&EPA
VI LIBRARY
"
DALLAS, TEXAS 75202
United States Environmental
Protection Agency
Office of
Pollution Prevention and Toxics
EPA 747-R-95-008
September 1997
METHODOLOGY FOR
XRF PERFORMANCE
CHARACTERISTIC SHEETS
rv r
-------
-------
EPA 747-R-95-008
September 1997
METHODOLOGY FOR XRF PERFORMANCE
CHARACTERISTIC SHEETS
Technical Branch
National Program Chemicals Division
Office of Pollution Prevention and Toxics
Office of Prevention, Pesticides, and Toxic Substances
U.S. Environmental Protection Agency
401 M Street, S.W.
Washington, D.C. 20460
-------
The material in this document has been subject to Agency technical and policy review and
approved for publication as an EPA report. Mention of trade names, products, or services
does not convey, and should not be interpreted as conveying, official EPA approval,
endorsement, or recommendation.
This report is copied on recycled paper.
-------
CONTRIBUTING ORGANIZATIONS
The methodology described in this report is part of a task funded by the U.S.
Environmental Protection Agency and the U.S. Department of Housing and Urban
Development. The task was managed by the U.S. Environmental Protection Agency. The
task was conducted collaboratively by three organizations under contract to the
Environmental Protection Agency: Battelle Memorial Institute, Midwest Research Institute,
and QuanTech. Each organization's responsibilities are listed below.
Battelle Memorial Institute
Battelle Memorial Institute (Battelle) was responsible for oversight of archive sample
maintenance and archive testing.
Midwest Research Institute
Midwest Research Institute (MRI) was responsible for the operations manual,
sample maintenance, collection of paint samples, laboratory analysis, and supervision of
testing.
QuanTech
QuanTech (formerly David C. Cox & Associates) was responsible for testing design,
data management, development of statistical methods, and the writing of this report and
the XRF Performance Characteristic Sheets.
U.S. Environmental Protection Agency
The U.S. Environmental Protection Agency (EPA) co-funded the task, managed the
task, reviewed task documents, and managed the peer review of this report. The EPA
Project Leader was John Schwemberger. The EPA Work Assignment Managers were
Sam Brown, John Scalera and John Schwemberger. The EPA Project Officers were Sam
Brown, Jill Hacker, Phil Robinson, and Sineta Wooten.
U.S. Department of Housing and Development
The Department of Housing and Urban Development (HUD) co-funded the task, and
was responsible for reviews of the XRF Performance Characteristic Sheets and for
contacts with the manufacturers of lead-based paint testing technologies. The key HUD
staff member was Bill Wisner.
HI
-------
ACKNOWLEDGMENTS
Special thanks are due to Mary McKnight of the National Institute of Standards and
Technology (NIST) for operating XRF instruments during archive testing and for her advice
and comments.
-------
TABLE OF CONTENTS
EXECUTIVE SUMMARY xix
1. INTRODUCTION 1
1.1 Background 1
1.2 Peer Review 4
2. COMPLIANCE WITH QUALITY ASSURANCE PROJECT PLAN OBJECTIVES 7
3. ARCHIVE FACILITY TESTING AND REPORTING 13
3.1 Initial Testing of New Instruments 13
3.2 Testing Practices 13
3.3 Analysis of Data and Release of the XRF Performance Characteristic Sheets 14
3.4 Future Improvements 15
4. SAMPLE DESCRIPTIONS AND SAMPLING PROTOCOL SUMMARY 17
4.1 Sample Descriptions 17
4.1.1 EPA/HUD Field Study Sample Descriptions 17
4.1.2 Archive Sample Descriptions 18
4.2 Component Makeup 20
4.3 Distribution of Lead Levels 20
4.4 Control Blocks and Quality Control Procedures 23
4.5 Order of Testing 24
4.5.1 Archive Sample Ordering 24
4.6 Testing Protocols 26
4.7 EPA/HUD Field Study Data vs. Archive Facility Data 27
4.8 EPA/HUD Field Study Data vs. National Survey Data 31
5. PCS DEVELOPMENT AND STATISTICAL METHODOLOGY 35
5.1 Bias and Precision Estimation 35
5.1.1 The XRF Measurement Model 35
5.1.2 Data Used in Model Estimation 38
5.1.3 Estimation Using XRF Instrument Testing Data 39
5.1.4 Combining XRF Results from Different Testing Scenarios 41
5.1.5 Standard Error Estimates 44
5.1.6 Relation to Previous Work 44
5.2 Laboratory Measurement of Lead Levels 44
5.3 Substrate Correction 45
5.3.1 Criteria for Recommending Substrate Correction 46
5.3.2 Measurement Range Subject to Substrate Correction 47
5.4 Inconclusive Ranges and Thresholds 48
5.4.1 Classification Using Inconclusive Ranges and Thresholds 49
5.4.2 Data Used in Inconclusive Range and Threshold Derivation 50
5.4.3 Model-Based Derivation of Inconclusive Ranges and Thresholds 50
5.4.4 Number of Decimal Places Reported 51
5.4.5 Criteria for Inconclusive Range and Threshold Derivation 51
5.4.6 Use of 0.5 mg/cm2 and 2.0 mg/cm2 as Pivotal Values 53
5.4.7 Properties of Inconclusive Ranges and Thresholds 54
5.4.8 Calculation of Inconclusive Ranges and Thresholds 55
5.4.9 Classification Performance of Inconclusive Ranges and Thresholds 60
vii
-------
5.5 XRF Calibration Check Methodology 60
5.6 XRF Re-evaluation Test Methodology 61
5.6.1 Re-Testing in Multifamily Housing 62
5.6.2 Re-Testing in Single-Family Housing 65
5.6.3 Examples of Re-Test Calculations 66
5.6.4 Error in Re-Test Formula 67
5.7 XRF Results That Do Not Consist of Fixed Reading Time Measurements 68
6.0 REFERENCES 69
APPENDIX A: XRF INSTRUMENT OPERATING PROTOCOLS A-1
A.1 Introduction A-1
A.2 Brief Description of Testing Areas A-1
A.3 Summary of Testing Protocols A-1
APPENDIX B: METHODOLOGICAL ISSUES IN THE ANALYSIS OF XRF MEASUREMENTS
WITH VARIABLE READING TIMES B-1
B.1 Introduction B-1
B.2 Variable Reading Time Measurements B-2
B.3 A Model for Variable Reading Time Measurements B-2
B.3.1 Precision Mode Readings B-3
B.3.2 Unlimited Mode Readings B-4
B.4 Nonparametric Inconclusive Ranges for Unlimited Mode Readings B-5
B.5 Substrate Correction B-6
APPENDIX C: XRF PERFORMANCE CHARACTERISTIC SHEETS ERRATA
C-1
C.1 Introduction C-1
C.2 Errors Appearing in the "Instructions for Evaluating XRF Testing" Section C-1
C.3 Errors Appearing in the MAP-3 PCS and Corrections C-1
C.4 Errors Appearing in the LeadStar and MAP 4 PCSs and Corrections C-2
C.5 Errors Appearing in the LPA-1 PCS and Corrections C-2
C.6 Errors Appearing in the Microlead I PCS and Corrections C-2
C.7 Errors Appearing in Titles of PCSs and Corrections C-2
APPENDIX D: XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED RESULTS ... D-1
D.1 Introduction D-1
D.2 XRF Performance Characteristic Sheet for the Advanced Detectors LeadStar and
Related Results D-2
D.3 XRF Performance Characteristic Sheet for the Radiation Monitoring Device LPA-1
and Related Results D-17
D.4 XRF Performance Characteristic Sheet for the Scitec MAP-3 and Related Results . D-45
D.5 XRF Performance Characteristic Sheet for the Scitec MAP 4 and Related Results . D-59
D.6 XRF Performance Characteristic Sheet for the Warrington Microlead I Revision 4
and Related Results D-76
D.7 XRF Performance Characteristic Sheet for the TN Technologies Pb Analyzer and
Related Results D-86
D.8 XRF Performance Characteristic Sheet for the Princeton Gamma-Tech XK-3 and
Related Results D-95
D.9 XRF Performance Characteristic Sheet for the Niton XL-309 Spectrum Analyzer
and Related Results D-105
VIM
-------
APPENDIX E: XRF PERFORMANCE CHARACTERISTIC SHEET: AN EXAMPLE E-1
E.1 Introduction E-1
E.2 Example E-1
IX
-------
LIST OF TABLES
Table 4-1. Number of Test Locations Per Substrate by Dwelling in the EPA/HUD Field Study ... 18
Table 4-2. The Originating City and Construction Date (if known) of Archive Samples
Categorized by Dwelling and Substrate Type 21
Table 4-3. The Number and Overall Percent of Samples in the EPA/HUD Field Study for Each
Type of Component 22
Table 4-4. The Number and Overall Percent of Samples in the EPA/HUD Archive Facility for
Each Type of Component 23
Table 4-5. Summary Statistics of Laboratory Modified NIOSH 70821 ICP Analyses Results
Reported in mg/cm2 Lead Units Categorized by Substrate For Samples From the
EPA/HUD Field Study and Samples In the EPA/HUD Archive Facility 25
Table 4-6. Bias Comparison Calculated Based on Measurements Taken by the Pb Analyzer in
Grouped-Substrate Order and in Random Order 26
Table 4-7. Precision Comparison Calculated Based on Measurements Taken by the Pb
Analyzer in Grouped-Substrate Order and in Random Order 26
Table 4-8. Calibration Check Comparison 28
Table 4-9. Bias Comparison for Measurements Taken by the Pb Analyzer 29
Table 4-10. Precision Comparison for Measurements Taken by the Pb Analyzer 30
Table 4-11. Inconclusive Ranges and Threshold Comparison 31
Table 4-12. Overall Misclassification and Inconclusive Rates Comparison 31
Table 4-13. Percentile and Mean for XRF Measurements for Public Housing Units by Sample
Location 33
Table 4-14. Percentile and Mean for XRF Measurements for Private Housing Units by Sample
Location 33
Table 5-1. Number of XRF Machines Used for PCS Development by Origin of Data for Each
XRF Instrument 42
Table 5-2. Frequencies and Percentages With 95% Confidence Intervals of False Positive
Results for XRF Readings Obtained When the Uncorrected XRF Reading is Greater
Than or Equal to 2.0 mg/cm2, 3.0 mg/cm2, or 4.0 mg/cm2, Based on the EPA/HUD
Field Study Data 48
Table A-1. Variations in the Beginning and End of Day Quality Control Samples Used for XRF
Testing of Archive Samples A-5
Table A-2. Variations in Measurements at Test Areas of Archive Samples A-6
Table D-1. Bias Estimates, and Their Standard Errors, of Advanced Detectors LeadStar
15-Second Fixed Mode Readings For Those Instruments With Software Versions
4.1 to 4.30 D-11
Table D-2. Precision Estimates, and Their Standard Errors, of Advanced Detectors LeadStar
5-Second Fixed Mode Readings For Those Instruments With Software Versions
4.1 to 4.30 D-12
Table D-3. Inconclusive Range For Advanced Detectors LeadStar 15-Second Fixed Mode
Readings For Results Where Substrate Correction Is Not Performed, But
Substrate Correction is Recommended For Those Instruments With Software
Versions Earlier Than Version 4.1 D-13
Table D-4. Bias Estimates, and Their Standard Errors, of Advanced Detectors LeadStar
15-Second Fixed Mode Readings For Those Instruments With Software Versions
Earlier Than Version 4.1 D-13
Table D-5. Precision Estimates, and Their Standard Errors, of Advanced Detectors LeadStar
5-Second Fixed Mode Readings For Those Instruments With Software Versions
Earlier Than Version 4.1 D-14
XI
-------
Table D-6. Classification Results For Advanced Detectors LeadStar Brief Mode (Uncorrected),
Classified Using Threshold Values Reported in the XRF Performance Characteristic
Sheet For Instruments With Software Versions 4.1 to 4.30
Table D-7. Classification Results For Advanced Detectors LeadStar 15-Second Fixed Mode
Readings (Uncorrected), Classified Using Inconclusive Ranges or Threshold
Values Reported in the XRF Performance Characteristic Sheet For Instruments
With Software Versions 4.1 to 4.30
Table D-8. Classification Results For Advanced Detectors LeadStar 15-Second Fixed Mode
Readings (Metal Substrates Corrected and Uncorrected), Classified Using
Inconclusive Ranges Reported in the XRF Performance Characteristic Sheet For
Instruments With Software Versions Earlier Than Version 4.1
Table D-9. Threshold Results For Radiation Monitoring Device LPA-1 30-Second Standard
Mode Readings For Results Where Substrate Correction Is Not Performed, But
Substrate Correction Is Recommended For Instruments For Those Instruments
Sold Prior to June 26, 1995 and Have Not Been Serviced Since June 26, 1995 ...
Table D-10. Threshold Results For Radiation Monitoring Device LPA-1 20-Second Standard
Mode Readings For Results Where Substrate Correction Is Not Performed, But
Substrate Correction Is Recommended For Instruments For Those Instruments
Sold Prior to June 26, 1995 and Have Not Been Serviced Since June 26, 1995 ...
Table D-11. Inconclusive Range Results For Radiation Monitoring Device LPA-1 Quick Mode
Readings For Results Where Substrate Correction Is Not Performed, But
Substrate Correction Is Recommended For Instruments For Those Instruments
Sold Prior to June 26, 1995 and Have Not Been Serviced Since June 26, 1995 . ..
