than 30% of mass 198
  443    17-23% of mass 4*2
ACTION: It is up to the reviewer's discretion, based
on professional judgement, to flag data associated
with tunes meeting expanded criteria, but not basic
criteria.   If only one element  falls  within  the
expanded criteria, no qualification may be needed.
On the other hand, if several data elements are in
the expanded windows, all associated data may merit
an estimated flag (J).  Note that the data reviewer
may still choose  to flag all data associated with a
tune not meeting contract criteria as rejected (R) if
it is deemed appropriate.
                         i
The most critical factors in! the DFTPP criteria are
the  non-instrument specific requirements that are
also not  unduly affected by the  location of the
spectrum on the chromatographic profile. The m/z
198/199 and 442/443 ratios jare critical. These ratios
are based on the natural abundances of Carbon 12
and Carbon 13 and should always be met. Similarly,
the m/z 68, 70, 197 and ml  relative abundances
indicate the condition of  the  instrument and the
suitability of the resolution! adjustment and are very
important. Note that all of the foregoing abundances
relate to adjacent ions; thej are relatively insensitive
to  differences in instrument design and position of
the spectrum  on the chromatographic profile.  For
the ions at m/z 51, 127, arid 275, the actual relative
 abundance is  not critical.  For instance, if m/z 275
 has a 40% relative abundance (criteria 10-30%) and
 other  criteria are met, the deficiency is minor. The
                                                     22

-------
       relative abundance of m/z 365 is an indicator of
       suitable instrument zero adjustment. If m/z 365
       relative abundance  is zero,  minimum  detection
       limits may be affected. On the other hand, if m/z
       365 is present, but  less than the 1%  minimum
       abundance criteria, the deficiency is not as serious.
4-3     Initial and Continuing Calibration Verification

       1. Do any compounds have an average response
         factor equal to zero?

       ACTION:   If yes, reject  (R) sample  data for
       associated compounds.

       2. Verify that all BNA compounds have Relative
         Response Factors of at least 0.05.

       ACTION: If any BNA compound has a Relative
       Response Factor of less than 0.05, flag positive
       results for that compound as estimated (J).  Flag
       non-detects for that compound as rejected  (R).

      3. Verify that all BNA compounds have a percent
         Relative Standard Deviation (%RSD) of <. 30%
         for the initial calibration.
              %RSD = s x  100
                       x
        where:

        s = standard deviation of 5 response factors
        x = mean of 5 response factors
     ACTION:  If any BNA compound has a %RSD of
     greater than 30%, flag positive results for that
     compound as estimated (J). Non-detects may be
     qualified  (J)  using  the reviewer's professional
     judgement.
        4. Verify that the percent difference (%D) is <. 25%
           for  all  BNA  compounds  in the  continuing
           calibration.

        ACTION: If any BNA compound has a %D between
        the initial and continuing calibration of greater than
        25%, flag all positive results for  the compound as
        estimated (J). Non-detects maybe qualified (J) using
        the reviewer's professional judgement.
 4.3.1   Internal Standards

        1. Verify that  all retention times  and  Internal
          Standard (IS) areas are acceptable.

        ACTION:  If an IS area is outside -50% or +100%
        of the associated standard, flag the positive results as
        estimated (J) for that sample fraction.  Non-detects
        for compounds quantitated using that IS are flagged
        with  the sample  quantitation  limit  flagged  as
        estimated (J) for that sample fraction.  If extremely
        low area counts are  reported, or  if performance
        exhibits a major abrupt drop-off, then a severe loss
        of sensitivity is indicated. Non-detects should then be
        flagged as rejected (R).

        If an IS retention time varies by more than 30
        seconds, the chromatographic profile for that sample
        must be examined to determine if any false positives
        or negatives exist.  For shifts of a large  magnitude,
        the reviewer may consider partial or total  rejection
        (R) of the data for that sample fraction.
4.4    Error Determination

       See Part I - Section 2.8 for QA samples to be used
       for error determination.
                                                  23

-------
4.4.1   Determination of Bias (% Recovery- Optional for
       QA-2; Mandatory for QA-3)

       4.4.1.1   Percent Recovery

                1. Were  at least eight spiked  sample
                  replicates for the matrix of interest
                  analyzed at the required frequency?

                ACTION:  If  no, flag  as recovery not
                determined (RND) all data for which
                spiked samples were not analyzed.

                2. Determine the average recovery of the
                  eight spiked replicates.  Is the average
                  recovery within the applicable control
                   limits (80% to 120%)?
        Cf,
% recovery for a single spiked sample =
   Spiked sample cone. - Sample cone.  x
                    Spike cone, added
                ACTION:    If recoveries  are  within
                applicable  control  limits,  no  bias is
                considered.  If % Recovery is less than
                80% or greater than 120%, the sample
                data should be flagged with a (J) estimate
                and a corresponding (-) or (+) sign to
                show direction of the bias. Adjustment of
                sample  values should  be  considered
                whenever there is consistent evidence of
                bias.

        4.4.1.2  Adjustment of Sample Values for Bias

                 1. Depending on bias direction, add or
                    subtract  the value (%  Bias x spike
                    concentration) to or from  the sample
                    values.  % bias is the reciprocal value
                    of % recovery (i.e., for 70% recovery
                    you have a negative 30% bias).  Use
                    the average % recovery from the total
                    number of matrix spikes analyzed.
                                                               This adjustment approach assumes a spiking
                                                               concentration equal
                                                               found in the sample.
                                                 4.4.2   Determination of Precision
                                                        Mandatory for QA-3)
                    to  the  concentration
                    (Optional for  QA-2;
                                                        4.4.2.1 Replicate Analysis
                                                               1. Was  a  minimum   of  eight  replicates
                                                                  analyzed? If yes, determine coefficient of
                                                                  variation. If no, flag data with precision
                                                                  not determined CPNDX for which replicate
                                                                  samples were not
                   analyzed.
                                                         4.4.2.2  Coefficient of Variation (Percent Relative
                                                                Standard Deviation)
1. The coefficient o£ variation (CV) is used
  in determining the precision of standard
  deviation. The CV expresses the standard
  deviation as  a percentage of the  mean
  (average) value  of  the, replicate values.
  The  CV is used to  determine a false
  positive or false rlegative value for results
  that are respectively greater than or less
  than a decision level concentration.
                                                                   Determine the coe fficient of variation using
                                                                   the following equation:

                                                                         CV =  s  *  100
                                                                where:
                                                                     =  the decision level concentration
                                                                 s = the sample standard deviation given by
                                                                 the equation:      |

                                                                                   |
                                                                 s* =    [(x, - x)2/(n - l)f2
                                                       24

-------
    *Note: When using a programmable calculator
    or computer statistics software,  be sure the
    above equation with (n - 1) is used and not (n)
    by itself. The equation using (n) is to determine
    the population standard deviation  (o) rather
    than the sample standard deviation (s).

    Apply the CV to the decision level to determine
    the false negative or false positive value  as
    follows:

    False positive value = Decision level value +
    (CV x decision level)

    False negative value  = Decision level value -
    (CV x decision level)

 Example:

    For an decision level = 50 ppm and CV = 20%

    False positive value = 50 ppm + (20% x
    50 ppm)
            = 50 ppm + (10 ppm)
            = 60 ppm
   False negative value = 50 ppm -
   50 ppm)
            = 50 ppm - (10 ppm)
            = 40 ppm
(20% x
For the above false positive example, any value
between 50 ppm  and 60 ppm  are  considered
suspect and should be reanalyzed. Values above
60 ppm are considered actionable. In many cases,
false positives have been considered actionable by
the  Agency  for  safety  reasons.    However,
depending on the action to be taken, this can be
costly and unjustifiable.  Consult the QA plan for
intended use of data and data quality objectives.

For the above  false negative example, any values
between 40 ppm  and  50 ppm are considered
                        suspect and should be reanalyzed. Values below 40
                        ppm are considered non-actionable.  In most cases,
                        the decision maker will be using the false negative
                        value as his decision level and  not  be concerned
                        about the  false positive value.   Whenever sample
                        values need to  be corrected for both  bias  and
                        precision, first correct the value for bias, then correct
                        the biased  value for precision.
                 4.5     Blanks
 1. Was a method blank extracted and analyzed for
    each set of samples or every 20 samples of similar
    matrix and similar extraction technique?

 ACTION:  If no, flag as estimated (J) all data for
 which  a method blank was not analyzed.   NOTE:
 If only one blank  was analyzed for more than 20
 samples, the first 20 samples  analyzed do not have
 to be flagged as estimated (J).

 2. Has the method blank for BNAs been run on the
    same GC/MS or GC system as the sample?

 ACTION: If no, flag as estimated (J) all results  that
 do not  have an associated blank.

 3. Are the concentrations of blank  contaminants for
   BNAs greater than the Required Detection Limit
   (RDL) of any BNA  compound?

 ACTION:  For sample values  reported at less than
 10 times the blank contamination level for common
 phthalate esters and 5 times the blank contamination
 level for other BNA compounds, flag as undetected
 (U).

NOTE:  In instances where more than one blank is
associated with a given sample,  quantification should
be  based upon a comparison with the associated
blank  having  the   highest  concentration  of  a
                                             25

-------
       contaminant. The results must not be corrected by
       subtracting any blank value.

4.6    Compound Identification

       1. Verify the following:

         -the  Relative  Retention  Time  (RRT)   of
         reported compounds is within 0.06 RRT units of
         the standard RRT.

         -all ions present in the standard mass spectrum
         at a relative intensity greater than 10% are also
         present in the sample mass spectrum.

         -all ions present in the sample, but not present
         in the standard are accounted for.

         -relative intensities of the ions specified above
         as  present in  the  sample  and  at a relative
         intensity  greater than  10% hi the standard,
         agree within 20% between the sample and the
         standard spectra.

       ACTION:    Use  professional  judgement  to
       determine  acceptability of the  data if the above
       criteria were not all met.  If it is determined that
       incorrect identifications were made, all such data
       should  be  reported  as  not  detected  with  an
       estimated (J) quantisation limit. Ions greater than
       10% in the sample spectrum but not present in the
       standard  spectrum  must be  considered   and
       accounted for.
 4.6.1  Tentatively Identified Compounds

       1. Verify the following:

          -all ions presenting the reference mass spectrum
          with a relative intensity greater than 10% are
          present in the sample mass spectrum.
         -relative intensities specified above agree within
         20%  between  the
         spectra.
                     sample  and  the reference
         -molecular ions present in the reference spectrum
         are present in the sample spectrum.
         -all tentatively identified compounds are reported
         with estimated quantitation and detection limits.

       ACTION: Use professional judgement to determine
       acceptability of the data if the above criteria are not
       all met.  If data  are considered to be unacceptable,
       the tentative ID  should be changed to "unknown".
4.7     Compound Ouantitation  and Reported  Detection
       Limits                   j

       1. Verify that the reported values, both positives and
         non-detects, have been correctly adjusted to reflect
         all  dilutions,  concentrations,  splits,   cleanup
         procedures,  dry  weight factors,  an any  other
         adjustments that have nc t been accounted for by
         the method.

         BNA for waters:  ug/L = (AjfUfV,)
          BNA for soils: ug/kg =
A;,. =

Is  =
RF =

V, =
V; =
Vs =
Ws =
D =
              area of characteristic ion for compound being
              measured
              area of characteristic ion for  the  internal
              standard          j
              amount of internal standard added (ng)
              daily response factor for the compound being
              measured
              volume of total extrjact (ul)
              volume injected (ul)
              volume of sample (jnl)
              weight of sample extracted (g)
              (100 - % moisture) /.ICO
       ACTION:  If incorrect valubs have been reported, it

       is essential that the correct values be determined.
                                                     26

-------
         The reviewer should contact  the laboratory  to
         verify any corrections made to  the data.
  4.8     Performance Evaluation Samples

         1. Were recovery limits within  those set by the
        .   EMSL lab?

        ACTION:   If  outside  the limits,  review on  a
        compound by compound basis.   If 50% of the
        compounds are outside  of confidence  limits  or
        were misidentified, all sample results should be
        rejected (R).
 4-9    Overall Assessment of Data

        It  is appropriate for the data reviewer to use
        professional judgment and express concerns and
        comments  on the validity of the  overall data
        package for a case. This is particularly appropriate
        for cases in which there are several QC criteria out
        of specification. The additive nature of QC factors
        which are out of specification is difficult to assess
        in an objective manner, but the reviewer has a
        responsibility to inform the user about data quality
        and data limitations.   This helps the user to avoid
        using data inappropriately, while not precluding
        consideration of the  data.   The  data reviewer
        would be greatly assisted in this endeavor if the
        data quality objectives were provided.
4.10   Optional QC Checks

4.10.1  Surrogate Recovery

       1. If either  two or more base  neutral or  acid
         surrogates were outside of specifications for any
         sample or blank, were the appropriate samples
         reanalyzed?
5.0

5.1
        ACTION:  If initial analysis  and reanalysis both
        have two or more surrogates putside of specifications
        for samples or blanks, estimate (J) all quantitation
        results, including detection limits.

        2. Does any  one surrogate have less  than 10%
          recovery?

        ACTION: If yes, flag as estimated (J) positive results
        for that fraction; flag negative results as rejected (R).
 VOAs by GC/MS Analysis

 Sample Holding Times

 1. Were any of the sample holding times exceeded?*

   Sample Holding Times  from date  of sample
   collection:

   Aromatic  (for water)  - 7  days (unpreserved), 14
   days (preserved)

   All other compounds - 14  days

   Soil, sludge, sediments - 14 days

ACTION:  If yes, flag as estimated (J) those values
above the Instrument Detection Limit (IDL). Values
that  are less than  the  IDL can  be  flagged  as
estimated   (UJ)   or  rejected (R)  based  on  the
reviewers professional judgement and the nature of
the sample and analyte.
      *Because  of their  long shelf lives,  performance
      evaluation  samples do not  have any associated
      holding times.
                                                    27

-------
52     GC/MS Tuning Criteria

       1. Has bromofluorobenzene (BFB) been run for
         every  12  hours  of  sample   analysis   per
         instrument?

       ACTION:  If no, reject (R) all associated data for
       that instrument which fall outside an acceptable
       12-hour time interval.

       2. Have the BFB ion abundance criteria been met
         for each instrument used?
         m/z

         50
         75
         95

         96
         173
         174
         175
         176
         177
Ion abundance criteria

15-40% of mass 95
30-60% of mass 95
Base   peak,   100%  relative
abundance
5-9% of mass 95
Less than 2% of mass 174
Greater than 50% of mass 95
5-9'% of mass 174
95-101% of mass 174
5-9% of mass 176
       ACTION:   If no, evaluate against expanded ion
       abundance criteria.
53
                                      but not basic criteria.  If anly  one element falls
                                      within the expanded criteria, no qualification may be
                                      needed. On the other hand,
                                      are in the expanded windows
                                if several data elements
                                 all associated data may
       merit  an estimated flag (J).  Note that the data
       reviewer may still choose toj flag all data associated
       with a tune not meeting contract criteria as rejected
       (R) if it is deemed appropriate.


       For BFB, the most important factors to consider are
       the empirical results that are1 relatively insensitive to
       location on the chromatographic profile and the type
       of instrumentation.    Therefore,  the critical ion
       abundance criteria for BFB are the m/z 95/96 ratio,
       the 174/175 ratio, the 176/177 and the 174/176 ratio.
       The relative abundances oil m/z 50 and  75 are of
       lower importance.
Initial and Continuing Calibration Verification
                         I

1. Do any compounds  have an  average  response
  factor equal to zero?
                         j"

ACTION:    If  yes, reject j (R)  sample  data  for
associated compounds.     j
       3. Have the appropriate expanded ion abundance
         criteria been met for each instrument used?
          m/z

          50
          75
          95

          96
          173
          174
          175
          176
          177
Ion abundance criteria

11-50% of mass 95
22-75% of mass 95
Base  peak,  100%   relative
abundance
5-9% of mass 95
Less than 2% of mass 95
Greater than 50% of mass 95
5-9% of mass 174
95-101% of mass 174
5-9% of mass 176
       ACTION:  It is up to the  reviewer's  discretion,
       based on professional judgement, to flag data
       associated with tunes  meeting expanded criteria,
       2. Verify that  all VOA compounds have Relative
         Response Factors of at least 0.05.
                                j-
       ACTION:  If any VOA compound has a Relative
       Response Factor of less  than 0.05, flag  positive
       results for  that compound as estimated  (J).   Flag
       non-detects for that compound as rejected (R).
       3. Verify that all VOA compounds have a percent
          Relative Standard Deviation (%RSD) of <. 30%
          for the initial calibration.:

          %RSD  = s x 100      '•
                                                    28

-------
        where:
                                                         5.4     Error Determination
        s = standard deviation of 5 response
        factors

        x = mean of 5 response factors

        ACTION: If any VGA compound has a %RSD of
        greater than 30%, flag positive results  for that
        compound as estimated (J).  Non-detects may be
        qualified (J)  using  the reviewer's  professional
       judgement.

       4.  Verify that the  percent difference (%D)  is <_
          25% for all VOA compounds hi the continuing
          calibration.

       ACTION:   If any VOA  compound has a  %D
       between the initial and continuing calibration of
       greater than 25%,  flag all positive  results for the
       compound as estimated  (J). Non-detects  may be
       qualified  (J)  using the  reviewer's professional
       judgement.

5.3.1   Internal Standards

       1. Verify that all  retention times and  Internal
         Standard (IS) areas are acceptable.

       ACTION: If an IS area is outside -50% or  +100%
       of the associated standard, flag the positive results
       as estimated (J)  for that  sample fraction. Non-
       detects  for compounds quantitated using  that IS
       are  flagged with the  sample quantitation limit
       flagged  as estimated (J) for that sample fraction.
       If extremely low  area  counts  are reported, or if
       performance exhibits a major abrupt drop-off, then
       a severe loss of sensitivity is indicated. Non-detects
       should then be flagged as rejected (R).  If an IS
       retention time varies by more than 30 seconds, the
       chromatographic profile for that sample must be
      examined to determine if any false positives or
      negatives exist. For shifts of a large magnitude, the
      reviewer may consider partial or total rejection (R)
      of the data for that sample fraction.
       See Part I - Section 2.8 for QA samples to be used
       for error determination.

5-4-!   Determination of Bias (% Recovery - Optional for
       QA-2; Mandatory for QA-3)

       5.4.1.1 Percent Recovery

              1.  Were  at   least  eight  spiked  sample
                 replicates  for  the   matrix  of  interest
                 analyzed at  the required frequency?

             ACTION:    If no, flag  as  recovery not
             determined CRND^ all data for which spiked
             samples were not analyzed.

             2. Determine the average recovery of the
                eight spiked replicates.  Is the average
                recovery  within the  applicable  control
                limits (80%  to 120%)?

         % recovery for a single spiked sample =

             Spiked sample cone. - Sample  cone.  x ^QQ
                        Spike cone, added

             ACTION:  If recoveries are within applicable
             control  limits, no bias  is considered.  If %
             Recovery is  less than 80% or greater  than
             120%, the sample data should be flagged with
             a (J) estimate and a corresponding (-) or ( + )
             sign  to  show  direction  of  the  bias.
             Adjustment  of  sample  values should  be
             considered  whenever  there  is  consistent
             evidence of bias.

      5.4.1.2  Adjustment of Sample Values for Bias

             1. Depending on  bias  direction,  add  or
               subtract  the  value  (%  Bias  x spike
               concentration) to or from the sample
                                                   29

-------
      values.  %  bias is the reciprocal value of %
      recovery  (i.e., for 70%  recovery  you  have a
      negative 30% bias). Use the average % recovery
      from the total number of matrix spikes analyzed.
      This  adjustment  approach  assumes a spiking
      concentration equal to the concentration found in
      the sample.

5.4.2  Determination of Precision (Optional for QA-2;
       Mandatory for QA-3)

       5.4.2.1  Replicate Analysis

               1. Was a minimum of eight replicates
                  analyzed? If yes, determine coefficient
                  of variation.   If no, flag data with
                  precision not determined CPND1 for
                  which  replicate  samples  were  not
                  analyzed.

       5.4.2.2   Coefficient of Variation (Percent Relative
                Standard Deviation)

                1. The coefficient of  variation (CV) is
                   used in determining the precision of
                   standard deviation. The CV expresses
                   the standard deviation as a percentage
                   of the mean (average)  value  of  the
                   replicate values.  The CV is used to
                   determine a  false  positive or false
                   negative  value  for results  that  are
                   respectively greater than or less than
                   a decision level concentration.