Table D-12. Threshold Results For Radiation Monitoring Device LPA-1 30-Second Standard
Mode Readings For Results Where Substrate Correction Is Not Performed, But
Substrate Correction Is Recommended For Instruments For Those Instruments
Sold or Serviced After June 26, 1995
Table D-13. Bias Estimates, and Their Standard Errors, of Radiation Monitoring Device LPA-1
20-Second Standard Mode Readings For Those Instruments Sold Prior to June 26,
1995 and Have Not Been Serviced Since June 26, 1995
Table D-14. Bias Estimates, and Their Standard Errors, of Radiation Monitoring Device LPA-1
30-Second Standard Mode Readings For Those Instruments Sold Prior to June 26,
1995 and Have Not Been Serviced Since June 26,1995
Table D-15. Bias Estimates, and Their Standard Errors, of Radiation Monitoring Device LPA-1
Quick Mode Readings For Those Instruments Sold Prior to June 26,1995 and
Have Not Been Serviced Since June 26, 1995
Table D-16. Bias Estimates, and Their Standard Errors, of Radiation Monitoring Device LPA-1
30-Second Standard Mode Readings For Those Instruments Sold or Serviced after
June 26, 1995
Table D-17. Bias Estimates, and Their Standard Errors, of Radiation Monitoring Device LPA-1
Quick Mode Readings For Those Instruments Sold or Serviced after June 26, 1995
Table D-18. Precision Estimates, and Their Standard Errors, of Radiation Monitoring Device
LPA-1 20-Second Standard Mode Readings For Those Instruments Sold Prior to
June 26, 1995 and Have Not Been Serviced Since June 26, 1995
Table D-19. Precision Estimates, and Their Standard Errors, of Radiation Monitoring Device
LPA-1 30-Second Standard Mode Readings For Those Instruments Sold Prior to
June 26, 1995 and Have Not Been Serviced Since June 26, 1995
ToL-!e D-20. Precision Estimates, and Their Standard Errors, of Radiation Monitoring Device
LPA-1 Quick Mode Readings For Those Instruments Sold Prior to June 26,1995
and Have Not Been Serviced Since June 26, 1995
D-15
D-15
D-16
D-30
D-30
D-30
D-30
D-31
D-32
D-33
D-34
D-35
D-36
D-37
D-38
XII
-------
Table D-21. Precision Estimates, and Their Standard Errors of Radiation Monitoring Device
LPA-1 30-Second Standard Mode Readings For Those Instruments Sold or
Serviced After June 26, 1995
Table D-22. Precision Estimates, and Their Standard Errors, of Radiation Monitoring Device
LPA-1 Quick Mode Readings For Those Instruments Sold or Serviced After
June 26, 1995
Table D-23. Classification Results For Radiation Monitoring Device LPA-1 30-Second
Readings (Metal Substrates Corrected and Uncorrected), Classified Using Threshold
Values Reported in the XRF Performance Characteristic Sheet For Instruments
Sold Or Serviced After June 26, 1995 and Compared to Laboratory Results in
mg/cm2 Lead Classified Using the 1.0 mg/cm2 Lead Federal Standard for Data
Taken From Testing Archived Building Components in July 1995
Table D-24. Classification Results For Radiation Monitoring Device LPA-1 Quick Mode
Readings (Uncorrected), Classified Using Inconclusive Ranges and Threshold
Values Reported in the XRF Performance Characteristic Sheet For Instruments
Sold Or Serviced After June 26, 1995 and Compared to Laboratory Results in
mg/cm2 Lead Classified Using the 1.0 mg/cm2 Lead Federal Standard For Data
Taken From Testing Archived Building Components in July 1995
Table D-25. Classification Results For Radiation Monitoring Device LPA-1 30-Second
Readings (Metal and Wood Substrates Corrected and Uncorrected),
Classified Using Threshold Values Reported in the XRF Performance
Characteristic Sheet For Instruments That Were Sold Prior to June 26, 1995 and
Have Not Been Serviced Since June 26,1995 and Compared to Laboratory
Results in mg/cm2 Lead Classified Using the 1.0 mg/cm2 Lead Federal Standard
For Data Taken From Testing Archived Building Components in March and
September 1995
Table D-26. Classification Results For Radiation Monitoring Device LPA-1 20-Second
Readings (Metal and Wood Substrates Corrected and Uncorrected),
Classified Using Threshold Values Reported in the XRF Performance
Characteristic Sheet For Instruments That Were Sold Prior to June 26, 1995 and
Have Not Been Serviced Since June 26,1995 and Compared to Laboratory
Results in mg/cm2 Lead Classified Using the 1.0 mg/cm2 Lead Federal Standard
For Data Taken From Testing Archived Building Components in March and
September 1995
Table D-27. Classification Results For Radiation Monitoring Device LPA-1 Quick Mode
Readings (Metal and Wood Substrates Corrected and Uncorrected), Classified
Using Inconclusive Ranges Reported in the XRF Performance Characteristic
Sheet For Instruments That Were Sold Prior to June 26, 1995 and Have Not Been
Serviced Since June 26, 1995 and Compared to Laboratory Results in mg/cm2
Lead Classified Using the 1.0 mg/cm2 Lead Federal Standard For Data Taken
From Testing Archived Building Components in March and September 1995
Table D-28. Inconclusive Ranges For Scitec MAP-3 15-Second K-Shell Readings Where
Substrate Correction Is Not Performed, But Substrate Correction Is Recommended
Table D-29. Inconclusive Ranges or Thresholds For Scitec MAP-3 K-Shell 60-Second Readings
Where Substrate Correction Is Not Performed, But Substrate Correction Is
Recommended
Table D-30. Bias Estimates, and Their Standard Errors, of Scitec MAP-3 K-Shell 15-Second
Readings
D-39
D-40
D-41
D-41
D-42
D-43
D-44
D-52
D-52
D-53
XIII
-------
Table D-31. Bias Estimates, and Their Standard Errors, of Scitec MAP-3 K-Shell 60-Second
Readings D-54
Table D-32. Precision Estimates, and Their Standard Errors, of Scitec MAP-3 K-Shell
15-Second Readings D-55
Table D-33. Precision Estimates, and Their Standard Errors, of Scitec MAP-3 K-Shell
60-Second Readings D-56
Table D-34. Classification Results For Scitec MAP-3 K-Shell 15-Second Readings (Metal
and Wood Corrected and Uncorrected), Classified Using the Inconclusive Ranges
Reported in the XRF Performance Characteristic Sheet and Compared to Laboratory
Results in mg/cm2 Lead Classified Using the 1.0 mg/cm2 Lead Federal Standard For
Data Taken During the EPA/HUD Field Study D-57
Table D-35. Classification Results For Scitec MAP-3 K-Shell 60-Second Readings (Metal
and Wood Corrected and Uncorrected), Classified Using the Inconclusive Ranges
and Threshold Value Reported in the XRF Performance Characteristic Sheet and
Compared to Laboratory Results in mg/cm2 Lead Classified Using the 1.0 mg/cm2
Lead Federal Standard For Data Taken During the EPA/HUD Field Study D-58
Table D-36. Bias Estimates, and Their Standard Errors, of Scitec MAP 4 K-Shell SCREEN
Mode Readings D-68
Table D-37. Bias Estimates, and Their Standard Errors, of Scitec MAP 4 K-Shell TEST Mode
Readings D-69
Table D-38. Precision Estimates, and Their Standard Errors, of Scitec MAP 4 K-Shell SCREEN
Mode Readings D-70
Table D-39. Precision Estimates, and Their Standard Errors, of Scitec MAP 4 K-Shell TEST
Mode Readings D-71
Table D-40. Classification Results For Scitec MAP 4 K-Shell UNLIMITED Mode Readings
(Uncorrected), Classified Using the Inconclusive Ranges Reported in the XRF
Performance Characteristic Sheet and Compared to Laboratory-Measured Lead
Levels in mg/cm2 Classified Using the 1.0 mg/cm2 Lead Federal Standard For Data
Taken From Testing Archived Building Components in February 1996 D-72
Table D-41. Classification Results For Scitec MAP 4 K-Shell SCREEN Mode Readings
(Drywall, Metal, and Wood Corrected and Uncorrected), Classified Using the
Inconclusive Ranges Reported in the XRF Performance Characteristic Sheet and
Compared to Laboratory-Measured Lead Levels in mg/cm2 Classified Using the
1.0 mg/cm2 Lead Federal Standard For Data Taken From Testing Archived
Building Components in February 1996 D-73
Table D-42. Classification Results For Scitec MAP 4 K-Shell TEST Mode Readings (Drywall,
Metal, and Wood Corrected), Classified Using the Inconclusive Ranges and
Threshold Value Reported in the XRF Performance Characteristic Sheet and
Compared to Laboratory-Measured Lead Levels in mg/cm2 Classified Using the
1.0 mg/cm2 Lead Federal Standard For Data Taken From Testing Archived
Building Components in February 1996 D-74
Table D-43. Misclassification of Indicated Positive and Negative Results and Retest
Percentages For Scitec MAP 4 K-Shell Readings (Uncorrected), Taken in
UNLIMITED, SCREEN, and TEST Modes and SCREEN-TEST-CONFIRM Mode
Sequence Compared to Laboratory-Measured Lead Levels in mg/cm2 Classified
Using the 1.0 mg/cm2 Lead Federal Standard For Data Taken From Testing
Archived Building Components in February 1996 Obtained From Archive Testing
Without Applying Substrate Correction Procedures, Inconclusive Ranges, or
Thresholds D-75
XIV
-------
Table D-44. Inconclusive Ranges For Warrington Microlead I Revision 4 Readings Where
Substrate Correction Is Not Performed, But Substrate Correction Is Recommended
D-82
Table D-45. Bias Estimates, and Their Standard Errors, of Warrington Microlead I Revision 4
Readings D-83
Table D-46. Precision Estimates, and Their Standard Errors, of Warrington Microlead I
Revision 4 Readings D-84
Table D-47. Classification Results For Warrington Microlead I Revision 4 Readings (All
Substrate Corrected and Uncorrected Except For Plaster Which is Uncorrected),
Classified Using the Inconclusive Ranges Reported in the XRF Performance
Characteristic Sheet and Compared to Laboratory Results in mg/cm2 Lead
Classified Using the 1.0 mg/cm2 Lead Federal Standard For Data Taken
During the EPA/HUD Field Study D-85
Table D-48. Bias Estimates, and Their Standard Errors, of TN Technologies Pb Analyzer
K-Shell 15-Second Readings D-92
Table D-49. Precision Estimates, and Their Standard Errors, of TN Technologies Pb Analyzer
K-Shell 15-Second Readings D-93
Table D-50. Classification Results For TN Technologies Pb Analyzer K-Shell 15-Second
Readings (Uncorrected), Classified Using Inconclusive Ranges Reported in
the XRF Performance Characteristic Sheet and Compared to Laboratory Results
in mg/cm2 Lead Classified Using the 1.0 mg/cm2 Lead Federal Standard For Data
Taken During the EPA/HUD Field Study and From Testing Archived Building
Components in January 1995 and in September 1995 D-94
Table D-51. Inconclusive Ranges For Princeton Gamma-Tech XK-3 Readings Where
Substrate Correction Is Not Performed, But Substrate Correction Is Recommended
D-101
Table D-52. Bias Estimates, and Their Standard Errors, of Princeton Gamma-Tech XK-3
Readings D-102
Table D-53. Precision Estimates, and Their Standard Errors, of Princeton Gamma-Tech XK-3
Readings D-103
Table D-54. Classification Results For the Princeton Gamma-Tech XK-3 K-shell 15-Second
Readings (Corrected and Uncorrected), Classified Using the Inconclusive
Ranges and Threshold Value Reported in the XRF Performance Characteristic
Sheet and Compared to Laboratory Results in mg/cm2 Lead Classified Using the
1.0 mg/cm2 Lead Federal Standard For Data Taken During the EPA/HUD Field
Study D-104
Table D-55. Bias Estimates, and Their Standard Errors, of Niton XL-309 Spectrum Analyzer
Readings D-113
Table D-56. Precision Estimates, and Their Standard Errors, of Niton XL-309 Spectrum
Analyzer D-113
Table D-57. Classification Results For the Niton XL-309 20-Second L-Shell Readings
(Uncorrected) and 120-Second K-Shell Readings (Uncorrected), Classified
Using the Threshold Values and Methodology Reported in the XRF
Performance Characteristic Sheet and Compared to Laboratory Results in
mg/cm2 Lead Classified Using the 1.0 mg/cm2 Lead Federal Standard For Data
Taken From Archive Testing D-114
XV
-------
LIST OF FIGURES
Figure A-1. General flow chart for XRF testing of archive samples A-3
Figure D-1. XRF Performance Characteristic Sheet for the Advanced Detectors LeadStar D-3
Figure D-2. XRF Performance Characteristic Sheet for the Radiation Monitoring Devices LPA-1. D-18
Figure D-3. XRF Performance Characteristic Sheet for the Scitec Corporation MAP-3 D-46
Figure D-4. XRF Performance Characteristic Sheet for the Scitec Corporation MAP 4 D-60
Figure D-5. XRF Performance Characteristic Sheet for the Warrington Microlead I revision 4. .. D-77
Figure D-6. XRF Performance Characteristic Sheet for the Tn Technologies Pb Analyzer D-87
Figure D-7. XRF Performance Characteristic Sheet for the Princeton Gamma-Tech XK-3 D-96
Figure D-8. XRF Performance Characteristic Sheet for the Niton XL-309 Spectrum Analyzer. . D-106
XVII
-------
EXECUTIVE SUMMARY
The methodology described in this report was developed as part of a task funded
by the U.S. Environmental Protection Agency (EPA) and the U.S. Department of
Housing and Urban Development (HUD) to collect information needed for the
development of federal guidance on testing paint for lead. The methodology was
specifically used to develop testing guidance that is supplemental information to be
used in conjunction with Chapter 7 of the HUD Guidelines for the Evaluation and
Control of Lead-Based Paint Hazards in Housing ("HUD Guidelines").