                   Determine the coefficient of variation
                   using  the following equation:
                         CV =
s x  100
XDL
                   where:
                                         s = the sample standard deviation given by
                                         the equation:
         s  =
                                                      - x)2|/(n -
*Note:  When using a programmable calculator
or computer statistics software, be sure the above
equation with (n -1) is used and not (n) by itself.
The equation  using (n)  is  to  determine the
population standard deviation (
-------
       Consult the QA plan for intended use of data and
       data quality objectives.

       For the above false negative example, any values
       between 40 ppm  and 50  ppm are  considered
       suspect and should be reanalyzed.  Values below
       40 ppm are considered non-actionable.  In most
       cases, the decision maker will be using the false
       negative  value as his decision level and not be
       concerned  about   the  false   positive   value.
       Whenever sample values need to be corrected for
       both bias and precision, first correct the value for
       bias, then correct the biased value for precision.
5.5     Blanks

       1. Was a method blank prepared and analyzed for
         each  set of samples or  every 20  samples of
         similar   matrix  and    similar   preparation
         technique?

       ACTION:  If no, flag as estimated (J) all data for
       which a method blank  was not analyzed.  NOTE:
       If only one blank was  analyzed for more than 20
       samples, the first 20 samples analyzed do not have
       to be flagged as estimated (J).

       2. Has the method blank for VOAs been run on
         the same GC/MS  or GC system as the sample?

       ACTION:  If no,  flag as estimated (J) all results
       that do not have an associated blank.
      3. Are   the  concentrations   of   any  blank
         contaminants for VOAs greater than the RDL
         of any VOA compound?

      ACTION: For sample values reported at less than
      10  times  the  blank  contamination  level   for
      methylene  chloride,  acetone,  toluene  and  2-
      butanone and 5 times the blank contamination
       level for other VOA compounds, flag as undetected
       (U).

       NOTE: In instances where more than one blank is
       associated with a given sample, quantification should
       be based  upon a comparison with the associated
       blank  having  the   highest  concentration  of  a
       contaminant.  The results must not be corrected by
       subtracting any blank value.
5.6     Compound Identification

       1. Verify the following:

         -the Relative Retention Time (RRT) of reported
         compounds  is  within  0.06  RRT  units  of the
         standard RRT.

         -all ions present in the standard mass spectrum at
         a relative intensity  greater  than 10% are also
         present in the sample mass spectrum.

         -all ions present in the sample, but not present in
         the standard are accounted for.

         -relative intensities of the ions specified above as
         present in the sample and at a relative intensity
         greater than 10% in the standard, agree within
         20% between the sample and the standard spectra.

      ACTION: Use professional judgement to determine
      acceptability of the data if the  above criteria were
      not  all met.   If it  is  determined that incorrect
      identifications were made, all such  data should be
      reported  as  not  detected  with an  estimated  (J)
      quantitation limit.   Ions greater  than  10%  in the
      sample spectrum  but not present in the standard
      spectrum must be considered and accounted  for.
                                                   31

-------
5.6.1   Tentatively Identified Compounds

       1. Verify the following:

         -all ions presenting the reference mass spectrum
         with a relative intensity greater than 10% are
         present in the sample spectrum.

         -relative intensities specified above agree within
         20%  between  the  sample  and the reference
         spectra.

         -molecular  ions  present  hi the  reference
         spectrum are present in the sample spectrum.

         -all  tentatively  identified   compounds  are
         reported with  estimated   quantitation  and
         detection limits.

       ACTION:     Use  professional judgement  to
       determine  acceptability of the data  if the  above
       criteria are not all met. If data are considered to
       be  unacceptable, the  tentative ID should  be
       changed to "unknown".
 5.7    Compound Ouantitation and Reported Detection
        Jrnit
        1. Verify that the reported values, both positives
          and non-detects, have been correctly adjusted to
          reflect  all  dilutions,   concentrations,  splits,
          cleanup procedures,  dry weight factors, an any    5.9
          other adjustments that have not been accounted
          for by the method.

          VGA for waters:  ug/L =
                                          .
                                  (Ais)(RF)(V0)
          Low level VOA for soils:
                            ug/kg =
                                    (Ais)(RF)(W5)(D)
          High level VOA for soils:
                            ug/kg = {
                                  (Ais)(RF)(Ws)(D)(Vi)
Au =

Is  =
RF =

V0 =
Ws =
D =
Vt =
              area of characteristic ion for compound being
              measured
              area of characteristic; ion for the internal
              standard           ,|
              amount of internal standard added (ng)
              daily response factor . for compound being
              measured           j
              volume of water purged (ml)
              weight of sample extracted (g)
              (100 - % moisture)/]100 or 1  on wet weight
              basis                ;
              volume of total extract  (ul)
          =   volume of extract added (ul) for purging
       ACTION: If incorrect values
       is  essential that the correct
       The reviewer should contact
                            have been reported, it
                            values be determined.
                            :he laboratory to verify
       any corrections made to the data.
5.8    Performance Evaluation Samples

       1. Were recovery limits within  those  set  by the
          EMSL lab?              i
                                  1
                                  I
                                  I
       ACTION:    If  outside  the! limits,  review on  a
       compound by  compound  b'asis.   If 50%  of the
       compounds are outside of confidence limits or were
       misidentified, all sample results should be rejected
       (R).
        Overall Assessment of Data I
                                  [
        It is appropriate  for  the Idata  reviewer  to  use
        professional judgment  and i express  concerns  and
        comments on the validity of the overall data package
        for a case. This is particularly appropriate for cases
        in  which there  are  several  QC criteria  out of
        specification.   The additive nature of  QC factors
        which are out of specificatioh is difficult to assess in
        an  objective  manner,  but| the  reviewer   has  a
        responsibility to inform the user about  data quality
                                                     32

-------
        and data limitations.  This helps the user to avoid
        using data inappropriately, while not precluding
        consideration of the data.  The data reviewer
        would be greatly assisted in this endeavor if the
        data quality objectives were provided.
 5.10    Optional PC Checks

 5.10.1  Surrogate Recovery

        1. If either one or more VOA surrogates were
          outside  of  specifications for any sample  or
          blank,   were  the   appropriate   samples
          reanalyzed?

       ACTION:  If initial  analysis and reanalysis both
       have   two   or  more  surrogates  outside   of
       specifications for samples or blanks,  estimate (J)
       all quantitation results, including detection limits.

       2. Does  any  surrogate  have  less  than 10%
          recovery?

       ACTION:   If yes,  flag as estimated  (J) positive
       results  for  that fraction; flag negative results as
       rejected (R).
6.0    Pesticides/PCBs

6.1     Sample Holding Times

       1. Were  any  of  the  sample  holding  times
         exceeded?

         Sample  Holding Times from date  of sample
         collection:

         Water -  7 days to extraction
         Soil, sludge, sediment - 14 days to extract

         All - analyze within 40 days after extraction
        ACTION: If yes, flag as estimated (J) those values
        above the Instrument Detection Limit (IDL), based
        on the reviewers professional judgement  and the
        nature of the sample and analyte.  Values that are
        less than the IDL can be flagged as estimated (UJ)
        or rejected (R) based on the reviewers professional
        judgement and the nature of the sample and analyte.
       Because of  their  long shelf lives,  performance
       evaluation  samples  do  not have  any associated
       holding times.
6-2     Instrument Performance

       1. Check the raw data to verify that DDT retention
         time is greater than 12 minutes on the standard
         chromatogram  and  that  there  is  adequate
         resolution  (>  25%)  between  peaks  of other
         pesticide standard compounds.

       ACTION: If the retention time of DDT is less than
       12 minutes (except on OV-1 and OV-101),  a close
       examination of the chromatography is necessary to
       ensure that  adequate  separation  of  individual
       components is achieved.  If adequate separation is
       not  achieved, flag all affected compound  data as
       rejected (R).

       2. Check  raw  data  to  verify  that retention time
         windows are  reported  and that all pesticide
         standards are within the established retention time
         windows.

      ACTION:  If the  standards do not  i'all within  the
      retention  time  windows,  professional  judgement
      should be used in the evaluation of associated sample
      results.

      3. Check the raw data  to verify that the percent
        breakdown for endrin and 4,4'-DDT, or the
                                                   33

-------
      combined percent breakdown,  does not exceed
      20% in all Evaluation Standard Mix B analysis.

      ACTION: If the DDT breakdown is greater than
      20%, beginning with the last in-control standard,
      flag all results for DDT as estimated (J). If DDT
      was not detected,but DDD and DDE are positive,
      then flag the quantitation limit for DDT as rejected
      (R).   Flag  results for  DDD  and/or DDE  as
      presumptively  present at  an estimated quantity
      (NJ).

      If the endrin breakdown is greater than 20%, flag
      all quantitative results for endrin as estimated (J).
      If endrin was  not detected, but endrin aldehyde
      and  endrin ketone  are  positive,  then flag  the
      quantitation limit for endrin as rejected (R). Flag
      results for endrin ketone as presumptively present
      at an estimated quantity (NJ).

      4. Check the raw data to verify that the percent
         difference in retention time for  the surrogate
         dibutylchlorendate (DEC) in all standards and
         samples in <. 2.0% for packed column analysis,
         <. 0.3%  for capillary  column analysis, and .<.
         1.5% for wide-bore capillary column analysis.

      ACTION:  If any of the percentages are  greater
      than indicated, the analysis may be flagged as
      rejected (R) for that sample. Qualification of the
      data is left up to the professional judgement of the
      reviewer.
63     Initial and Continuing Calibration Verification

       1.  Verify that the %RSD of the calibration factor
          for aldrin, endrin, DEC and DDT are less than
          or equal to  10%  for the initial  calibration
          linearity check.

       ACTION:  If criteria for linearity is not met, flag
       all associated quantitative results as estimated (J).
  %RSD  = s x 100
             x
  where:
  s = standard deviation of 5 response factors
  x = mean of 5 response factors
2. If toxaphene or DDT series was identified and
  quantitated, verify that a three-point calibration
  was established.

ACTION: If no, flag as estimated (J) positive results
for toxaphene or DDT.      ;
                            I'
3. Verify the proper 72-hour analytical sequence as
  follows:                   |
                            |
  Standard Mix A, Standard Mix B, Standard Mix
   C (individual standard mix A, individual standard
   mix B, may be one mix), Toxaphene, Aroclors
   1016/1260, (Aroclor 1221, Aroclor 1232, once per
   month), Aroclor 1242, Aroclor 1248, Aroclor 1254,
   5 samples, Standard Mix B,i5 samples, Individual
   Standard Mix A or  B, 5 samples, repeat starting
   from Standard Mix B, must end with individual
   Standard Mix A and B.    !
                            [
                            I
 ACTION:   If the proper standards have not been
 analyzed and the sequence followed, use professional
 judgement to determine the severity of the effect and
 qualify the data accordingly.  |
                            i
 4. Review the pesticide sample1 data to verify whether
   the standard was used as a' quantitation standard
   or as a confirmation standard.
                            i
 ACTION:  If the  %D for standard analysis is greater
 than 15% on the quantitation column or greater than
 20% on the confirmation column, flag all associated
 positive sample results as estimated (J).
                                                     34

-------
6.4    Error Determination

       See Part I - Section 2.8 for QA samples to be used
       for error determination.

6.4.1   Determination of Bias (% Recovery - Optional for
       QA-2; Mandatory for QA-3)

       6.4.1.1   Percent Recovery

               1. Were  at  least eight spiked sample
                  replicates for the matrix of Interest
                  analyzed at the required frequency?

               ACTION:   If no, flag  as recovery not
               determined TRND)  all data  for which
               spiked samples were not analyzed.

               2. Determine the average recovery of the
                  eight spiked replicates. Is the average
                  recovery within the applicable control
                  limits (80% to 120%)?

       % recovery for a single spiked sample =
         Spiked sample cone. - Sample  cone.  x ^QQ
                          Spike cone,  added

               ACTION:     If  recoveries  are  within
               applicable  control limits, no bias  is
               considered.  If %  Recovery is less than
               80% or greater than 120%, the sample
               data should be flagged with a (J) estimate
               and a corresponding (-) or (+) sign to
               show direction of the bias. Adjustment of
               sample  values  should  be  considered
               whenever there is consistent evidence of
               bias.

      6.4.1.2   Adjustment of Sample Values for Bias
                                                             values. % bias is the reciprocal value of
                                                             % recovery (i.e., for 70% recovery you
                                                             have a negative 30%  bias).  Use the
                                                             average % recovery from the total number
                                                             of  matrix  spikes   analyzed.     This
                                                             adjustment approach assumes a spiking
                                                             concentration equal to the concentration
                                                             found in the sample.
                                          6.4.2   Determination of Precision  (Optional  for  QA-2;
                                                 Mandatory for QA-3)

                                                 6.4.2.1 Replicate Analysis

                                                        1. Was  a minimum  of eight  replicates
                                                           analyzed?  If yes, determine coefficient of
                                                           variation.  If no, flag data with precision
                                                           not determined (PND}. for which replicate
                                                           samples were not analyzed.

                                          6.4.2.2 Coefficient of Variation (Percent Relative Standard
                                                 Deviation)

                                                        1. The coefficient of variation (CV) is used
                                                           in determining the precision of standard
                                                           deviation. The CV expresses the standard
                                                           deviation as a percentage of the  mean
                                                           (average) value  of  the replicate values.
                                                           The  CV  is  used to  determine a  false
                                                           positive or false negative value for results
                                                           that are respectively greater than or less
                                                           than a decision level concentration.

                                                           Determine  the  coefficient of variation
                                                           using the following equation:

                                                                CV =  s  x 100
                                                                       XDL
1.  Depending on bias direction, add or
   subtract the value (% Bias x  spike
   concentration) to or from the sample
                                     35

-------
          where:
          X   = the decision level concentration
           DL
          s  =  the  sample standard  deviation
          given by the equation:
  *Note: When using a programmable calculator
  or computer statistics software,  be sure the
  above equation with (n - 1) is used and not (n)
  by itself. The equation using (n) is to determine
  the population standard deviation (a) rather
  than the sample standard deviation (s).

  Apply the CV to the decision level to determine
  the false negative  or false positive value as
  follows:

  False positive value = Decision level value  +
  (CV x decision level)

  False negative value  = Decision level value -
  (CV x decision level)

Example:

  For an decision level = 50 ppm and CV = 20%

  False positive value = 50 ppm + (20% x
  50 ppm)
           = 50 ppm + (10 ppm)
           = 60 ppm

   False negative value = 50  ppm - (20% x
   50 ppm)
           = 50 ppm - (10 ppm)
           = 40 ppm

For the  above false positive example, any value
between 50  ppm  and 60 ppm are considered
suspect and should be reanalyzed. Values above 60
      ppm are considered actionable.  In many cases, false
      positives have  been considered actionable  by the
      Agency for safety reasons.  Hoiwever, depending on
      the  action  to  be taken, this]  can be costly  and
      unjustifiable.  Consult the QA |plan for  intended use
      of data and data quality objectives.
                                  i.
      For the above false  negative j example, any values
      between 40 ppm and 50 ppm are considered  suspect
      and should be reanalyzed. Values below 40 ppm are
      considered  non-actionable.  |In most cases, the
      decision maker will be using th'e false negative value
      as his decision level and not be concerned about the
       false positive value.  Whenever sample values need
       to be corrected  for  both  bias and precision, first
       correct  the value for bias,  then correct the biased
       value for precision.          |
6.5    Blanks                      |
                                   !
       1. Verify  that  method blank  analysis  has  been
          reported  per  matrix, per concentration level, at
          the proper frequency, for  each GC system used
          to analyze samples, for each extraction batch.
       ACTION:   If  the  proper type and  frequency of
       method  blank  have  not , been  analyzed,  use
       professional judgement to determine  the  effect on
       the data.    -                I
        2. Verify that all blank analyses contain less than the
          Required Detection Limits (RDL) of any pesticide
          or interfering peak.
        ACTION:  Any pesticide detected in the sample and
        also  detected  in  any  associated blank,  must  be
        qualified  as  non-detect (U)  when  the  sample
        concentration   is   less  than  5X   the   blank
        concentration.
        NOTE:  In instances where more than one blank is
        associated with a given sample, quantification should
                                              36

-------
       be based upon a comparison with the associated
       blank  having  the highest  concentration  of  a
       contaminant. The results must not be corrected by
       subtracting any blank value.
6.6     Compound Identification

       1. Verify that positive identifications have GC/MS
         confirmation or dissimilar column analysis (the
         3%  OV-1  column  cannot   be  used  for
         confirmation if both  dieldrin and  DDE  are
         identified).

       ACTION:   If  the qualitative  criteria  for dual
       column or GC/MS confirmation were not met, all
       reported positive  results should be flagged as
       presumptively present  at an estimated quantity
       (NJ).

       2. If   multipeak   pesticides   (chlordane   and
         toxaphene)/PCBs were reported,  were  the
         retention times and relative peak height ratios
         of major component peaks compared against
         the appropriate standard chromatograms.

      ACTION:  If multipeak pesticides/PCBs exhibit
      marginal pattern-matching quality  professional
      judgement should be used to establish whether the
      differences  are attributable  to  environmental
      "weathering".    If  the presence of  a multipeak
      pesticide/PCB is strongly suggested, results should
      be reported as presumptively present (N).

      3. Verify that  the  sample  chromatogram  agree
        with the correct  daily standard chromatogram,
        and that the retention time windows match.

      ACTION: If the chromatograms do not agree, and
      the retention time windows vary significantly, the
      reviewer  must  use professional judgement  to
      determine the flags that should be applied and the
      usefulness of the data.
6-7     Compound Ouantitation and Reported  Detection
       1. .Verify that the reported values, both positives and
         non-detects, have been correctly adjusted to reflect
         all   dilutions,  concentrations,   splits,  cleanup
         procedures, dry weight  factors, an  any other
         adjustments that have not been accounted for by
         the method.
         Pesticide/PCBs for waters: ug/L =
         Pesticide/PCBs for soils: ug/kg  =
                                        (As)(Ws)(D)(Vi)
      AX = area of quantitation peak(s)
      Is =  amount of standard injected (ng)
      Vt< =  volume of total extract (ul)
      Vj =  volume injected (ul)
      Vs =  volume of sample (ml)
      Ws = weight of sample extracted (g)
      D  =  (100 - % moisture)/100 or 1 for wet weight
      basis
      As =  Area of external standard

      ACTION: If incorrect values have been reported, it
      is essential that the correct values be determined.
      The reviewer should contact the laboratory to verify
      any corrections made to the data.
                                                  37

-------
6.8    Performance Evaluation Samples

       1. Were recovery limits within those set by the
         EMSL lab?

       ACTION:   If outside the limits, review  on  a
       compound by compound  basis.  If  50%  of the
       compounds are  outside of confidence limits or
       were misidentified, all sample  results should be
       rejected (R).
 6.9     Overall Assessment of Data

        It is appropriate  for  the  data  reviewer to use
        professional judgment and express concerns and
        comments on the validity of the  overall data
        package for a case. This is particularly appropriate
        for cases in which there are several QC criteria out
        of specification. The additive nature of QC factors
        which are out of specification is difficult to assess
        in an objective manner, but the  reviewer has a
        responsibility to inform the user about data quality
        and data  limitations.  This helps the user to avoid
        using data inappropriately, while not precluding
        consideration  of  the  data.  The  data reviewer
        would be greatly assisted in this endeavor if the
        data quality objectives were provided.
 6.10   Optional OA Checks

 6.10.1 Snrrnorata Recovery

        1. Verify that the recoveries are within the control
           limits.

        ACTION:  If not, check the raw data for possible
        interferences.
                                                               2. If  recoveries  are out  of  control  limits,  use
         professional  judgement
         appropriate action.
to  determine   the
       ACTION:   If zero surrogate, pesticide recovery is
       reported, determine whether tjhe surrogate is outside
       its retention time window,  it yes, use professional
       judgement  in the evaluation1 of this data.   If the
       surrogate is not present, flag; all negative results as
       rejected (R).
7.0    PCBs

7.1    Sample Holding Times

       1. Were any of the sample holding times exceeded?
                                  I
          Sample  Holding  Times  ifrom  date of sample
          collection:               !
          Water - 7 days to extract [,
          Soil, sediment, sludges - l|4 days to extract
                                  i
          All - analyze within 40 days after extraction
        ACTION:  If yes, flag as estimated (J) those values
        above the Instrument Detection Limit (IDL), based
        on the reviewers professional judgement and the
        nature of the sample and arialyte. Values that are
        less than the IDL can be flagged as estimated (UJ)
        or rejected (R) based on thp reviewers professional
        judgement and the nature ofj the sample and analyte.
                                  i
                                  l
        Because  of their  long  sh'elf  lives, performance
        evaluation  samples  do  no|t have  any associated
        holding times.
                                                          1.1    Instrument Performance   !