Information collected from an EPA/HUD field study conducted in 1993 provided
background for Chapter 7 of the HUD Guidelines. This field study came about due to
the passage of Title X (Section 1017 of the Residential Lead-Based Paint Hazard
Reduction Act of 1992), which mandated that the federal government establish
guidelines for lead-based paint hazard evaluation and reduction. From the field study
came detailed information EPA and HUD needed in order to provide up-to-date field
testing guidance for portable X-ray fluorescence (XRF) instruments. The results from
the field study are reported in two documents. A summary report, entitled A Field Test
of Lead-Based Paint Testing Technologies: Summary Report (EPA 747-R-95-002a),
contains an overview of the results from the field study. This report is available from the
National Lead Information Center (1-800-424-LEAD). The other report is a technical
report entitled A Field Test of Lead-Based Paint Testing Technologies: Technical
Report (EPA 747-R-95-002b) which presents results from the field study in detail. This
report can be obtained by calling the National Technical Information Service (NTIS) at
703-487-4650 and ordering the report by its NTIS reference number, PB96-125026.
A primary conclusion of the EPA/HUD field study is that testing by K-shell XRF
instruments, with laboratory confirmation of inconclusive XRF results, and with
substrate correction in cases where this is effective in reducing bias, is a viable way to
test for lead-based paint. This approach can produce satisfactory results for classifying
the paint on architectural components using the federal threshold of 1.0 mg/cm2. These
findings were incorporated into Chapter 7 of the HUD Guidelines.
Since it was anticipated that there would be ongoing improvements in the
performance of XRF instrumentation as the demand for testing for lead-based paint
increased, Chapter 7 was written to defer to easily updated documents for providing
testing guidance for specific instruments. The documents are called XRF Performance
Characteristic Sheets (PCSs). PCSs are XRF instrument model specific documents
intended to provide up-to-date testing guidance and performance information. This
information includes the specification of conclusive and inconclusive XRF results. Each
XRF Performance Characteristic Sheet (PCS) also states whether or not substrate
correction is recommended, and, if necessary, provides supplemental guidance on field
xix
-------
substrate correction procedures. The PCSs also provide calibration check values to be
used in conjunction with one of the NIST Standard Reference Material paint films and a
procedure for evaluating XRF testing.
This report documents the methodology used to develop and produce the XRF
Performance Characteristic Sheets.
xx
-------
1. INTRODUCTION
1.1 Background
This report documents the methodology used for developing the XRF
Performance Characteristic Sheets (PCSs) produced jointly by the U.S. Environmental
Protection Agency (EPA) and the U.S. Department of Housing and Urban Development
(HUD). This document is to supersede previous draft versions of this document. On
several occasions a different draft version was released to coincide with the release of
a new PCS. No further testing or PCS development is planned under an EPA contract.
To date, eight XRF Performance Characteristic Sheets have been released by
EPA and HUD and are reproduced in appendix D of this document. The PCSs are
supplemental to the HUD Guidelines for the Evaluation and Control of Lead-Based
Paint Hazards in Housing ("HUD Guidelines") mandated by Section 1017 of the
Residential Lead-Based Paint Hazard Reduction Act of 1992. The HUD Guidelines
provide comprehensive information on how to evaluate and respond to lead-based
paint hazards in residential housing. In particular, Chapter 7 of the HUD Guidelines
describes how to conduct a lead-based paint inspection. The XRF Performance
Characteristic Sheets contain instrument-specific information needed to conduct a lead-
based paint inspection as delineated by Chapter 7.
The data for the development of the XRF Performance Characteristic Sheets
come from two sources. The first source is the EPA/HUD field study, which was
conducted in 1993 to obtain the data and information necessary to develop federal
guidance on lead-based paint testing. The results of the field study are published in
two reports. A summary report, entitled A Field Test of Lead-Based Paint Testing
Technologies: Summary Report (EPA 747-R-95-002a), contains an overview of the
results from the field study. A technical report, entitled A Field Test of Lead-Based
Paint Testing Technologies: Technical Report (EPA 747-R-95-002b), contains details
of the study procedures and results. (The summary report is available from the
National Lead Information Center by calling 1-800-424-LEAD and requesting the report
by its title and EPA report number, EPA 747-R-95-002a. The technical report is
available from the National Technical Information Center (NTIS) by calling 703-487-
4650 and ordering the report by its NTIS reference number, PB96-125026. The HUD
Guidelines are available from HUD USER by calling 1-800-245-2691 or, in the
Washington D.C. area, 301-251-5154.
The second source of data for the XRF Performance Characteristic Sheets is
XRF testing using an archive of samples collected during and after the EPA/HUD field
study. During the field study, samples of painted housing components were removed
from the houses and testing locations, repairs were made to the houses, and the
samples themselves were packaged and shipped to a laboratory facility. Some
1
-------
additional reference samples were acquired from EPA's Office of Research and
Development, and added to the archive sample materials.
The archive materials were originally intended for quality assurance and data
verification subsequent to the EPA/HUD field study and for testing a few new XRF
instruments or chemical test kits that might become commercially available after the
study was completed. However, at the request of HUD, the archive samples became
the basis for an interim testing program for new XRF instruments while the National
Institute of Standards and Technology (NIST) proceeded with the development of a
protocol for laboratory testing of new XRF instruments. Accordingly, only non-
destructive technologies for testing lead in paint were evaluated by the archive testing,
so as to preserve the samples. Portable XRF instruments were the only technology
that meet the requirement of non-destructiveness.
In most cases, the data used for developing the XRF Performance Characteristic
Sheets came from either the EPA/HUD field study or from archive testing; in one case,
data were available from both the field study and archive testing. The source of the
data for the XRF Performance Characteristic Sheet (PCS) development is indicated on
each PCS.
In the EPA/HUD field study, 1,290 test locations were marked on architectural
components with an indelible-ink-inscribed standardized template in three different
cities: ten single-family houses in Denver, Colorado, four units of a multifamily housing
development in Louisville, Kentucky, and eight units of a multifamily housing
development in Philadelphia, Pennsylvania. About ten percent of the 1,290 test
locations were selected for collection for the archive facility. Building materials that
contained the targeted test locations were collected and later assembled into the
EPA/HUD archive facility.
At the EPA/HUD archive facility, the samples are attached to eight by four foot
plywood sheets. The plywood sheets were arranged in a rectangular configuration
approximately 32 feet long by 20 feet wide and 8 feet high, at least four feet within the
exterior walls, and at least four feet from adjacent objects. All materials used to attach
the samples to the plywood sheets were kept as far as possible from positions lying
directly behind XRF test areas. To minimize interference with XRF testing, plywood
was removed from the back side of the sample locations except for a few samples.
These exceptions include all brick and concrete samples which were first placed in a
wood box before attaching them to the plywood sheet. Furthermore, for those samples
that needed additional support from behind, such as drywall, Styrofoam was placed into
the hole from where the plywood had been removed. Unique numbers were assigned
to each sample for identification purposes. The lead levels for each sample were
determined from laboratory analysis. XRF instruments tested for lead in paint at the
designated areas on each sample.
-------
PCS information was derived from analyzing the XRF testing results and
corresponding lead levels from data collected during the EPA/HUD field study or from
the archive facility testing. The PCSs provide inconclusive range values and threshold
values; state whether or not substrate correction is recommended; and, if necessary,
provide supplemental guidance on field substrate correction procedures. Inconclusive
ranges and thresholds are defined and illustrated in section 5.4. Substrate correction is
discussed in section 5.3. The values reported for the inconclusive ranges and
thresholds were derived from either bias-corrected XRF results or from uncorrected
results, depending on whether correcting for substrate bias is recommended. Bias and
precision estimates shown in the PCSs, however, were derived from uncorrected
results.
Much of the information reported on a PCS consists of estimates obtained by
fitting a model to XRF instrument testing data. The model, which was originally
developed for the EPA/HUD field study, describes the statistical properties of XRF
measurements as a function of the lead level. From the model, it is possible to estimate
the bias and precision of an XRF instrument at pre-specified lead levels; determine
whether or not a benefit from substrate correction is demonstrated; and, derive
inconclusive ranges and thresholds for the classification of XRF measurements relative
to the 1.0 mg/cm2 federal standard. For example, with an inconclusive range of 0.8 to
1.3, an XRF result less than or equal to 0.8 indicates a lead level below the federal
standard. An XRF result greater than or equal to 1.3 indicates a lead level greater than
or equal to the federal standard. An XRF result strictly between 0.8 and 1.3 is regarded
as inconclusive, and referred to laboratory testing for a final resolution. Inconclusive
ranges and thresholds are described more fully in section 5.4 of this document.
The remaining sections of this document provide greater detail about XRF
Performance Characteristic Sheets and their development. Section 2 explains
compliance with Quality Assurance Project Plan Objectives. Section 3 explains archive
policy issues related to testing the archive samples and PCS development. Section 4
provides descriptions of samples from both the EPA/HUD field study and the archive.
Section 5 provides details of the statistical methodology used to derive the numerical
information reported on a PCS. The model used for this purpose is presented, and
details concerning its application to XRF measurements obtained from both the field
study and archive testing are discussed. Appendix A provides information about XRF
testing. Appendix B discusses the statistical treatment of XRF measurements taken
with variable reading times. Appendix C lists errors and corrections in previously
released XRF Performance Characteristic Sheets. Appendix D provides all of the eight
XRF Performance Characteristic Sheets released to date, with all corrections made,
and related empirical results corresponding to each PCS. Appendix E is an example of
a sample PCS that is shorter than the current version.
-------
The XRF Performance Characteristic Sheets (PCS) for the Princeton Gamma-
Tech XK-3, the Scitec MAP-3, and the Warrington Microlead I Revision 4 were
developed from data collected during the EPA/HUD field study. A PCS for the TN
Technologies Pb Analyzer was developed from the field study and from two separate
testing events at the archive facility. PCSs for the Niton XL-309, the Radiation
Monitoring Device LPA-1, the Advanced Detectors (formerly Xsirius) LeadStar, and the
Scitec MAP 4 were developed from archive testing data. Archive testing has been
conducted with the Princeton Gamma-Tech XK-3 and the Warrington Microlead I
Revision 4, but the testing took place after the release of the first edition of the PCS for
those two instruments. Results of archive testing for the XK-3 and the Microlead I
Revision 4 have not been included in the development of the PCS for those two
instruments.
1.2 Peer Review
This report was reviewed independently by five members of a peer review panel.
Peer review comments which are important for interpreting the report are discussed
below.
A reviewer commented on the distribution of lead levels in the archive samples
and choice of the pivotal values 0.5 mg/cm2 and 2.0 mg/cm2 for determining
inconclusive ranges and thresholds. These pivotal values were chosen because they
produced inconclusive ranges and thresholds that were most similar to those obtained
under a direct and more complicated derivation, which involves integrating against a
lognormal density.
Another comment was related to the equal weighting of results from the
EPA/HUD field study and archive testing. In making the comment, however, the
reviewer noted that combining results of the field study and archive study by weighting
by sample size would cause the field study to dominate the archive testing. Upon
consideration of the comment, equal weighting of the archive testing and the field study
results was deemed appropriate.
One reviewer pointed out that the report indicated that in certain cases
uncorrected XRF results as high as 3.0 mg/cm2 could be false positives (that is, the true
lead level could in fact be less than 1.0 mg/cm2 as measured by laboratory analysis).
The XRF Performance Characteristic Sheets have always used 4.0 mg/cm2 as the cut-
off for substrate correction. Section 5.3.2, Measurement Range Subject to Substrate
Correction, discusses this issue.
Another reviewer commented on a perceived lack of "real world" testing, and
raised a number of concerns, such as effects of electrical wiring, piping, fasteners,
-------
grease, dirt, contact papers, irregular surfaces, and electromagnetic fields and the
decision to test instruments with radioactive sources no older than six months. The
archive was constructed almost exclusively with "real world" painted components from
houses and dwelling units built before 1978. Numerous different textures, surfaces,
paint layerings, and substrate combinations are present in the archive. Although it has
been beyond the scope of archive testing to test the effects of wiring, piping, fasteners,
and electromagnetic fields, federal guidance has tended to recommend avoiding testing
where and when there are obvious potential interferences. The decision to use sources
no more than six months old has been done for two reasons: 1) to test instruments
under the most similar conditions possible and 2) to perform instrument testing as
quickly as possible. Testing with an older source can make use of a PCS by increasing
the reading time as necessary, according to the age and half-life of the source.
Whether older sources raise performance issues, aside from requiring longer reading
times, has not been explored.
There were many comments regarding the clarity of the presentation, especially
technical issues. Revisions for clarity were made in most of those sections indicated by
the reviewers. However, some sections were unavoidably technical and detailed. One
reviewer commented that the XRF Performance Characteristic Sheets were too detailed
and confusing, and commented favorably on the proposed shortened version of a PCS
provided in one of the appendices of this report. The proposed shorter form will not be
pursued by EPA at this time. The shorter form or a similar form may be pursued in the
future.
There were review comments on editorial mistakes in the XRF Performance
Characteristic Sheets and on simple calculation errors in some of the empirical results
tables associated with the XRF Performance Characteristic Sheets. Corrections were
made where necessary. Other corrections to the PCSs, based on internal review, were
also made. Appendix C contains a summary of the corrections to the PCSs. Appendix
D contains all the XRF Performance Characteristic Sheets released to the public as of
August 31, 1997, with all corrections made.
-------
2. COMPLIANCE WITH QUALITY ASSURANCE PROJECT PLAN OBJECTIVES
The development of Performance Characteristic Sheets follows from objectives
laid out in an addendum to "Quality Assurance Project Plan for Comparative Field Study
of Methodologies Used to Detect Lead in Paint" dated December 14, 1994. These
objectives are listed and discussed below:
The first and primary objective was to provide a limited evaluation of new XRF
instruments that were not in the full study and to report the results to HUD.
The archive facility was designed to meet this objective. Archive testing is the
basis for the development of PCSs for instruments that were not tested in the full
(EPA/HUD field) study. For instruments that were tested in the full study, PCSs have
been developed using the study data, possibly in combination with archive testing data.
The addendum listed eight "study objectives" that further elaborated the goals of
the first objective:
(1) Estimate the bias of the new XRF instruments at 0 and 1.0 milligrams per square
centimeter to within plus or minus 0.2 with 95% confidence, overall, and to within
plus or minus 0.4 with 95% confidence on wood, metal, and plaster.
Tables in appendix D of this document report estimated standard errors for the
bias at the 0.0 and 1.0 mg/cm2 lead levels, for each XRF instrument that has a
PCS. These standard errors, reported by substrate, seldom exceed 0.1. A 95%
confidence interval is formed by adding and subtracting 2 times the standard
error from the bias estimate, which suggests that objective (1) is achieved on a
substrate-specific basis. The standard error of a pooled estimate, consisting of a
weighted average of substrate-specific estimates, would satisfy the plus or minus
0.2 criterion for those instruments that have a PCS.
(2) Estimate the precision of the new XRF instruments at 0 and 1.0 milligrams per
square centimeter to within plus or minus 3 times what was in the full study (with
95% confidence), overall and on wood, metal, and plaster.
Estimates from the full EPA/HUD field study are based on a sample size of about
1,200. Estimates for new XRF instruments are based on archive tests, each of
which contributes a sample size of about 150. The sample size ratio, based on a
single archive test, is approximately 8, and the square root of this ratio is less
than 3. In addition, precision estimates for newer XRF instruments have been
-------
found to be lower than for instruments tested in the field study. These factors
suggest that objective (2) is met by the PCS development process.
(3) Develop an operating characteristic curve over all substrates and on wood,
metal, and plaster for each new XRF, Estimate the threshold probability and
50% point with precision no more than three times the standard errors of similar
parameter estimates in the full study.
The PCS development process does not make use of the information that an
operating characteristic curve would provide, including the threshold probability
and 50-percent point. Operating characteristic curves and quantities derived
from them are therefore not estimated.
(4) Estimate the inconclusive region for each XRF using order statistics, overall and
on wood, metal, and plaster. Compute misclassification rates and inconclusive
rates overall and for wood, metal, and plaster substrates.
The estimation of inconclusive regions is central to the PCS development
process. These regions consist of inconclusive ranges and thresholds that are
derived from a model that is fit to testing data. Misclassification and inconclusive
rates are reported in appendix D of this document.
A model-based approach is used instead of order statistics, because sample
sizes are usually too small to obtain stable estimates using the latter approach.
A model-based approach overcomes this difficulty by imposing reasonable
assumptions on the data. Estimated percentiles are obtained as simple
functions of the estimated model parameters. Calculation of inconclusive ranges
and thresholds is fully explained in section 5.4 of this document.
(5) Estimate the bias and precision of each new XRF on the NIST SRMs and
develop control limits for usage of these instruments in the field.
Control readings are made in the archive testing of an XRF instrument, using the
NIST Standard Reference Material paint films (SRMs). The NIST SRMs and
their role in archive testing are more fully discussed in section 4.4 of this
document. Readings made on red NIST SRMs, which have a lead level of 1.02
mg/cm2, over wood control blocks are used to estimate bias and precision.
These bias and precision estimates are then used to derive the calibration check
bounds that appear on a PCS, as described in section 5.5 of this document.
8
-------
(6) Estimate time for a nominal reading using the standard operating protocol for
each new instrument.
For instruments that have a variable-time reading mode, a statistical summary of
reading times is given on the PCSs. As explained in appendix E, this information
is provided to give a more complete accounting of the precision of XRF
measurements taken with variable reading times. In fixed-time reading mode,
the situation is simpler: actual reading times are longer than nominal reading
times by a factor that depends on the age and type of radioisotope used by the
instrument. The reading time does not raise precision issues for an instrument in
fixed-time reading mode, and for this reason information about the actual versus
nominal reading time is not provided on the PCSs for these instruments.
(7) Determine the influence of paint thickness on XRF measurements.
This objective was not addressed by the PCS development process, because
paint thickness is not a controllable factor in nondestructive lead testing.
Variation in XRF measurements that is due to paint thickness is treated as a part
of the total variation attributed to an instrument.
(8) Assess the performance of auxiliary indicators of XRF performance, such as the
"absorption index" of the Niton XL.
The auxiliary indicators referred to are typically specific to a given instrument. To
include an assessment of the performance of each of these indicators would
require the PCSs to depart from a standardized format. For this reason the
PCSs are not designed to address every data feature of an XRF instrument, but
only those most directly related to performance issues that are common to all
XRF instruments. The PCS development process retains the flexibility to include
new instrument features, as it has done, for instance, with variable-time reading
modes, which have been adopted by a number of instruments.
The second objective was to deal with spatial variability in the paint samples.
This was done by taking a second sample for ICP analysis so as to better
characterize the lead in the paint in the XRF test area, by using this second
sample to validate interpolation models, and by taking XRF readings at additional
locations on the archived material to determine the impact of additional areas on
XRF measurements.
-------
A second set of ICP measurements was taken on each of the archive samples,
and XRF measurements are taken on as many as three separate locations per sample
in archive testing. These additional measurements have not been used to date in the
PCS development process. The main obstacle to their usage is the need for further
research into how the additional data can be used to effect an improvement in
estimation. This is discussed in section 5.2 of the present document for the case of
laboratory duplicate ICP measurements.
The addendum listed three "study objectives" that further elaborated the
goals of the second objective:
(1) Estimate the lead in paint in the primary XRF testing area using an average of
samples near the primary XRF testing area.
The average of multiple ICP measurements is not currently used as an estimate
of the true lead level in the PCS development process, because it is not clear
that the average constitutes an improvement over the use of a single
measurement. This point is further explained in response to (2) below.
(2) Investigate interpolation models for spatial variation in paint using paint chip
samples at varying distances.
Further investigation of the spatial variation in lead levels across a sample was
suggested by the results of the EPA/HUD field study. The method used to
account for this variation will play a key role in determining how multiple ICP
measurements can be combined to obtain an improved estimate of the lead
level. The development of a model for this purpose was investigated, but is not
complete at this time, and remains an area for possible future research.
(3) Characterize differences between XRF testing at one area versus XRF testing at
three areas on a sample.
Variation in XRF measurements across an archive sample has been noted.
Some of this variation may be attributed to spatial variation in lead leve!3, as
discussed above. Other sources of variation are attributable to the XRF
instrument. The characterization of these sources of variation remains an area
for future research.
10
-------
The third and last objective was to determine if the order of substrate materials
has an effect on XRF readings.
In the archive testing of an XRF instrument, the samples are tested in a
randomized order. As a result, consecutive readings usually involve a change in
substrate materials. The issue of whether these substrate changes introduce effects
was investigated with the TN Technologies Pb Analyzer, which was the first XRF
instrument tested at the archive facility. A preliminary set of measurements was taken
on the archive samples ordered by substrate. No significant differences were found
between the two sets of measurements. This issue is discussed in section 4.5 of the
present document.
11
-------
3. ARCHIVE FACILITY TESTING AND REPORTING
This chapter describes in general terms the testing of the archive samples and
the release of the resulting XRF Performance Characteristic Sheets. Testing and
publication practices are described, as are recommendations for improvements in the
future.
3.1 Initial Testing of New Instruments
Underlying the XRF Performance Characteristic Sheets is the concept of
responding to changes in technology. The PCSs are regarded as documentation that
can be supplemented, replaced, or updated as new XRF instruments enter the market.
As stated above, testing with the archive samples has been restricted to non-
destructive technologies, and at this time portable XRF instruments are the only paint
testing technology that meets the requirement of being non-destructive for archive
testing purposes. Archive testing policy has placed the first priority on testing new
instruments from manufacturers who have not had an XRF Performance Characteristic
Sheet published in the past. The second priority has been on testing instruments that
can be ascertained to be different from past and current instruments by a review of
product literature and operations manuals. Changes to instruments that are cosmetic in
nature or otherwise do not affect the readings of lead in paint have not called for
archive testing to date. However, some instruments that were extensively tested in the
EPA/HUD field study were tested at the archive at the request of the National Institute
of Standards and Technology (NIST). Data from these instruments was collected
following protocols that allow the data to be used for PCS development.
3.2 Testing Practices
After an instrument has been designated for testing, a protocol for archive testing
is developed, based on the operations manual for the instrument and the data
requirements for producing a PCS. The protocol indicates which operation modes of
the instrument will be used, if there is more than one mode.
Operation of the XRF instruments during archive testing has been carried out
either by Dr. Mary McKnight of NIST or by operators from independent testing
companies under subcontract to an EPA contractor. Testing has been supervised and
monitored by staff from organizations under contract to EPA. Operators from testing
companies have been required to obtain training with the instrument they operate prior
to archive testing. Instruments with sources no more than six months old are preferred,
and most of the archive testing has been done with such instruments. Two rounds of
13
-------
testing, with different machines1 of the instrument type, are preferred in order to obtain
information about variation across machines at minimal cost. In practice, for a variety
of reasons, PCSs have been released with only one round of testing, or with more than
two rounds of testing. All rounds of testing that are available at the time of PCS
development are used.
Instruments operated by Dr. McKnight have been owned by NIST or, in one
case, loaned to NIST. Instruments operated by the personnel from testing companies
may be owned or leased by the companies. For subcontractor testing, manufacturers
have been asked by archive facility staff to furnish the names of three or more testing
companies that might bid on the testing subcontract. It is plausible that the
manufacturers loan new machines to the subcontractors for use in archive testing, as
would be expected for a new model or version of the instrument.
An instrument may fail to pass a calibration check, experience battery failure, or
may for other reasons be precluded from further use during testing. When such cases
have arisen, the instrument in question has been replaced by a back-up machine so
that testing may be completed on all archive samples. For subcontractor testing, it has
become a standard policy to require that a back-up machine be available on-site. If an
instrument fails during testing, all readings since the last successful quality control (QC)
check are repeated. Quality control procedures are discussed in section 4.4.
3.3 Analysis of Data and Release of the XRF Performance Characteristic
Sheets
After completion of testing, the test data are analyzed to meet the reporting
requirements in the XRF Performance Characteristic Sheets. The key elements for the
PCS are the determination of the inconclusive range or threshold for each substrate,
the determination of substrates for which substrate correction is recommended, and the
determination of the tolerance values for calibration checks specified by Chapter 7 of
the HUD Guidelines. Estimates of bias and precision of the instrument are also
published as part of the PCS. Each Sheet is reviewed, and after resolution of
comments, the Sheet is made available to the public through the National Lead
Information Center. Copies of the Sheets can be requested by calling 1-800-434-LEAD
and requesting the XRF Performance Characteristic Sheets.
1lt is necessary to distinguish a type of instrument (model, make, version, etc.) from
a physical unit. The term "instrument" is used to refer to either where doing so is not
ambiguous. Otherwise, the term "instrument" is used to refer to a type, and the term
"machine" is used to refer to a physical unit.
14
-------
Although it is not stated on the PCS, testing with a PCS that is current at the time
of testing is considered by EPA and HUD to be valid testing, regardless of any
subsequent changes in PCS methodology or in XRF instrumentation. However, it is
expected that a tester will always make sure that he or she is using a current PCS at
the time when conducting a lead-based paint inspection.
3.4 Future Improvements
Based on experiences with testing, a number of areas of archive facility testing
and PCS development are subject to change. The PCSs themselves could be
simplified in the future (an example of a simplified PCS is in appendix E). Because of
the increased availability of XRF modes that have a variable instead of a fixed time, an
analysis approach for variable time modes has been developed. (This approach is
described in appendix B). The issue of making recommendations for substrate
correction based on archive samples is one that requires additional research.