                                                                 1. Examine  standard  chromatograms  to  assure
                                                                    adequate quantitation pe'ak resolution.
                                                      38

-------
       ACTION: If there is inadequate peak separation
       (<25% quantitation peak resolution), flag the data
       as rejected (R).

       2. Examine raw data and spot check the surrogate
         compound retention times.

       ACTION:  If the  retention time  shift for  the
       surrogate compound  exceeds 2.0% for packed
       columns, 0.3% for  capillary columns,  1.5% for
       wide-bore capillary columns,  the  data may be
       rejected (R), but the qualification is left up to the
       professional judgement of,the reviewer.
73     Initial and Continuing Calibration Verification

       1. Verify that the Aroclors of interest have been
         analyzed  at  a  minimum  of three  different
         concentrations (e.g., Aroclor 1260 analyzed at
         1.0, 5.0 and 10.0 ppm).

       ACTION: If no, flag data as estimated (J).

       2. Verify that the %RSD of the calibration factor
         for all Aroclors is less than or equal to 10% for
         the initial linearity check.

         %RSD = sx 100
                    x

         where:

         s = standard deviation of 5 response factors
         x = mean of 5 response factors

       ACTION: If criteria for linearity is not met, flag
       all associated quantitative results as estimated (J).

       3. Verify that the continuing calibration for each
         Aroclor of interest was  analyzed daily.

       ACTION: If no, flag all associated sample results
       as estimated (J).
       4. Verify %D between calibration factors.

       ACTION: If the %D for standard analysis is greater
       than 15% on the quantitation column or greater than
       20% on the confirmation column, flag all associated
       positive sample results as estimated (J).
7.4    Error Determination

       See Part I - Section 2.8 for QA samples to be used
       for error determination.

7.4.1   Determination of Bias (% Recovery - Optional for
       QA-2; Mandatory for QA-3)

       7.4.1.1  Percent Recovery

              1. Were  at  least   eight  spiked  sample
                replicates  for  the  matrix  of  interest
                analyzed at the required frequency?

          ACTION:  If no, flag as recovery not determined
          (RND) all data for which spiked samples were not
          analyzed.

  .  .'. .        2. Determine the average  recovery of the
                eight spiked  replicates.   Is the average
                recovery within  the applicable  conlrol
                limits (80% to 120%)?

          % recovery for a single spiked sample  =
              Spiked sample cone. - Sample cone. x ^QQ
                      Spike cone,  added

          ACTION:   If recoveries  are within  applicable
          control limits,  no  bias  is  considered.    If %
          Recovery is  less  than  80%  or greater  than
          120%,the sample data should be flagged with a
          (J) estimate and a corresponding (-) or ( + ) sign
          to show direction of the bias.   Adjustment of
          sample values  should be considered  whenever
          there is consistent evidence of bias.
                                                    39

-------
r
                     7.4.1.2  Adjustment of Sample Values for Bias

                             1. Depending on bias direction, add or
                                subtract  the  value (%  Bias x spike
                                concentration) to or from the sample
                                values. % bias is the reciprocal value
                                of % recovery (i.e., for 70% recovery
                                you have a negative 30% bias).  Use
                                the average % recovery for the total
                                number  of  matrix spikes analyzed.
                                This adjustment approach assumes a
                                spiking  concentration  equal to  the
                                concentration found in the sample.

              7.4,2   Determination of Precision (Optional for  QA-2;
                     Mandatory for QA-3)

                     7.4.2.1   Replicate Analysis

                             1. Was a minimum of eight  replicates
                                analyzed? If yes, determine coefficient
                                of variation.   If  no,  flag data with
                                precision not  determined fPNDV  for
                                which replicate  samples  were  not
                                analyzed.

                     7.4.2.2   Coefficient of Variation (Percent Relative
                             Standard Deviation)

                             1. The coefficient  of variation (CV)  is
                               used in determining the precision of
                               standard deviation. The CV expresses
                               the standard deviation as a percentage
                               of the mean  (average) value of the
                               replicate values.  The  CV is used to
                               determine a  false positive or  false
                               negative value for  results  that  are
                               respectively greater than or less than
                               a decision level concentration.

                               Determine the coefficient of variation
                               using the  following equation:
                 CV =  s  x.lOO
          where:
XDL  =
                     decision level concentration
          s =  the sample standard deviation given
          by the equation:  |
                                   ,14
             - x)f/(n , 1)]
        *Nqte:    When  using  a  programmable
        calculator or computer statistics software, be
        sure the above equation with (n - 1) is used
        and not (n) by itself. I The equation using (n)
        is  to  determine  the  population standard
        deviation (a)  rather than the sample standard
        deviation (s).

        Apply the  CV  to  jthe  decision level to
        determine the false negative or false positive
        value as  follows:     j
                          . [
        False positive value = Decision level value +
   r"  *  (CV x decision level)
   f - ; .      . -    . - •.   .    •• ,\: .    .- •
        False negative value  =  Decision level value -
      -. •  (CV x decision level)
 .,--....•      .       |:'     - •••  .
Example:                  ! •
                           r
                           i
       For an decision level = 50 ppm and CV =
•.   • •  20%                j:
                           |
False positive value = 50 ppm + (20% x 50 ppm)
                = 50 ppm 4- (10 ppm)
                = 60 ppm  !
False negative value  = 50 ppm - (20% x 50 ppm)
                = 50 ppm -j (10 ppm)
                = 40 ppm  |.
                           I.
For the above  false positive example,  any value
between 50 ppm and 60 ppm fare considered suspect
                                                                  40

-------
,7.5
       and should be reanalyzed. Values above 60 ppm
       are considered actionable.  In many cases, false
       positives have been considered actionable by the
       Agency for safety reasons. However, depending on
       the action to be taken,  this can be  costly and
       unjustifiable.  Consult the QA plan for intended
       use of data and data quality objectives.

       For the above false negative example, any values
       between 40 ppm  and 50 ppm  are  considered
       suspect and should be reanalyzed.  Values below
       40 ppm are considered non-actionable.  In most
       cases, the  decision maker will be using the false
       negative value as his  decision level and not be
       concerned  about   the   false   positive  value.
       Whenever sample values need to  be corrected for
       both bias and precision, first  correct the value for
       bias, then  correct the biased value for precision.
 Blanks

 1. Verify that method blank analysis  has been
,  , reported per matrix, per concentration level, at
   the proper frequency, for each GC system used
   to analyze samples, for each extraction batch.

 ACTION:   If the proper type and  frequency of
 method blank  have  not  been  analyzed,  use
 professional judgement to determine the effect on
 the data.
                                                         NOTE:  In instances where more than one blank
                                                         is  associated with  a given sample, quantification
                                                         should -be  based upon  a comparison  with  the
                                                         associated blank having the highest concentration of
                                                         a contaminant.  The results must not be corrected by
                                                         subtracting any blank value.
7.6    Compound Identification

       1. Review the data to  confirm that positive results
         were identified  using the correct retention time
         window,  peak  height  ratio,  and  "fingerprint"
         pattern.  Determine which peak(s) were used to
         quantitate each Arbclor and verify that the finger-
         print pattern matches the standard chromatogram.

       ACTION: If the reported positive results were  not
       identified correctly, professional judgement should
       be used to qualify  the data.

       2. Verify that  dual column confirmation of positive
         results identify the same Aroclor or that the  lab
         performed GC/MS  confirmation of PCB results
         that were greater than 10 ng/ul.

       ACTION: If the .qualitative criteria for  dual column
       or GC/MS confirmation were not met, all reported
       positive results should  be flagged as presumptively
       present at an estimated quantity (NJ).
       2. Verify that all blank analyses contain less than
         the Required  Detection Limits (RDL) of any
         PCB or interfering peak.

       ACTION: Any PCB detected in the sample and
       also  detected in any associated blank, must be
       qualified  as non-detect (U) when  the  sample
       concentration  is less  than  5  times  the blank
       concentration.
                                                    41

-------
 7.7    Compound Ouantitation and Reported Detection
       Limits

       1. Verify that the reported values, both positives
          and non-detects, have been correctly adjusted to
          reflect  all  dilutions,  concentrations,  splits,
          cleanup procedures, dry weight factors, and any
          other adjustments that have not been accounted
          for by the method.
          PCBs for waters:  ug/L = fA..₯LW,1
          PCBs for soils:  ug/kg = (AjfLVV,)
                                 (As)(Ws)(D)(Vi)
 7.9     Overall Assessment of Data
                                  I
        It  is  appropriate  for  tie I data  reviewer to use
        professional judgment  and! express  concerns and
        comments on the validity of the overall data package
        for a case. This is particularly appropriate for cases
        in  which there  are several QC criteria out  of
        specification.   The  additive, nature of QC factors
        which are out of specification is difficult to assess in
        an  objective  manner, but the  reviewer has  a
        responsibility to inform the user about data quality
        and data limitations.  This helps the user to  avoid
        using  data  inappropriately, while not  precluding
        consideration of the data.  The data reviewer would
                                  i.
        be greatly assisted in this endeavor if the data quality
        objectives were provided.    !
       Ax = area of quantitatibn peak(s)
       I, =  amount of standard injected (ng)
       V, «• volume of total extract (ul)
       Vj = volume injected (ul)
       V, = volume of sample (ml)
       W, = weight of sample extracted (g)
       D = (100 - % moisture)/100
       A, = Area of external standard

       ACTION: If incorrect values have been reported,
       it is essential that the correct values be determined.
       The  reviewer should contact  the laboratory to
       verify any corrections made to the data.
7.8    Performance Evaluation Samples

       1.  Were recovery limits within those set by the
          EMSL lab?

       ACTION:   If  outside the  limits,  review on  a
       compound  by compound  basis.   If 50% of the
       compounds  are outside of  confidence  limits or
       were misidentified,  all sample results should be
       rejected (R).
7.10    Optional PC Checks       j
                                  1

7.10.1  Surrogate Recovery        !
                                  i

        1. Verify that the recoveries are within the control
          limits.                   !
                           • • •     i
 •'           •                     i
   .     ACTION:  If not, check the raw data for possible
;-.,. .      interferences.
       2. If  recoveries  are out  of  control  limits,  use
          professional  judgement •  to  determine   the
          appropriate action.       |
       ACTION:  If zero surrogate pesticide recovery is
       reported, determine whetherithe surrogate is outside
       its retention time window.  If yes, use professional
       judgement in  the  evaluation  of this data.  If the
       surrogate  is not present, flag all negative  results as
       rejected (R).               !
                                                     42

-------
8.0    2,3,7,8-TCDD
                                         •m/z
                Ion abundance criteria (continued)
8.1    Sample Holding Times

       1.  Were any  of  the  sample  holding  times
          exceeded?*

          To extract - 6 months from sample collection

          To analysis - 40 days from extraction

       ACTION: If yes, flag as estimated (J) those values
       above  the  Instrument Detection Limit (IDL).
       Values that are less than the IDL can be flagged
       as  estimated  (UJ) or rejected (R) based on the
       reviewers professional judgement and the nature of
       the sample and  analyte.