15
-------
4. SAMPLE DESCRIPTIONS AND SAMPLING PROTOCOL SUMMARY
4.1 Sample Descriptions
4.1.1 EPA/HUD Field Study Sample Descriptions
For the EPA/HUD field study, primary among considerations for the selection of
housing units was the selection of units with a wide distribution of surface types and the
selection of units that were likely to represent those that are currently being routinely
tested for lead-based paint. Therefore, both multifamily housing units and single-family
housing units were included in the field study to generate results that represent both
these types of housing.
With the cooperation of the public housing authority and following site
inspections by study team members, test locations in the EPA/HUD field study were
selected in housing units in three different cities. In Louisville, Kentucky, test locations
were selected in four units in a multifamily housing development, two units each inside
two buildings of the development. In Denver, Colorado, test locations were selected in
ten single-family homes and in Philadelphia, Pennsylvania, eight units in a multifamily
housing development were selected inside two buildings of the development.
Within the housing units chosen for the study, test locations were selected and
marked. A test location was an area on a painted building component where
measurements for lead-based paint were taken. Examples of building components
included walls, baseboards, doors, and window frames. The material underlying the
test locations was classified as being one of the following six substrates: brick,
concrete, metal, drywall, plaster, or wood.
A breakdown of the selected test locations for each city in the EPA/HUD field
study follows. In Louisville, a total of 25 test locations per housing unit were selected
for testing for a total of 100 locations. A total of 75 test locations per house were
selected in Denver and 55 locations per housing unit were selected in Philadelphia.
Thus, a total of 1,290 test locations were selected for the EPA/HUD field study. Table
4-1 provides an additional breakdown of test locations for each of the six substrates.
17
-------
Table 4-1. Number of Test Locations Per Substrate by Dwelling in the EPA/HUD
Field Study.
CITY
Louisville
Denver
Philadelphia
DWELLING
1
2
1
2
3
4
5
6
7
8
9
10
1
2
3
4
5
6
7
8
YEAR
BUILT
1937
1937
1943
1948
1952
1905
1949
1948
1952
1890
1949
1947
1942
1942
1942
1942
1942
1942
1942
1942 __,
NUMBER OF TEST LOCATIONS PER SUBSTRATE
Brick
0
0
20
0
0
0
3
0
0
21
21
16
2
2
0
0
2
2
2
2
Concrete
4
4
4
1
2
15
0
6
10
15
1
44
15
15
15
15
15
15
15
15
Drywall
4
7
4
0
8
25
20
18
16
1
0
13
2
2
2
2
0
0
0
0
Metal
17
11
4
10
2
3
6
8
12
6
9
2
16
12
16
19
16
16
16
16
Plaster
9
11
15
20
22
10
0
0
0
13
21
0
11
12
13
13
18
18
18
18
Wood
16
17
28
44
41
22
46
43
37
19
23
0
9
12
9
6
4
4
4
4
4.1.2 Archive Sample Descriptions
The EPA/HUD archive facility is a collection of 158 samples that has been
assembled as a mechanism for testing and evaluating XRF instruments. Each archive
sample consists of a substantial piece of painted substrate that has been kept intact as
much as possible, and displaying the testing template, if applicable.
Ten percent of the 1,290 test locations in the EPA/HUD field study were targeted
for archiving. Given the likelihood of breakage, slightly more than ten percent were
selected for removal in order to collect at least the targeted number of samples. The
selection of test locations were made by a field statistician under constraints such as
ease of removal and transport, willingness of the public housing authority to allow the
18
-------
removal, and lead levels in the paint as measured by prior XRF testing data2. None of
the selected samples were marked as archive samples until all testing was completed.
Archive sample selections were made using a randomized stratification sampling
scheme; the stratification was based on lead levels provided by XRF measurements.
Of the selected test locations, (approximately) one-fourth of them were selected from
test locations with measured lead levels less than 0.5 mg/cm2 lead, another one-fourth
of them were selected from test locations with measured lead levels equal to and
greater than 3.0 mg/cm2 lead, and the remaining one-half of the test locations were
selected from test locations with measured lead levels equal to and greater than 0.5
mg/cm2 lead but less than 3.0 mg/cm2 lead. The removal of wood, metal, drywall, and
plaster test materials was relatively straightforward. (However, several of the drywall
and plaster samples were very fragile and became unusable due to breakage during
removal or shipping). Brick and, especially, concrete test materials were much more
difficult to remove.
After the samples were removed from the houses and testing locations, the
samples themselves were packaged and shipped to a site that became the archive
facility. Next, the samples were mounted onto four by eight foot plywood sheets. The
plywood sheets were arranged in a rectangular configuration approximately 32 feet long
by 20 feet wide and 8 feet high, at least four feet within the exterior walls, and at least
four feet from adjacent objects. All materials used to attach the samples to the plywood
sheets were kept as far as possible from positions lying directly behind XRF test areas.
To minimize interference with XRF testing, plywood was removed from the back side of
the sample locations except for a few samples. These exceptions include all brick and
concrete samples, which were first placed in a wood box before attaching them to the
plywood sheets. Furthermore, for those samples that needed additional support from
behind, such as drywall, Styrofoam was placed into the hole resulting from the removal
of the plywood. Unique numbers were assigned to each sample for identification
purposes. The lead levels for each sample were determined from laboratory analysis.
XRF instruments tested the designated area on each sample for lead.
The total number of test samples from the EPA/HUD field study that actually
became mounted in the archive facility was 132 samples. From all four units in
Louisville, a total of 13 field study test samples were collected and mounted at the
archive facility. From nine houses in Denver, 76 field samples were collected and
mounted and from three units in Philadelphia, 43 field samples were collected and
mounted.
Selections were made in the field, after XRF testing had been done but before paint
samples were analyzed by the laboratory. Thus, at the time, the best available source
of lead level measurements were provided by XRF instruments.
19
-------
Twenty-six additional samples were mounted and became archive samples.
During the EPA/HUD field study in Louisville, building components were retrieved from
construction site waste areas; nine of them became archive samples. Thirteen samples
were added simply by marking test areas on the back sides of 13 already-mounted
archive samples. The last additions were four wood samples that were donated to the
archive facility from another lead-based paint research project funded by EPA's Office
of Research and Development. Thus, the archive facility includes a total of 158 archive
samples. Table 4-2 provides a summary of the origin, date, and the number of archive
samples per substrate type by dwelling.
4.2 Component Makeup
Samples from the EPA/HUD field study and samples in the archive facility were
categorized into the building components from which they were made. These
components, for example, in a bedroom, could be the ceiling, floor, walls, a door, its
casing, the window sash, or casings. Table 4-3 displays number and overall percent of
samples in the field study samples for each type of component. Table 4-4 provides the
same information for the archive samples.
4.3 Distribution of Lead Levels
In the EPA/HUD field study, paint samples were collected at 1,290 sampling
locations in the three cities. There were 100 locations in two multifamily buildings in
Louisville, 750 in ten single-family houses in Denver, and 440 in eight multifamily units
in Philadelphia. Each sample was analyzed using a modified NIOSH 7082 method3
applied to a 0.5 gram subsample of the sample (if it weighed more than 0.5 gram),
taken after homogenization of the sample. Table 4-5 presents summary statistics in
mg/cm2 lead by substrate, aggregated across cities from the 1,290 test locations in the
field study and from the 158 test locations in the EPA/HUD archive facility.
Comparisons of the medians to the arithmetic means shown in Table 4-5 indicates that
the lead level distribution is skewed toward high levels in most cases.
3The NIOSH Method 7082 is designed to prepare and analyze air filter samples for
analysis of many inorganic elements, including lead. Modifications were needed to
make it applicable to processing paint samples. Details are provided in the technical
report to the EPA/HUD field study.
20
-------
Table 4-2. The Originating City and Construction Date (if known) of Archive Samples
Categorized by Dwelling and Substrate Type
ORIGIN
Louisville
Denver
Philadelphia
Backside
Donated
DWELLING'
1
2
unknown"
2
3
4
5
6
7
8
9
10
5
6
8
n/a
unknown'
YEAR
BUILT
1937
1937
1937
1948
1952
1905
1949
1948
1952
1890
1949
1947
1942
1942
1942
n/a
unknown'
NUMBER OF TEST LOCATIONS PER SUBSTRATE
Brick
0
0
0
0
0
0
0
0
0
3
0
0
0
0
0
0
0
Concrete
0
0
0
0
0
1
0
0
0
0
0
1
0
0
0
0
0
Drywall
0
0
1
0
3
0
0
2
2
1
0
5
0
0
0
0
0
Metal
1
2
5
5
0
1
2
1
1
2
1
0
6
6
5
0
0
Plaster
1
2
3
1
1
0
0
0
0
4
1
0
8
7
10
7
0
Wood
6
1
0
9
9
3
2
7
7
3
2
0
5
1
4
6
4
The dwelling numbering scheme is described in the technical report to the field study.
bBuilding components were retrieved from the construction site waste areas within the multifamily
development where the EPA/HUD field study was being conducted. Even though the exact unit from
which these samples were taken is not known, it is reasonable to assume that all units in the housing
development were constructed at the same time.
GSamples donated from another lead-based paint research project funded by EPA's Office of Research
and Development.
21
-------
Table 4-3. The Number and Overall Percent of Samples in the EPA/HUD Field Study
for Each Type of Component.
COMPONENT
Ext. door frame
Door frame
Exterior door
Interior door
Ext. window sill
Int. window sill
Window casing
Bar top
Threshold
Shelf
Closet wall
Shelf support
Wall trim
Wall
Support column
Ceiling beam
Ceiling hatch door
Floor baseboard
Elec. box cover
Duct
Stair riser
Stair stringer
Coat rack support
Medicine cabinet
Side of bar
Railing cap
NO. OF
SAMPLES
58
9
7
9
2
3
27
2
5
47
8
19
13
513
47
54
3
43
9
34
17
2
1
4
1
7
OVERALL
PERCENT
4.5
0.7
0.5
0.7
0.2
0.2
2.1
0.2
0.4
3.6
0.6
1.5
1.0
39.8
3.6
4.2
0.2
3.3
0.7
2.6
1.3
0.2
0.1
0.3
0.1
0.5
COMPONENT
Gutter
Downspout
Door jamb
Door casing
Door
Window sill
Window sash
Facia
Shutter
Header
Drawer
Cabinet
Closet door
Closet shelf
Foundation
Mail box or slot
Heating register
Pipe
Plumbing access
Rafter support
Rafter
Floor
Roof flashing
Closet baseboard
Awning
Unknown
NO. OF
SAMPLES
16
13
30
5
84
34
9
12
2
17
2
24
5
3
20
3
1
8
7
2
6
13
2
14
6
8
OVERALL
PERCENT
1.2
1.0
2.3
0.4
6.5
2.6
0.7
0.9
0.2
1.3
0.2
1.9
0.4
0.2
1.6
0.2
0.1
0.6
0.5
0.2
0.5
1.0
0.2
1.1
0.5
0.6
22
-------
Table 4-4. The Number and Overall Percent of Samples in the EPA/HUD Archive
Facility for Each Type of Component.
COMPONENT
Door frame
Door: exterior
Door: interior
Threshold
Shelf
Closet wall
Shelf support
Wall trim
Wall
Ceiling hatch door
Floor baseboard
Electrical box cover
Duct
Medicine cabinet
Gutter
Downspout
Door jamb
Door casing
Door
Cabinet
Closet shelf
Mail box or slot
Pipe
Closet baseboard
Awning
Unknown
NO. OF SAMPLES
3
•1
3
1
11
1
5
5
50
1
16
2
2
3
4
5
1
1
16
6
1
2
1
5
3
9
OVERALL
PERCENT
1.9
0.6
1.9
0.6
7.0
0.6
3.2
3.2
31.6
0.6
10.1
1.3
1.3
1.9
2.5
3.2
0.6
0.6
10.1
3.8
0.6
1.3
0.6
3.2
1.9
5.7
4.4 Control Blocks and Quality Control Procedures
Control blocks were developed to permit the investigation of quality control (QC)
procedures and for calibration check development. Control blocks were constructed
from various building materials representing the commonly encountered substrates.
XRF measurements were taken on the control blocks while covered with known levels
of lead-in-paint films enclosed in plastic. The plastic film is a Standard Reference
Material (SRM 2579) available from the U.S. National Institute of Standards and
Technology (NIST) and referred to as NIST SRM films. XRF measurements were also
taken on control blocks that were not covered with the NIST SRM films.
Control blocks were constructed of six different types of materials: brick,
concrete (with aggregate), drywall, metal, plaster, and wood (pine). The brick control
block was a "standard" baked clay brick. The other five control blocks measured
approximately four by four inches on the surface and varied in thickness, depending on
their composition.
23
-------
For each XRF testing day, all control block measurements made by a specific
XRF operator were always performed on the same set of control blocks at a fixed
location and the blocks were always placed on twelve inches of Styrofoam support.
These restrictions were made to assure that differences observed in control block data
would be free from influence from physical differences (including control block
differences) and to eliminate any effects from underlying materials.
Three types of control block measurements were taken: 1) beginning of test day,
2) periodically during the day, and 3) at the end of the test day. For information on
control block testing for specific XRF instruments and locations, refer to the testing
protocols described in appendix A.
4.5 Order of Testing
For the EPA/HUD field study, XRF testing was performed in a specific testing
order with respect to substrate type. All like substrates in a unit were grouped together
for testing before moving to the next substrate. An analysis of the data collected for the
field study gave indication that the grouped-substrate ordering did not affect the
accuracy of XRF measurement data. (A detailed description of the ordering and
analysis is provided in the technical report to the field study). XRF measurements of
archive samples taken in grouped-substrate order were compared to XRF
measurements of archive samples taken in random order. Similarly, results showed
that the ordering did not affect XRF measurements. These results are given below.