       *Because of their long shelf lives, performance
       evaluation samples do not have any associated
       holding times.


82     Instrument Performance

       1.'Verify that a performance'check "solution was
         run at the beginning of each 8-hour shift and at .
         the end of the final 8-hour period.

       ACTION:  If no, use professional judgement to
       qualify data.

       2. Have the ion abundance criteria been met for
         each instrument used?
         m/z

         51
         68
         69
         70
         127
Ion abundance criteria

30-60% of mass 198
Less than 2% of mass 69
(reference only)
Less than 2% of mass 69
40-60% of mass 198
                                          197
                                          198
                                          199
                                          275
                                          365
                                          441
                                          442
                                          443
               Less than 1% of mass 198
               Base peak, 100% relative abundance
               5-9% of mass 198
               10-30% of mass 198
               Greater than 1% of mass 198
               Present but less than mass 443
               Greater than 40% of mass 198
               17-23% of mass 442
                                       ACTION: If no, use professional judgement to flag
                                       all associated data.

                                       3. Is the resolution of the valley between 2,3,7,8-
                                         TCDD and the peak representing all other TCDD
                                         isomers <. 25%?  (where, Valley (%)  = X/Y x
                                         100 and X is measured from the valley of the least
                                         resolved adjacent  isomer to  the  baseline,  Y =
                                         peak height of 2,3,7,8-TCDD).

                                       ACTION:   If no,  use professional judgement to
                                       qualify all positive  sample data  associated with the
                                       standard.
Initial Calibration

1. Verify the following:

  -the five 2,3,7,8-TCDD standards have been run.

  -the ratios of ions 320 to 322 for 2,3,7,8-TCDD
  and 332 to 334 for 13C!2-2,3,7,8-TCDD is  >. 0.67
  and <_ 0.87.

  -signal-to-noise ratios for ions 257, 320, 322 and
  328 is >_ 2,5 and the signal to noise ratios for ions
  332 and 334 is >. 10.

  -the ions 257, 320, 322 for 2,3,7,8-TCDD reached
  a maximum within three seconds of 13C12-TCDD
  ions 332 and 334.
                                                   43

-------
         -during the unlabeled 2,3,7,8-TCDD calibration
         the  percent  Relative  Standard  Deviation
         (%RSD) of relative response factors for the five
         calibration concentrations is less than or equal
         to 15%.

         -during the ^Cl^SJ.S-TCDD  calibration the
         %RSD of relative response factors for the three
         calibration concentrations is less than or equal
         to 15%.

       ACTION: If the calibration curve standards fail
       the acceptance criteria, use professional judgement
       to qualify associated data.
8.4     Continuing Calibration

       1. Verify the following:

         -the calibration standard has been run for every
         eight hour shift.

         -the ratios of ions are 320 to 322 for 2,3,7,8-
         TCDD and 332 to 334 for 13C12-2,3,7,8-TCDD >.
         0.67 and <. 0.87.

         -the signal to noise ratios for ions are 257, 320,
         322 and 328 >. 25 and the noise ratios for ions
         332 and 334 >. 10.

         -the ions  are  257, 320, 322 for 2,3,7,8-TCDD
         reached a maximum within  three seconds of
         13C,2-TCDD ions 332 and 334.

         -the percent difference of the relative response
         factor is ±. 30% of the initial calibration.

       ACTION:  If the calibration standard fails  the
       above  acceptance   criteria,  use   professional
       judgement to qualify associated  data.
8.5    Error Determination      i
                                 t
       See Part I - Section 2.8 for QA samples to be used
       for error determination.    J
                        =   •      i.
8.5.1   Determination of Bias (% Recovery - Optional for
       QA-2; Mandatory for QA-3)
                                 I
       8.5.1.1 Percent Recovery   |
                                 j
              1.  Were   at   least  ; eight  spiked  sample
                 replicates  for the matrix  of  interest
                 analyzed at the required frequency?

              ACTION:   If no, [flag  as  recovery  not
              determined (RND^ all data for which spiked
              samples were not analyzed.
                                 \
                                 I
              2.  Determine  the average  recovery of the
                 eight spiked replicates.   Is  the  average
                 recovery  within  the applicable  control
                 limits (80% to 120%)?
                                 i -
% recovery for a single spiked sample =
                   -              i'
       Spiked sample cone. - Sample cone.  x JQQ
                     Spike cone, added
                                 i •
             , ACTION: If recoveries are within applicable
              control limits, no bias is considered.  If %
              Recovery is less than 80% or greater than
              120%, the sample data should be flagged with
              a (J) estimate and a corresponding (-) or (+)
              sign   to   show  direction  of  the   bias.
              Adjustment  of  sample values  should  be
              considered whenever  there  is consistent
              evidence of bias.    '
                                 i
                                 i
       8.5.1.2 Adjustment of Sample Values for Bias
                                 I
              1.  Depending on  bias direction,  add  or
                 subtract the  value  (% Bias  x  spike
                 concentration) to,  or  from  the  sample
                 values.  % bias is the reciprocal value of
                                                    44

-------
                 % recovery (i.e., for 70% recovery you
                 have  a negative 30% bias).  Use the
                 average  %  recovery from the  total
                 number  of  matrix  spikes analyzed.
                 This  adjustment approach  assumes  a
  ,               spiking  concentration  equal  to  the
                 concentration found in the sample.

 8.5.2   Determination of Precision (Optional for QA-2;
        Mandatory for QA-3)

        8.5.2.1   Replicate Analysis

                1. Was a minimum of eight replicates
               ,  analyzed?      If   yes,   determine
   '         ,      coefficient  of variation.   If no, flag
                 i data with, precision not determined
                  CPND). for which replicate samples
                  .were not analyzed.

        8.5.2.2   Coefficient   of   Variation   (Percent
                Relative Standard Deviation)

                1. The  coefficient of variation (CV) is
                  used in determining the precision of
                  standard .deviation. The CV expresses
                  the standard deviation as a  percentage
-••'."•           of the  mean {average)  value  of the
   '••.*••          replicate values.  The CV is used to
                  determine  a false  positive  or false
   ;,     •          negative value  for results that are
                  respectively greater than or less than
                  a decision level concentration.

                  Determine the  coefficient of variation
                  using the following equation:

   :                     CV  = s .x-100
           where:    .    ; '  •

           XDL  .= .the decision level concentration

           s = the sample standard deviation given by
           the equation:
               '      ' '     *
           s*  =     [ft - x)2/(N - l)f2
   *Note:  When using a programmable calculator
   or computer statistics software, be sure the above
   equation with (n - .1) is used and not (n) by itself.
   The  equation using  (n)  is  to determine  the
   population standard deviation (a) rather than the
   sample standard deviation (s).
   Apply the CV to the decision level to determine
   the false negative or false positive value as follows:

   False positive value = Decision level value + (CV
   x decision level)

•   False negative value  = Decision level value - (CV
   x decision level)   ' ,,

.^Example:.      .

   For an decision level = 50 ppm and CV = 20%

   False positive value = 50 ppm.+ (20% x 50 ppm)
           =: 50 ppm  + (10 ppm)
           =  60 ppm
   False negative value = 50 ppm - (20% x
                nnm - (10 ppm)
  50 ppm
= 40 ppm
                              50 ppm)
                                                                For  the  above..false positive example,  any value
                                                                between 50 ppm and 60 ppm are considered suspect
                                                                and should be reanalyzed. Values above 60 ppm are
                                                                considered actionable.  In many cases, false positives
                                                     45

-------
       have been considered actionable by the Agency for
       safety reasons. However, depending on the action
       to be taken, this can be costly and unjustifiable.
       Consult the QA plan for intended use of data and
       data quality objectives.

       For the above false negative example, any values
       between 40 ppm  and 50  ppm  are considered
       suspect and should be reanalyzed. Values below
       40 ppm are considered non-actionable. In most
       cases, the decision maker will be using the  false
       negative value as  his decision level and not be
       concerned  about  the  false   positive  value.
       Whenever sample values need to be corrected for
       both bias and precision, first correct the value for
       bias, then correct the biased value for precision.
8.6    Blanks
       1.  Has a method blank, spiked with the internal
          standards, been analyzed with each case?

       ACTION:    If  the  method  blank  contains
       contaminants at the method detection limit of the
       matrix of interest, the blank must be reanalyzed.
       If the contaminated method blank was extracted
       along with  a  batch of samples the associated
       positive samples  must  be reanalyzed.  If  the
       samples were not reanalyzed or if contamination is
       present in the second analysis, all positive sample
       results less than 5 tunes the concentration in the
       blank are flagged as non-detects (U).

       2.  Has a reagent  blank been analyzed along with
          each case?

       ACTION:  The reagent blank should be free of
       contamination.  If the level is >  0.10  ppb,  use
       professional judgement to qualify associated data.
8.7    Internal Standard Requirements
                                  I

       1. Did ion 332 or 334 fail the relative ion intensity
         criteria (.>  0.67  and <. 0.87)?   If yes, was the
         sample reanalyzed?       '

       ACTION: If initial analysis and reanalysis both have
       ions 332 or 334 outside the relative ion intensity
       criteria,  reject all  quantitation results, including
       detection limits.            I
8.8    Identification of 23.7.8-TCDD
       1. Verify the following:
          -the retention time of the, sample component is
          within three seconds of the retention tune of the
          13C12 - 2,3,7,8-TCDD.     j
                                  I '
                                  [
          -the integrated ion currents detected for m/z 257,
          320, and 322 maximize simultaneously.
                                  i
                                  i
          -the ion ratio of 320 to 322| and 332 to 334 is >. .67
          and < .87.               !
                   • :              i
                                , • .1.
          •  .           ,   -      .;:!.', '    •••
          -the integrated ion current for each analyte and
          surrogate compound (m/z; 257, 320, 322 and 328)
          are at least 2.5 times background noise.
                                  [
                                  I
          -internal  standard  ions  are  at  least  10 times
          background noise. (The integrated ion current or
          the internal standard ions must not saturate the
                                  i-
          detector.)
                                  i

          -if the above requirements were not met, then
          reanalyze the samples.

       ACTION: If initial analysis aiid reanalysis both have
       the sample outside  the abovp limits, 2,3,7,8-TCDD
       was not qualitatively identified, reject (R) all positive
       results.                     i
                                                     46

-------
       ACTION:  If no, reject all quantitation results,
       including detection limits.
 8.9    Performance Evaluation Samples

       1. Were recovery limits within those set by the
          EMSLlab?
        ACTION:   For  37 Cl4-2,3,7,8-TCDD  the  ion 328
        must have a signal to noise ratio  of  >_ 2.5.  The
        surrogate recovery must be >_ 60 and <_ 140 percent.
        If the signal to noise ratio for ion 328 does not meet
        acceptance criteria, reject positive and ND data. If
        surrogate recovery is outside acceptance limits, use
        professional judgement to qualify associated data.
       ACTION:   If  outside  the  limits,  review on  a
       compound by compound basis.  If 50% of the
       compounds are outside of  confidence limits  or
       were misidentified, all sample  results should be
       rejected (R).
8.10   Overall Assessment of Data

       It  is  appropriate for the data reviewer to use
       professional judgment and express concerns and
       comments  on the  validity  of  the  overall data
       package for a case. This is particularly appropriate
       for cases in which there are several QC criteria out
       of specification. The additive nature of QC factors
       which are out of specification is difficult to assess
       in  an objective manner,  but  the reviewer has a
       responsibility to inform the user about data quality
       and data limitations.  This helps the user to avoid
       using data  inappropriately, while  not precluding
       consideration of the data.   The data reviewer
       would be greatly assisted in this endeavor if the
       data quality objectives were provided.
8.11    Optional OC Checks
9.0    Generic Data Validation Procedures
9.1     GC Analyses (i.e., Herbicides,  Organophosphate,
        Pesticides)

9.1.1   Sample Holding Times

        1. Were any of the sample holding times exceeded?*

          Sample holding times can generally be found in
          the analytical  method, or in  the appropriate
          reference, such as the 40CFR Part 136, MCAWW,
          or SW846.

        ACTION: If yes, flag as estimated (J) those values
        above the Instrument Detection Limit (IDL). Values
        that  are' less than the IDL can  be  flagged  as
        estimated (UJ)   or  rejected (R)  based  on  the
        reviewers professional judgement and  the nature of
        the sample and analyte.

        *Because of their long  shelf lives,  performance
        evaluation  samples do  not  have  any associated
        holding times.
8.11.1  Surrogate Recovery
9.1.2   Instrument Performance
       1. Was surrogate outside of specifications for any
         samples?  If yes, were the appropriate samples
         reanalyzed?
       1. Check the raw data to verify that there is adequate
         resolution (> 25%) between peaks of the standard
         compounds.
                                                    47

-------
      ACTION: If adequate separation is not achieved,
      flag all affected compound data as rejected (R).
      2. Check raw data to verify that retention time
         windows are reported and that  all standard
         compounds are within the established retention
         time windows.

      ACTION:  If the standard compounds do not fall
      within the retention  time  windows, professional
      judgement should be used in the evaluation of
      associated sample results.
9.1.3   Initial and Continuing Calibration Verification

       1. Verify that the %RSD of the calibration factor
         for the calibration compounds are less than or
         equal to 10% for the initial calibration linearity
         check.

       ACTION: If criteria for. linearity is not met, flag
       all associated quantitative results as estimated (J).
       2. Verify the proper analytical sequence was run as
          required.         ,

       ACTION: If the proper standards have not been
       analyzed  and  the  sequence   followed,  use
       professional judgement to determine the severity
       of the effect and qualify the data accordingly.
       3. Review  the sample data to verify whether a
          standard was used as a quantitation standard or
          as a confirmation standard.

       ACTION:  If the %D for standard analysis is
       greater than 15% on the quantitation column or
       greater than 20% on the confirmation column, flag
       all associated positive sample results as  estimated
       (J).

9.1.4   Error Determination        j

       See Part I - Section 2.8 for QA samples to be used
       for error determination.     J
    . -         .'-•-.          [
       9.1.4.1 Determination  of  Bias  (% Recovery  -
              Optional for QA-2; Mandatory for
              QA-3)              I
 ....     -,                      f
                                  i
              9.1.4.1.1  Percent Recovery
                                  i
                       1. Were at least eight spiked sample
                         replicates  for the  matrix  of
                         interest analyzed "at the required
                         frequency?

                       ACTION:  if no, flag as recovery
                       not determined fRNDI all data for
                       which spiked samples  were  not
                       analyzed.
                       2. 'Determine the average recovery
                          of the eight spiked replicates. Is
                          the average recovery within the
                       *   applicable control limits (80% to
                          120%)? i
          % recovery for a single spiked sample =
               Spiked sample cone. '<- Sample cone.  x IQQ
                           Spike cone, added
                         ;   ,      I
                       ACTION:  Jf recoveries are within
                       applicable control limits, no bias is
                       considered.^ If % Recovery is less
                       than 80% or greater than 120%, the
                       sample data should be flagged with
                       a (J) estimate and a corresponding
                       (-) or (+) s'ign to show direction of
                       the  bias.  [Adjustment of sample
                       values   should   be   considered
                                                     48

-------
                whenever  there  is  consistent
                evidence of bias.

         9.1.4.1.2 Adjustment of Sample Values
                 for Bias

                 1. Depending on bias direction,
                    add or subtract the value (%
                    Bias x spike concentration)
                    to  or  from  the  sample
                    values.  , This  adjustment
                    approach  assumes a spiking
                    concentration equal to  the
                    concentration found in  the
                    sample.

 9.1.4.2   Determination of Precision (Optional for
         QA-2; Mandatory for  QA-3)

         9.1.4.2.1 Replicate Analysis

                 1. Was  a minimum  of four
                    replicates for QA-2 or eight
                    replicates  for   QA-3
                    analyzed?  If yes, determine
        .  .,,-;,;  ,     coefficient of variation.   If
                    no, flag data with precision
'  ,  .   .     '        not determined (PND),  for
                    which replicate samples were
                    not analyzed.

         9.1.4.2.2 Coefficient    of   Variation
                  (Percent   Relative
                  Deviation)
Standard
                  1. The coefficient  of variation
                    (CV) is used in  determining
                    the  precision  of standard
                    deviation. The CV expresses
                    the standard deviation as a
                    percentage  of  the  mean
                  .  (average)   value   of   the
                    replicate values. The CV is
                                       used  to  determine  a  false
                                       positive or false negative value
                                       for results that are respectively
                                       greater than or  less  than a
                                       decision level concentration.

                                     Determine  the  coefficient  of
                                     variation using  the  following
                                     equation:

                                     CV =  s x 100
                                            X       '

                                     where:
                                     XDL    =    the  decision  level
                                     concentration

                                     s   =   the  sample  standard
                                     deviation given by the equation:
                                     s*  =
   *Note:  When using a programmable calculator
   or computer statistics software, be sure the above
   equation with (n -1) is used and not (n) by itself.
   The  equation using  (n)  is  to  determine  the
   population standard deviation (a) rather than the
   sample standard deviation (s).

   Apply the CV to the decision level to determine
   the  false negative  or false  positive value  as
   follows:

   False positive value = Decision  level value +
   (CV x decision level)

   False negative value = Decision level value - (CV
   x decision level)

Example:

   For an decision  level = 50 ppm and CV  = 20%
                                              49

-------
          False positive value = 50 ppm + (20% x
          50 ppm)
                           = 50 ppm + (10 ppm)
                           = 60 ppm

          False negative value = 50 ppm - (20% x
          50 ppm)
                           = 50 ppm - (10  ppm)
                           = 40 ppm

       For the above false  positive example, any value
       between  50 ppm  and 60 ppm are considered
       suspect and should be reanalyzed. Values above 60
       ppm are considered  actionable.  In many cases,
       false positives have been considered actionable by
       the  Agency  for  safety  reasons.    However,
       depending on the action to be taken, this can be
       costly and unjustifiable.  Consult the QA plan for
       intended use of data and data quality objectives.

       For the above false negative example, any values
       between  40 ppm  and 50 ppm are considered
       suspect and should be reanalyzed. Values below
       40 ppm are considered non-actionable.  In most
       cases, the decision maker will be using the false
       negative value as his decision level  and not be
       concerned  about   the  false   positive  value.
       Whenever sample values need to be corrected for
       both bias and precision, first correct the value for
       bias, then correct the biased value for precision.
9.1.5   Blanks
       1. Verify that method blank analysis has  been
         reported per matrix, per concentration level, at
         the proper frequency, for each GC system used
         to analyze samples, for each extraction batch.

       ACTION:   If the proper type and frequency of
       method blank  have  not  been  analyzed,  use
       professional judgement to determine the effect on
       the data.
       2. Verify that all blank analyses contain less than the
         Required Detection  Limits   (RDL)  of  any
         compound or interfering peak.

       ACTION:   Any compound detected in the sample
       and also detected in any associated blank, must be
       qualified  as  non-detect   (U) when  the  sample
       concentration is less than 5X the blank concentration.
       NOTE:  In instances where more than one blank is
       associated with a given sample, quantification should
       be based upon a comparison  with the associated
       blank  having  the  highest concentration  of  a
       contaminant.  The results must not be corrected by
       subtracting any blank value.;
                                 i
9.1.6   Compound Identification    i
       	*^	    i
                                 i
                                 r •
       1. Verify that positive identifications have dissimilar
         column analysis.

       ACTION:  If the qualitative criteria for dual column
       were not met, all reported positive results should be
       flagged  as  presumptively present at  an estimated
       quantity (NJ).
                  :.'*•'..,       :'T' i;      •
                                 \ (
       2. If multipeak compounds wefe reported, were the
 ;        retention times and relative peak height ratios of
         major component peaks compared  against the
         appropriate standard chroniatograms.