Therefore, for PCS development, archive samples were tested in a random order,
irrespective of substrate, to better match the ordering that would typically occur in a
lead-based paint inspection.
4.5.1 Archive Sample Ordering
In January 1995, the Pb Analyzer XRF was used for two rounds of archive
testing. For the first round of testing, the archive samples were tested using a grouped-
substrate ordering. For the second round, the archive samples were tested in random
order. A comparison of the two types of ordering was made by comparing the XRF
measurements taken during each round. Estimates of bias and precision, for non-
substrate corrected (uncorrected) XRF results, at four levels of lead in paint: 0.0, 0.5,
1.0, and 2.0 mg/cm2 were computed using a regression model for each round of testing.
(The issue of substrate correction is discussed in section 5.3). Table 4-6 provides the
bias estimates and Table 4-7 provides the precision estimates for each round of testing.
Comparisons of the values in these tables indicates that the order of testing has little
effect on these estimates.
24
-------
LJ j
CO ^^
(Jj "^^
^UJ
^ 0
^ E
T3 .
CO "o
CD 3
fi/ ^T^
^^ CyJ
CO
CO "fl>
C Q
< Z)
o|
r\i LJJ
CO 0
O .c
f^. *-•
__ C
5|
0 LJ.
0
"O "7^
^5 «3
•^ o
|>*U_
O CD
-i-" -t->
CO CO
11
5 W
O ^"»
1-
•^la?
S 0 |
^^ (U Cu
CO fj d)
P >
ill
CO D <
i
•
co Q
S
=J
^
S
UJ
^ P
^™ Ml
K O
a:
UJ
a.
o
H
s<
p
z
5
UJ
s
UJ
0(1 a:
UJ
a.
s
S
I
UJ
||
CO
UJ
1-
2
to
m
to
a
i-
to
CM
CO
CO
O
in
^:
,
5
CO
CM
t-.
O
•
in
CO
o
o
o
CO
a
m
o
o
o
d
,
m
o
o
o
o
•
§
o
o
o
CM
£
i_
o
c
o
O
CO
o
o
CO
d
in
CM
d
o
CM
o
S
o
o
o
d
S
8
o
*
I
Q
C
f
<-
<
CO
o
*-
r--
co
CO
CO
in
•^
m
CO
o
CM
CO
o
o
d
o
o
o
8
2
i
CO
00
CO
o
CO
in
d
PI
T~
CO
o
CM
O
CO
d
CM
O
O
O
CO
CO
CO
Q.
CO
00
in
£
d
CO
CM
CM
CM
CO
CD
CM
s?
o
CM
00
O
d
S
o
o
o
CO
CO
•o
o
1
CO
^
O)
o
S
in
CO
d
CO
CO
a>
o
CO
o
o
d
T—
O
o
o
o
S
1
CO
CO
CO
'-
00
CO
in
CO
d
r-
o
CO
o
S
d
8
o
o
o
CO
CM
CM
ID
S:
o
o
O
c-
o
S
d
00
o
d
§
o
CM
O
O
CM
O
O
d
o
o
o
o
5
"S
Q
•c
"a
il
J
T—
S
CO
o
CO
d
=2
o
CO
o
o
d
CM
o
o
o
o
h-
CM
0)
5j
CO
CO
CM
CO
CM
d
CM
o
lr~
CO
CM
O
O
d
CO
0
8
o
CM
"55
CO
Q-
§
CO
^1
d
CO
CD
CO
T-^
S
r—
^
o
S
d
o
o
o
o
00
T3
0
1
CO
cu
N
^
CO
CO
CO
E
tn
0
CD
T3
in
93
in
in
cu
fi
1
0
a
o
£
0)
S
CO
"c
a
in
•o
CO
CM
CU
SI
IT
LO
CN
-------
Table 4-6. Bias Comparison Calculated Based on Measurements Taken by the Pb
Analyzer in Grouped-Substrate Order and in Random Order.
LEAD LEVEL
(mg/cm2)
0.0
0.5
1.0
2.0
BIAS (mg/cm2)
Grouped-Substrate
Order
0.00
0.19
0.37
0.75
Random
Order
0.03
0.21
0.38
0.73
Table 4-7. Precision Comparison Calculated Based on Measurements Taken by the
Pb Analyzer in Grouped-Substrate Order and in Random Order.
LEAD LEVEL
(mg/cm2)
0.0
0.5
1.0
2.0
PRECISION' (mg/cm2)
Grouped-Substrate
Order
0.11
0.37
0.52
0.72
Random
Order
0.12
0.35
0.47
0.66
'Precision at 1 standard deviation
The nonparametric sign test was used to test XRF measurements for differences
in results from the two rounds of testing. The XRF measurements from the first round
of testing were paired with XRF measurements from the same sample in the second
round of testing. A total of 153 measurement pairs were used in the sign test. In 75
pairs, the measurements taken in random order were greater than measurements taken
in grouped-substrate order, in 68 pairs the measurements taken in grouped-substrate
order were greater, and in 10 pairs the measurements were equal. This result is
consistent, at the 5 percent level, with the null hypothesis that there is no effect
attributable to the order in which the measurements were taken.
4.6 Testing Protocols
Testing protocols for each XRF instrument under evaluation were developed
from manufacturer-supplied documents such as the manufacturers' instrument
operating manuals. Efforts were made to perform XRF testing in a manner consistent
with manufacturer guidance subject to the restriction that no destructive testing is
26
-------
permitted. Sampling locations were tested using extensive before, during, and after
quality control checks on NIST SRM films placed over control blocks of representative
substrates. Testing at each sampling location included XRF measurements on one or
more painted surfaces plus XRF measurements on a bare substrate area both with and
without being covered by the 1.02 mg/cm2 lead NIST SRM film. Appendix A provides
additional information about the XRF instrument testing protocols.
4.7 EPA/HUD Field Study Data vs. Archive Facility Data
XRF Performance Characteristic Sheets were developed using data collected
either from the EPA/HUD field study or from testing archive samples. It should be
noted that the archive samples are not a subset of those in the field study, but a smaller
set that largely overlaps the field study samples as well. A comparison of the field study
data to the archive facility data can be made by comparing the performance of the Pb
Analyzer in the field study with its performance in archive testing. Results from the Pb
Analyzer were selected for this comparison since the instrument used to test the
archives was the same physical instrument used to test a majority of the test locations
in the field study. Furthermore, the same individual operated the Pb Analyzer for all of
the testing on archives and field samples.
The data collected from the EPA/HUD field study (and reported in A Field Test of
Lead-Based Paint Testing Technologies: Technical Report, EPA 747-R-95-002b, May
1995) are designated below as "FIELD". The data collected from the archive testing is
designated below as "ARCHIVE".
Comparisons of the information presented on PCSs are the basis for making
comparisons of the performance of the Pb Analyzer. Results from each of the following
are examined and discussed below: 1) substrate correction recommendations, 2)
calibration checks, 3) inconclusive ranges and thresholds, 4) bias, 5) precision, and 6)
empirical results. In general, these results do not indicate better performance on the
archive samples. However, some differences were observed and are stated below.
1) Substrate correction recommendation: There is no need of, or benefit from,
substrate correction as indicated by the results from both the archive testing and the
field study.
2) Calibration checks: Calibration checks, computed from results of the field
study and archive testing, are presented in Table 4-8. A calibration check consists of
an interval bounded at the lower end by a "minus value" and at the upper end by a "plus
value." The minus and plus values, which are estimated from measurements taken on
red NIST SRM films (1.02 mg/cm2 lead level) placed over wood control blocks, are used
to determine if an XRF machine is operating within acceptable calibration parameters,
27
-------
as explained in section 5.5. It is seen in Table 4-8 that the widths of the two intervals
are the same, but the interval from the field study is more nearly centered about zero.
Table 4-8. Calibration Check Comparison Computed from Measurements Taken by
the Pb Analyzer.
TYPE
Minus value
Plus value
CALIBRATION CHECK (mg/cm2)
Field
-0.3
+0.4
Archive
-0.2
+0.5
3) Bias: The bias of the field study measurements and the archive testing
measurements are provided below in Table 4-9. These bias figures are estimates
derived from fitting a model to the data, as explained in section 5.1. A comparison of
the biases from the two sources shows that in all but one case, the absolute value of
the bias estimates for the field measurements were less than those for the archive
measurements. On the archive, Pb Analyzer measurements exhibited very little bias at
lead levels near 0.0 mg/cm2, but the bias was positive and became progressively larger
as the lead level increased. Therefore, some differences in bias can be observed when
comparing Pb Analyzer measurements on field study samples to archive samples. It is
important to note, however, that other instruments have demonstrated both high and
low bias on the archive samples. The XL demonstrated low bias on the archive
samples. The LPA-1, on one occasion, demonstrated low bias and on another
occasion, demonstrated high bias on the archive samples. These biases may be
observed in the PCSs provided in appendix D. Thus, there is a lack of evidence to
conclude that the bias with the archive samples differs from the field study samples
bias.
4) Precision: The precision of the field study measurements and the archive
testing measurements are provided in Table 4-10. These precision figures are
estimates derived from fitting the same model to the data that gave the bias estimates
presented in Table 4-9. As occurred for the bias comparison, in all but one case the
precision estimates of the field measurements were less than or equal to those for the
archive measurements for the Pb Analyzer. However, other XRF instruments have
measured field and archive samples with varying degrees of precision as shown in the
PCSs in appendix D. Therefore, there is a lack of evidence to conclude that the
precision of measurement of archive samples differs from that of the field study.
28
-------
Table 4-9. Bias Comparison for Measurements Taken by the Pb Analyzer.
LEAD LEVEL
(mg/cm2)
0.0
0.5
1.0
2.0
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
BIAS (mg/cm2)
Field
0.0
0.0
0.0
1.0
0.0
0.0
0.0
0.0
0.1
0.0
0.0
0.0
0.0
0.0
0.2
0.0
-0.1
0.3
0.0
0.0
0.4
0.0
-0.3
0.6
Archive
0.0
0.0
0.0
0.0
0.0
0.0
0.2
0.2
0.2
0.2
0.2
0.2
0.4
0.4
0.4
0.4
0.4
0.4
0.7
0.7
0.7
0.7
0.7
0.7
5) Inconclusive ranges and thresholds: The inconclusive ranges and thresholds
computed from results of the field study and the archive testing are provided in Table 4-
11. Section 5 provides details on how these values were computed. Comparison of
the results from the two sources shows differences in thresholds and inconclusive
ranges. Thresholds are present for brick, concrete, and plaster substrates for the field
results whereas no thresholds are shown for any of the substrates for the archive
results. For the other three substrates, the inconclusive ranges are all wider for the
archive results. However, the thresholds and inconclusive ranges determined from
EPA/HUD filed study data are included within the inconclusive ranges determined from
archive data.
29
-------
Table 4-10. Precision Comparison for Measurements Taken by the Pb Analyzer.
LEAD LEVEL
(mg/cm2)
0.0
0.5
1.0
2.0
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
PRECISION" (mg/cm2)
Field
0.1
0.1
0.1
0.2
0.1
0.1
0.3
0.3
0.2
0.3
0.2
0.3
0.4
0.4
0.3
0.4
0.2
0.5
0.5
0.5
0.4
0.5
0.3
0.6
Archive
0.1
0.1
0.1
0.1
0.1
0.1
0.3
0.3
0.3
0.3
0.3
0.3
0.5
0.5
0.5
0.5
0.5
0.5
0.7
0.7
0.7
0.7
0.7
0.7
"Precision at 1 standard deviation
6) Empirical results: The overall misclassification (error) rates and inconclusive
rates that resulted from applying the Table 4-11 inconclusive ranges and thresholds to
the test data are given in Table 4-12. Two of the three rates are lowest for the field
study: the false positive rate and the inconclusive rate.
30
-------
Table 4-11. Inconclusive Ranges and Threshold Comparison for Measurements
Taken by the Pb Analyzer.
DATA SOURCE
EPA/HUD field study
Archive Testing
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
THRESHOLD
(mg/cm2)
1.0
1.0
None
None
0.9
None
None
None
None
None
None
None
INCONCLUSIVE
RANGE (mg/cm2)
None
None
0 9 to 1 .2
0.9 to 1.1
None
0.9 to 1.3
0.9 to 1.4
0.9 to 1.4
0.9 to 1.4
0.9 to 1.4
0.9 to 1.4
0.9 to 1.4
Table 4-12. Overall Misclassification and Inconclusive Rates Comparison for
Measurements Taken by the Pb Analyzer.
DESCRIPTION
Field study results
Archive testing results
FALSE POSITIVE
20/717(2.8%)
7/113(6.2%)
FALSE NEGATIVE
12/220(5.5%)
1/41 (2.4%)
INCONCLUSIVE
21/1190= 1.8%
7/154 =4.6%
4.8 EPA/HUD Field Study Data vs. National Survey Data
In 1990, a HUD-sponsored national survey of lead-based paint in housing
("National Survey") was conducted and has been reported in Report on the National
Survey of Lead-Based Paint in Housing, Base Report, EPA 747-R-95-003, April, 1995;
Report on the National Survey of Lead-Based Paint in Housing, Appendix I: Design
and Methodology, EPA 747-R-95-004, April, 1995; and Report on the National Survey
of Lead-Based Paint in Housing, Appendix II: Analysis, EPA 747-R-95-005, April, 1995.
These reports are available from the National Lead Information Center (1-800-424-
LEAD). The National Survey was carefully designed to give representative samples of
the private and public housing stocks in the U.S. It is therefore of interest to determine
the extent to which the lead levels represented in the EPA/HUD field study agree with
those from the National Survey.
31
-------
Summary statistics for XRF measured lead levels reported for the National
Survey were compared to results found in the EPA/HUD field study. This comparison
found that the reported lead levels for the two studies appear to be similar. It should be
noted that the comparison was limited in scope due to the fact that the two studies had
different objectives. Some of the similarities and differences between the two studies
as well as a brief description of the comparison and results are described below.
XRF measurements reported by the MAP-3 XRF instruments in Denver and
Philadelphia for the EPA/HUD field study were used to compare to the MAP-3 XRF
instrument results reported for the National Survey. The measurements taken from the
field study are averages of three screen mode values whereas the measurements
taken in the National Survey were single test mode readings. The results reported for
Denver were compared to the National Survey results for private housing. Similarly, the
results reported for Philadelphia for the field study were compared to the National
Survey public housing results. This was done since all housing tested in Denver were
single-family homes and the housing tested in Philadelphia were part of a multifamily
housing development.
Prior to comparing the MAP-3 measurements taken from the two studies, it was
necessary to account for the censoring of the readings taken for the National Survey.
The instruments used for the National Survey never reported a reading less than zero
(that is, negative readings were never reported), even though readings less than zero
were otherwise possible. Thus, all negative readings reported for the EPA/HUD field
study were set equal to zero before summary statistics were computed.
Tables 4-13 and 4-14 below provide summary statistics for censored results for
readings taken by the two MAP-3 "field classifications" in the EPA/HUD field study. A
field classification represents a set of 1,190 readings taken at all of the sample
locations in Denver and Philadelphia. The fact that there were two field classifications
for the MAP-3 means that there are two complete sets of readings for this instrument on
a common set of samples. One MAP-3 machine made all readings in the first field
classification (Field Study I); two different MAP-3 machines were used in the second
field classification (Field Study II). Results for the two field classifications are shown in
Tables 4-13 and 4-14. Also provided in these tables are National Survey results, as
reported in Appendix II of the above mentioned publication. A comparison of the
summary statistics reveals similar results, particularly when comparing means and
medians. However, some differences may be observed when comparing the statistics
for the common areas. The lead levels of the common areas in the National Survey are
more varied and have higher means and medians.
32
-------
Table 4-13. Percentile and Mean for XRF Measurements for Public Housing Units by
Sample Location (mg/cm2).
STATISTIC
Minimum
1 %
5%
10%
25%
Median
75%
90%
95%
99%
Maximum
Mean
Std. Dev.
No. of Samples
INTERIOR
Field
Study I
0.00
0.00
0.00
0.00
0.00
0.27
0.86
2.23
3.56
5.23
6.64
0.72
1.15
385
Field
Study II
0.00
0.00
0.00
0.00
0.00
0.14
0.84
2.11
3.25
5.00
6.54
0.66
1.10
385
National
Survey
0.00
0.00
0.00
0.00
0.05
0.21
0.68
1.74
2.64
4.78
12.76
0.58
2.81
1,731
COMMON AREAS
Field
Study I
0.00
0.00
0.00
0.00
0.00
0.00
0.30
2.56
4.50
6.66
6.66
0.62
1.43
55
Field
Study II
0.00
0.00
0.00
0.00
0.00
0.00
0.32
2.71
4.60
7.36
7.36
0.65
1.51
55
National
Survey
0.00
0.00
0.00
0.00
0.06
0.31
1.08
2.42
4.58
23.28
23.96
1.17
7.90
L_ 553
Table 4-14. Percentile and Mean for XRF Measurements for Private Housing Units by
Sample Location (mg/cm2).
STATISTIC
Minimum
1 %
5%
10%
25%
Median
75%
90%
95%
99%
Maximum
Mean
Std. Dev.
No. of Samples
INTERIOR
Field
Study I
0.00
0.00
0.00
0.00
0.00
0.05
0.39
1.18
3.61
18.58
26.73
0.90
3.24
447
Field
Study II
0.00
0.00
0.00
0.00
0.00
0.03
0.38
1.08
3.70
18.79
25.59
0.87
3.08
447
National
Survey
0.00
0.00
0.00
0.00
0.03
0.19
0.60
1.66
4.49
10.18
21.82
0.81
1.95
4,273
EXTERIOR
Field
Study I
0.00
0.00
0.00
0.00
0.00
0.35
2.53
6.35
12.83
22.05
28.03
2.28
4.44
303
Field
Study II
0.00
0.00
0.00
0.00
0.00
0.36
2.56
5.90
9.68
22.95
29.94
2.14
4.24
303
National
Survey
0.00
0.00
0.00
0.00
0.05
0.42
1.85
5.81
9.30
2771
53.81
2.07
4.64
1,047
33
-------
5. PCS DEVELOPMENT AND STATISTICAL METHODOLOGY
Numerical entities on a PCS related to XRF instrument performance are
estimates obtained from field testing data. This section describes the statistical
methodology used in PCS-related estimation. The estimates obtained by applying the
methodology described in this section are given in appendix D. Empirical classification
results based on these estimates are also given in appendix D.
Some of the material in this chapter is unavoidably technical, in particular
sections 5.1.1, 5.1.6, 5.1.7, and 5.6.1. Readers who do not have a background in
statistics, or are not interested in mathematical details, can skip over these sections
with little loss of comprehension of the remaining material.
5.1 Bias and Precision Estimation
The PCS reports estimates of bias and precision, for uncorrected XRF results, at
four levels of lead in paint: 0.0, 0.5, 1.0, and 2.0 mg/cm2. If substrate correction is
recommended on some or all substrates, estimates of bias and precision of the
corrected measurements at the 0.5 mg/cm2 and 2.0 mg/cm2 lead levels are also
required to calculate an inconclusive range or threshold, although the estimates do not
appear in the PCS but are shown in appendix D. Since estimation is based on XRF
results obtained on field samples, two fundamental issues have to be addressed in
order to estimate bias and precision at the four indicated lead levels:
The lead levels of field samples, in the archive as in the EPA/HUD field study,
are distributed towards lower values, with diminishing concentration of samples
near any fixed lead level as the lead level increases;
Lead levels are themselves estimated by laboratory analysis of paint samples
using ICP-AES ("ICP"), which accounts for some of the apparent imprecision of
XRF results.
These two factors make it impossible to directly observe the bias and precision
of XRF results under field conditions, or at pre-specified lead levels. Estimation of
these quantities is performed using the model introduced in the EPA/HUD field study.
5.1.1 The XRF Measurement Model
The XRF measurement model is a mathematical expression that relates XRF
results to the true level of lead in paint, accounting for the fact that the lead level is
35
-------
estimated by ICP. The form of the model used in PCS development is defined by the
following two equations:
(1 ) XRF = a +
log(ICP) = log(Pb) + 5.
In this model, XRF and ICP are observable quantities representing the XRF
measurement and the ICP-measured lead level respectively. The true lead level,
represented by Pb, is not observable. The terms e, rand 5 are independent, normal
random variables each having mean 0, and variances given by Var(s) = c, Var(j) = d,
and Var(d) = o/. The model parameters (a,b,c,d) describe the linkage of an XRF
measurement to the true lead level:
• The expression a + b-(Pb) defines a linear regression relationship between the
mean XRF measurement and the true lead level. The intercept a represents the
mean XRF measurement when the lead level is 0.0 mg/cm2. The slope b
measures the change in the mean XRF measurement per unit change in the true
lead level.
The expression s + r(Pb)'A is the residual (error) term in the linear regression.
This residual has mean zero, and variance given by the expression c + d-(Pb).
In other words, the residual variance is not constant with respect to the lead
level, but increases linearly with it. When the lead level is 0.0 mg/cm2 the
residual variance is equal to c. Parameter d measures the change in variance
per unit change in the true lead level.
Under the model, the bias (8) and precision (SD) at a fixed lead level Pb = p can
be expressed as follows:
(2) B(p) = a + (b-1)-p
SD(p) = [c + d-p]1/2
The quantities B(p) and SDfpj are conceptual, representing true, innate characteristics
of an XRF instrument that can only be inferred from data obtained with the instrument.
This is done by using the data to obtain estimates of the model parameters (a,b,c,d).
Bias and precision estimates can then be obtained at any lead level p by substitution,
36
-------
using the above formulas for B(p) and SD(p). In PCS development, estimates for B and
SD at the lead levels p = 0.0, 0.5, 1 .0 and 2.0 (mg/cm2) are reported.
The term a/ is the variance of the error caused by using log(ICP) as a substitute
for log(Pb). A value for this term must be obtained externally, as it was in the EPA/HUD
field study. Based on an analysis of laboratory duplicate ICP measurements from the
field study, it was found that os- .3 for Denver samples and o5= .2 for Philadelphia and
Louisville samples were reasonable values to use in PCS-related work.
The model represented by (1) differs slightly from the model developed in the
EPA/HUD field study. The model used in the field study has the following form:
(3) XRF = a +
log(ICP) = log(Pb) + 5,
which leads to the following expressions for the bias and precision:
(4) B0(p) = a + (b-1)-p
SD0(p) = [c + d-p2]1".
The only difference between the two forms of the XRF measurement model
represented by (1) and (3) is the manner in which the precision is expressed. In (2),
SD(p) increases at a rate approximately proportional to the square root of p. In (4),
SD0(p) increases at a rate approximately proportional to the lead level p itself. As a
result, SD0(p) can be substantially larger than SD(p) for high lead levels. It was found
that the precision estimates provided by SD0(p) and SD(p) were similar for lead levels
up to 1 .0 mg/cm2, and that both closely agreed with nonparametric estimates within this
lead range. At the 2.0 mg/cm2 lead level, SD0(p) usually gave higher estimates than
SD(p). The decision to use SD(p) instead of SD0(p) in PCS development was
motivated by the fact that the estimates provided by SD(p) at the 2.0 mg/cm2 lead level
agreed more closely with nonparametric estimates4 than the estimates provided by
SD0(p).
Parameter estimates in the XRF measurement model are obtained using
maximum likelihood. It is assumed that the true distribution of lead levels is lognormal,
"The nonparametric procedure used is based on fitting monotone (also referred to as
"isotonic") regressions, as described in the technical report to the EPA/HUD field study.
37
-------
and that os is known. For Denver samples and EPA's Office of Research and
Development samples, o5 = 0.3, and for Philadelphia and Louisville samples, o5 = 0.2,
are used in PCS-related estimation. The joint density of XRF measurement ("XRF")
and ICP is then approximated using numerical integration at every (XRF, ICP) data pair.
The maximum likelihood estimate (MLE) is obtained by maximizing the product of the
densities using a Newton-Raphson algorithm. An estimated covariance matrix of the
MLE is obtained upon convergence of the algorithm using asymptotic theory, which
gives standard errors of the parameter estimates and of related quantities. The
constraints c,d > 0 are imposed in order to ensure feasible solutions. These details are
further elaborated in the technical report to the EPA/HUD field study.
5.1.2 Data Used in Model Estimation
Only XRF results for which the representative ICP-measured lead level is less
than 4.0 mg/cm2 are used in estimating the XRF measurement model parameters. XRF
results on brick and concrete samples in the archive are excluded from model
estimation. Treatment of brick and concrete results in archive testing is discussed in
section 5.1.3. The designation of a representative ICP measurement for a sample is
discussed in section 5.2. The 4.0 mg/cm2 upper limit is imposed because
characteristics of instrument performance relevant to PCS development and addressed
with the model involve lead levels of 2.0 mg/cm2 or less. And, because a number of
XRF instruments do not read above a certain value, restricting the range of lead levels
makes it less likely that truncated XRF measurements would complicate a model-based
analysis. The 4.0 mg/cm2 upper limit is the same as that adopted for substrate
correction when needed, as explained in section 5.3.2.
XRF results identified as outliers are also excluded from analyses. Outliers,
based on the derivation of nonparametric standardized residuals as defined in the
EPA/HUD field study technical report, are also excluded from model estimation. An
outlier is indicated if the nonparametric standardized residual exceeds 3.3 in absolute
value. This criterion typically results in the designation of no more than one
measurement (out of more than 150) as an outlier.
Measurements made on one of the archive samples are excluded from model
estimation because of concern regarding the accuracy of the representative ICP
measurement. This sample has a wood substrate, and an ICP measurement less than
0.1 mg/cm2. With restriction of the data to ICP less than or equal to 4.0 mg/cm2,
exclusion of brick and concrete archive samples, and exclusion of the mentioned wood
38
-------
sample, there are 139 measurements per archive test available for model fitting: 14
drywall, 38 metal, 34 plaster, and 53 wood.