       ACTION:  If multipeak compounds exhibit marginal
       pattern-matching quality  professional judgement
       should be used to establish whether the differences
       are  attributable to  environmental "weathering".  If
       the  presence of a multipeak compound is  strongly
       suggested,   results    shoul(i   be   reported   as
       presumptively present (N).  i
                                 i,
       3. Verify that the sample difpmatogram agree with
         the correct daily standard chromatogram,  and that
         the retention time windows match.
                                                    50

-------
       ACTION: If the chromatograms do not agree, and
       the retention time windows vary significantly, the
       reviewer  must  use  professional  judgement to
       determine the flags that should be applied and the
       usefulness of the data.

9.1.7   Compound Quantitation and Reported Detection
       Limits

       1.  Verify that the reported values, both positives
          and non-detects, have been correctly adjusted to
          reflect  all dilutions,   concentrations,  splits,
         •cleanup procedures, dry weight factors, an any
          other adjustments that have not been accounted
          for by the method.

          For waters: ug/L = (AjfLW,)
         For soils: ug/L = (AjfLW,)
                           (As)(Ws)(D)(Vi)

       A< = area of quantitation peak(s)
       I, = amount of standard injected (ng)
       V, = volume of total extract (ul)
       V; = volume injected (ul)
       Vs = volume of sample (ml)
       Ws =  weight of sample extracted, (g),
       D = (100 ,-,% moisture)/100; or 1 for wet weight
       basis                       •       .  .
       AS = Area of external standard

       ACTION: If incorrect values have been reported,
       it is essential that the correct values be determined.
       The  reviewer  should contact the  laboratory to
       verify any corrections made to the data.

9.1.8   Performance Evaluation Samples

       1. Were  recovery limits within those set by the
         EMSL lab?

       ACTION:   If outside the limits,  review on  a
       compound by  compound basis.  If 50% of the
       compounds  are outside of  confidence  limits or
       were  misidentified, all  sample  results should be
       rejected (R).

9.1.9   Overall Assessment of Data

       It is  appropriate  for the data reviewer to use
       professional judgment and express  concerns and
       comments on the validity of the overall data package
       for a case.  This is particularly appropriate for cases
       in which there are  several QC criteria out of
       specification.  The additive nature of QC factors
       which are out of specification is difficult to assess in
       an  objective  manner,  but  the reviewer has  a
       responsibility to inform the user about data quality
       and data limitations.  This helps the user to  avoid
       using data  inappropriately, while not precluding
       consideration of the data.  The data reviewer would
       be greatly assisted in this endeavor if the data quality
       objectives were provided.
92,    Non-Metal Inorganic Parameters (i.e., anions, pH,
       TOC, nutrients)

9.2.1   Sample Holding Times

       1. Were any of the sample holding times exceeded?*

         Sample Holding Times can generally be found in
         the  analytical  method,  or in  the  appropriate
         reference, such as the 40CFR Part 136, MCAWW,
         or SW846.

       ACTION:  If yes, flag as estimated (J)  those values
       above the Instrument Detection Limit (IDL). Values
       that are less  than the  IDL can be  flagged as
       estimated  (UJ)  or  rejected  (R)  based  on the
       reviewers professional judgement and the nature of
       the sample and analyte.
                                                    51

-------
       *Because of their long shelf lives, performance
       evaluation samples do not have any associated
       holding times.

9.2.2   Initial and Continuing Calibration Verification

       1. Verify that the %RSD of the calibration factor
         for the calibration compounds are less than or
         equal to 10% for the initial calibration linearity
         check.

       ACTION:  If criteria for linearity is not met, flag
       all associated quantitative results as estimated (J).

       2. Verify the proper analytical sequence was run
         as required.

       ACTION:  If the proper standards have not been
       analyzed   and   the  sequence   followed,   use
       professional judgement to determine the severity
       of the effect and qualify the data accordingly.

9.2.3   Error Determination

       See Part I - Section 2.8 for QA samples to be used
       for error determination.

       9.2.3.1   Determination  of Bias (%  Recovery -
               Optional for QA-2; Mandatory for
               QA-3)

               9.2.3.1.1  Percent Recovery
                        1. Were at least  eight spiked
                          sample  replicates  for  the
                          matrix of interest analyzed at
                          the required frequency?

                        ACTION: If no. flag as recovery
                        not determined (RND) all data
                        for which spiked samples were
                        not analyzed.
               2. Determine the average recovery
                  of the eight spiked replicates.  Is
                  the- average recovery within the
                  applicable control limits (80% to
                  120%)?  ;

% recovery for a single spiked sample =
  Spiked sample cone. - Sample cone. x ^QQ
               Spike cone, added

               ACTION: If recoveries are within
               applicable control limits, no bias is
               considered.  [If % Recovery is less
               than 80% or greater than 120%, the
               sample data should be flagged with
               a (J) estimate and a corresponding
               (-) or (+) sign to show direction of
               the bias.  Adjustment  of sample
               values   should   be  considered
               whenever   there   is   consistent
               evidence of bias.

       9.2.3.1.2 Adjustment  of Sample Values for
               Bias        ;'

               1. Depending on bias direction, add
                 or subtract the value (% Bias x
                 spike concentration) to or from
                 the sample values. % bias is the
                 reciprocal value  of %  recovery
                 (i.e., for 70% recovery you have
                 a negative 30% bias).  Use the
                 average % recovery from  the
                 total number of  matrix spikes
                 analyzed. •   This  adjustment
                 approach i assumes   a   spiking
                 concentration  equal   to   the
                 concentration  found  in   the
                 sample.
                                                    52

-------
9.23.2   Determination of Precision (Optional for
         QA-2; Mandatory for QA-3)

 '        9.2.3.2.1 Replicate Analysis
                  s   =   the  sample  standard
                  deviation given by the equation:
                 1. Was a  minimum of eight
                    replicates  analyzed?  If yes,
                    determine   coefficient   of
                    variation.   If no, flag data
                    with precision not determined
                    (PND).  for which replicate
                    samples were not analyzed.

         9.2.3.2.2 Coefficient of Variation (Percent
                 Relative Standard Deviation)

                 1. The coefficient of variation
                    (CV) is used in determining
                    the  precision  of standard
                    deviation. The CV expresses
                    the standard deviation as a
                    percentage  of  the   mean
                    (average)   value  of  the
                    replicate values.  The CV is
                    used to determine a  false
                    positive  or false negative
                    value for   results that  are
                    respectively greater than or
                    less  than   a decision  level
                    concentration.

                    Determine the coefficient of
                    variation using the following
                    equation:

                    CV =  s x 100
                          XDL

                    where:

                    XDL  =   tne decision  level
                    concentration
   *Note: When using a programmable calculator
   or computer statistics software, be sure the above
   equation with (n - 1) is used and not (n) by itself.
   The  equation using  (n)  is  to  determine the
   population standard deviation (a) rather than the
   sample standard deviation (s).

   Apply the CV to the decision level to determine
   the false negative or false positive value as follows:

   False positive value = Decision level value + (CV
   x decision level)

   False negative value = Decision level value - (CV
   x decision level)

Example:

   For an decision level = 50 ppm and CV = 20%

   False positive value = 50 ppm + (20% x
   50 ppm)
          = 50 ppm + (10 ppm)
          = 60 ppm

   False negative value =  50 ppm - (20% x
   50 ppm)
          = 50 ppm - (10 ppm)
          = 40 ppm

For the  above  false positive example, any value
between 50 ppm and 60 ppm are considered suspect
and should be reanalyzed. Values above 60 ppm are
considered actionable.  In many cases, false positives
have been considered actionable by the Agency for
safety reasons. However, depending on the action to
be  taken, this  can  be  costly and  unjustifiable.
                                              53

-------
       Consult the QA plan for intended use of data and
       data quality objectives.

       For the above false negative example, any values
       between 40 ppm  and 50  ppm are considered
       suspect and should be reanalyzed.  Values below
       40 ppm are considered non-actionable.  In most
       cases, the decision maker will be using the  false
       negative value  as his decision level and not be
       concerned  about  the  false  positive  value.
       Whenever sample values need to be corrected for
       both bias and precision, first correct the value for
       bias, then correct the biased value for precision.
9.2.4   Blanks
       1.  Verify  that method blank  analysis has been
          reported per matrix, per concentration level, at
          the proper frequency, for analytical system used
          to analyze samples, for each extraction batch.

       ACTION:  If the proper type and frequency of
       method  blank  have not  been analyzed,  use
       professional judgement to determine the effect on
       the data.

       2.  Verify that all blank analyses contain less than
          the Required Detection Limits (RDL) of  any
          compound or interfering peak.

       ACTION: Any compound detected in the sample
       and also detected in any associated blank, must be
       qualified  as  non-detect  (U)  when the sample
       concentration  is   less  than   5X   the   blank
       concentration.

       NOTE: In instances where more than one blank
       is  associated with a given sample, quantification
       should be based upon  a comparison  with  the
       associated blank having the highest concentration
       of a  contaminant.    The results  must not be
       corrected by subtracting any  blank value.
9.2.5   Compound  Quantitation  and Reported Detection
       Limits

       1. Verify that the reported values, both positives and
          non-detects, have been correctly adjusted to reflect
          all  dilutions,  concentrations,  splits,  cleanup
          procedures, dry weight i factors, and any other
          adjustments that have not been accounted for by
          the method.             >
       ACTION: If incorrect values have been reported, it
       is essential that the correct values be determined.
       The reviewer should contact the laboratory to verify
       any corrections made to the data.
                                 !
9.2.6   Performance Evaluation Samples
                                 i
       1. Were recovery limits  within those  set  by the
          EMSLlab?             '.

       ACTION:   If outside  the limits, review on a
       compound by  compound basis.   If 50%  of the
       compounds are outside of confidence limits or were
       misidentified, all  sample results should be rejected
       (R).                      I
                                 i
9.2.7   Overall Assessment of Data

       It is  appropriate for the data  reviewer  to  use
       professional judgment and express concerns  and
       comments on the validity of the overall data package
       for a case. This is particularly appropriate for cases
       in  which there  are several  QC  criteria  out of
       specification.  The additive nature of  QC  factors
       which are out of specification is difficult to assess in
       an objective  manner,  but  the  reviewer  has a
       responsibility to inform  the user about data quality
       and data limitations.  This helps the user to avoid
       using data  inappropriately,  while not precluding
       consideration of the data. The data reviewer would
       be greatly assisted in this endeavor if the data quality
       objectives were provided.   (
       » U.S. Government Printino Otficat 1990-748-159/004S8
                                                     54

-------