5.1.3 Estimation Using XRF Instrument Testing Data
The evaluation of XRF instruments in the EPA/HUD field study differed from
archive testing in several important respects. The field study was conducted at more
than 1,000 sample locations, in contrast to the approximately 150 samples available for
testing in the archive. In most cases, several machines of the same instrument type
were typically used in the field study, with different operators5. For example, 5 different
Microlead I machines were used by four different operators in the field study. Machines
and/or operators were changed according to conditions that prevailed during the
progress of the field study. In contrast, a single archive test of an XRF instrument
normally consists of one machine used by one operator. If multiple rounds of archive
testing are conducted with an XRF instrument, possibly with different machines and/or
operators, each set of XRF results is obtained on common samples. Two rounds of
testing are desired for PCS development so as to capture information on XRF inter-
machine variability at minimum cost.
This section describes how model estimates are obtained with a single machine
of an XRF instrument type, either in the EPA/HUD field study or in the archive. An
explanation of the methods used to combine results obtained with multiple machines of
a given XRF instrument type is provided in section 5.1.4.
Model estimates based on the EPA/HUD field study data were obtained by
substrate, with several qualifications related to brick. For the Microlead I, model
estimates for brick were not obtained, because the model did not adequately describe
the features of this instrument on brick substrates. The estimates reported are
averages and standard deviations obtained for individual machines, which were then
pooled. Only XRF results for samples with ICP measurements less than 1.0 mg/cm2
were used in this analysis, because it was found that Microlead I measurements were
essentially unaffected by the lead level within this range, thereby justifying the use of
simple summary statistics. The sparseness of lead levels between 1.0 mg/cm2 and 4.0
mg/cm2 made it difficult to infer whether this condition persisted for lead levels in the
5Again, the distinction between "instrument" and "machine" is emphasized:
"instrument" refers to a particular make or brand, "machine" to a specific physical unit
of a particular make or brand.
39
-------
wider range. For the MAP-3, brick and concrete data were combined, because the
model did not give stable estimates for brick alone under the ICP less than or equal to
4.0 mg/cm2 range restriction.
Fitting XRF measurement models separately by substrate cannot give reliable
bias and precision estimates based on archive testing data, due to the small sample
sizes by substrate indicated in section 5.1.2. Consequently, it is necessary to pool data
across all substrates for the purpose of estimation. In order to recognize substrate
differences, three indicator (dummy) variables are used for drywall, metal and plaster
substrates within the scope of a single model. To illustrate, if a sample has a plaster
substrate, the indicator variables for drywall and metal are set to 0, and the indicator
variable for plaster is set to equal to 1. If a sample has a wood substrate, all three
indicator variables are set to 0. As explained in section 5.1.2, brick and concrete
sample measurements are not used in model estimation. The use of indicator variables
allows separate, substrate-specific bias estimates to be obtained. The estimates are
based on a common slope parameter (b) and on common variation parameters (c and
cf). Thus, the difference in bias estimates at two different lead levels is the same for all
substrates. Similarly, the precision estimates are the same for all substrates.
The indicator variable model has 7 parameters (a,b,D,M,P,c,d), with D, /Wand P
denoting coefficients for the drywall, metal, and plaster substrate indicator variables,
respectively. Bias estimates for wood are obtained by setting the three indicator
variables to zero. The following demonstrates how bias and precision estimates are
obtained, by substrate, at lead level p (mg/cm2):
SUBSTRATE BIAS PRECISION
Drywall a + (b-1)-p + D [c + d- p]1/2
Metal a + (b-1)-p + M [c + d- p]1/2
Plaster a + (b-1)-p + P [c + d- p]m
Wood a + (b-1)-p [c + d-pf2
Separate substrate estimates are given only if the indicator variable model has a
significant (5% level) likelihood ratio test (LRT) with respect to the model having no
indicator variables. The LRT statistic has an approximate chi-square distribution with
three degrees of freedom. If the LRT statistic is less than 7.8, which is the 5% critical
value, it is concluded that no significant differences between the 4 substrates are found,
and the same bias is reported for each of the substrates.
40
-------
Estimation for brick and concrete substrates requires a different approach, since
the archive contains only 3 brick and 2 concrete samples. Applying the 4.0 mg/cm2
cutoff to these substrates would eliminate 2 of the 3 brick samples. Increasing the
cutoff to 15 mg/cm2 adds back one brick sample, which provides some, albeit limited,
information on how XRF instruments perform at a high lead level.
The following methodology is used to obtain combined estimates for brick and
concrete substrates. The statistic T = Yw - a - b-Xw is computed using the 2 brick and 2
concrete measurements on samples with ICP less than or equal to 15 mg/cm2, where
Yw is the weighted average XRF result; Xw is the weighed average ICP measurement;
and a and b are model parameters based on drywall, metal, plaster and wood samples.
The weights are reciprocals of [c + d-x,]1/i, where x, is one of the four ICP
measurements, and c and d are model parameters. An adjustment to the bias
estimates for brick and concrete is made only if the absolute value of T exceeds two
times its estimated standard error. If this condition is satisfied, estimates of the form
a + (b-1)-p + Tare reported for the bias on brick and concrete. If this condition is not
satisfied, the estimates for brick and concrete are the same as those reported for wood.
5.1.4 Combining XRF Results from Different Testing Scenarios
An important finding of the EPA/HUD field study is that variation in performance
between different machines of an XRF instrument type can be substantial. Archive
testing continues to validate this observation. Intermachine variation is recognized in
the methods used to combine XRF results from different testing scenarios. This section
explains how the results of XRF instrument testing are combined using different
machines in the field study; using different archive tests (typically with different
machines); and, combining field study and archive testing data. The distinction
between machines and instruments is again emphasized; the beginning of section 5.1.3
explains the distinction and its differing implications for field study and archive testing.
PCSs were developed from EPA/HUD field study, or archive data, or both.
PCSs for the Princeton Gamma-Tech XK-3, the Scitec MAP-3, and the Warrington
Microlead I Revision 4 were developed from data collected during the EPA/HUD field
study. A PCS for the TN Technologies Pb Analyzer was developed from the field study
and from two separate testing events at the archive facility. PCSs for the Niton XL-309,
the Radiation Monitoring Device LPA-1, the Advanced Detectors LeadStar, and the
Scitec MAP 4 were developed from archive testing data. Table 5-1 reports the number
of machines that were used in the EPA/HUD field study, upon which testing data in
41
-------
PCS development is based and shows where the testing was done. The LeadStar PCS
guidance distinguishes between instruments with software versions earlier than version
4.1 and those with software version 4.1 through version 4.3. Thus, on Table 5-1, there
is a separate entry for each software version. Similarly, the LPA-1 PCS provides
guidance for two categories of instruments. Guidance is provided for instruments
purchased before June 26, 1995 and have not been serviced since June 26, 1995 and
also for instruments sold or serviced after June 26, 1995. So there is a separate entry
for each LPA-1 instrument category for the LPA-1 on Table 5-1.
Table 5-1. Number of XRF Machines Used for PCS Development by Origin of Data
for Each XRF Instrument.
XRF INSTRUMENT MODEL3
XK-3
MAP-3
MAP 4
Microlead I
Pb Analyzer
XL
LPA-1 b
LPA-1 c
Loadstar"
LeadStar6
ORIGIN OF DATA
EPA/HUD FIELD STUDY
3
4
0
5
2
0
0
0
0
0
ARCHIVE FACILITY
0
0
2
0
2
1
2
1
2
2
aPCS guidance applies to all versions of each model except as noted below.
b Used for developing PCS guidance for LPA-1 instruments purchased before 6/26/95 and had not been
serviced since 6/26/95.
°Used for developing PCS guidance for LPA-1 instruments sold or serviced after June 26, 1995.
d Used for developing PCS guidance for LeadStar instruments with software versions earlier than
version 4. 1 .
6 Used for developing PCS guidance for LeadStar instruments with software versions 4.1 to version 4.3.
42
-------
Three of the four K-shell instruments tested in the EPA/HUD field study (MAP-3,
Microlead I and XK-3) exhibited significant differences between machines. For these
instruments, model parameters were estimated separately for each machine, and the
results pooled by averaging with weights based on sample sizes. Pooling results,
instead of data, avoided inflating the estimated SDs of the instrument due to
intermachine differences. It also avoided the problem of mixing repeated
measurements taken at the same sample locations, which applies to all of the K-shell
instruments except the Pb Analyzer. The use of unequal weights in pooling reflects the
circumstances by which different machines came into use during the course of the field
study. Variation in sample sizes, sample lead levels, and possibly other factors that
could affect the reliability of estimates obtained for individual machines, motivated the
decision not to weight estimates equally.
Sample sizes by machine for a given instrument varied by substrate, and in
some cases were too small for accurate model estimation. Estimates for an individual
machine were obtained provided that there were at least 25 observations available. For
the Microlead I, two machines exhibited problems on certain substrates. For one
machine, data were used only in obtaining pooled estimates for brick. For the other
one, data were used only in obtaining pooled estimates for brick and drywall.
If the PCS is based on separate archive tests made with the same type of
instrument, the estimated model parameters are averaged to produce bias and
precision estimates. Precision estimates are not inflated due to intermachine or other
differences that may have existed between tests. Ranges are reported along with the
bias and precision estimates if data from more than one archive test are used. The
model form with the least aggregation across substrates (see section 5.1.3) is used as
the basis for pooling. For example, if the first archive test produces a model parameter
estimate indicating significant substrate differences, but the second archive test of the
same XRF instrument does not, the same model with indicator variables for drywall,
metal and plaster is estimated with both sets of data separately before pooling. The
unweighted average used in pooling assigns equal importance to each archive test that
was conducted.
If the PCS is based on a combination of EPA/HUD field study and archive
testing, pooling is conducted in two stages: (1) weighted averaging of estimated model
parameters obtained by machine in the field study, as described above in this section;
(2) unweighted averaging of the field study estimate obtained in (1) with the estimated
model parameters obtained for each archive test. The second stage treats the field
study evaluation of the XRF instrument, with different machines, essentially the same
43
-------
as one archive test, despite the larger sample size of the field study. This is done to
prevent the field study evaluation from dominating the archive testing evaluation in PCS
development.
5.1.5 Standard Error Estimates
An approximate covariance matrix of estimated model parameters, based on
non-repeated measurements with a single machine, is obtained using standard
maximum likelihood theory. For bias and precision estimates derived from the model
parameters, standard error (SE) estimates are obtained from the estimated covariance
matrix. When model estimates are pooled using repeated measurements at common
sample locations, which occurs after multiple rounds of archive testing, conservative SE
estimates are obtained using the triangle inequality, which states that the SE for the
average of two estimates is less than or equal to the average of the SEs.
5.1.6 Relation to Previous Work
The XRF measurement model represented by (1) is an example of a nonlinear,
heteroscedastic measurement error model. Fuller (1987) is a standard reference for
measurement error models, and it describes the use of maximum likelihood estimation
for these models. Provided that the explanatory variable (in the present case, the true
lead levels of the tested samples) can be regarded as randomly selected from a
population and regularity conditions are met, maximum likelihood estimates are
asymptotically normal and efficient. The use of numerical integration to approximate
the likelihood function is recognized as a necessity when the required probability
densities cannot be obtained in closed form.
5.2 Laboratory Measurement of Lead Levels
In the EPA/HUD field study, a primary paint sample was defined for each sample
location (field sample). A laboratory ICP measurement based on the primary paint
sample6 constituted the representative lead-level measurement for the field sample.
6A modified NIOSH Method 7082 was used to prepare the paint samples. The
NIOSH Method 7082 is designed to prepare and analyze air filter samples for analysis
of many inorganic elements, including lead. Modifications were needed to make it
44
-------
This ICP measurement was, in most cases, the only one taken. It serves as the "best"
estimate of the lead level in PCS development that is based on the field study data.
By contrast, all archive samples have one primary and at least one additional
ICP measurement. An archive sample that was also in the EPA/HUD field study
inherits the ICP measurement taken on the primary paint sample in the field study as its
primary ICP measurement. Primary ICP measurements for archive samples that were
not in the field study are based on paint samples collected according to protocols
similar to those used in the field study. The paint sample collection protocols are
described in detail in the technical report to the EPA/HUD field study. A second set of
paint samples was collected from each archive sample after the archive was
assembled. Secondary ICP measurements are based on these additional paint
samples.
Secondary ICP measurements were made so that, when combined with the
primary ICP measurements, improved estimates of the lead levels of paint at XRF
testing locations may be obtained. The manner in which this combination should be
made is, at the time this document is published, an unresolved issue that may be
addressed in the future. At the present time, the primary ICP measurements are used
as representative lead-level measurements for the archive samples, as they were in the
field study evaluation.
5.3 Substrate Correction
Substrate correction is recommended for an XRF instrument if data suggest that
it can reduce bias to a significant degree. Substrate correction is performed by
subtracting from the XRF results a correction value determined separately in each
house for single-family housing or in each development for multifamily housing, for each
substrate recommended for correction. To obtain the correction value, three
measurements are taken at each of two areas of the substrate type in question, with
red NIST SRM film (1.02 mg/cm2 lead level) placed over the areas that had been
scraped clean of their paint covering. The correction value is the average of the six
readings minus 1.02 mg/cm2. Substrate correction may be recommended on all, some,
or no substrates.
applicable to processing paint samples. Details are provided in the technical report to
the EPA/HUD field study.
45
-------
5.3.1 Criteria for Recommending Substrate Correction
For PCSs based on EPA/HUD field study data, substrate correction is
recommended in accordance with the study conclusions on the effectiveness of red
NIST SRM average correction as described in the technical report to the field study.
For PCSs based on archive testing data, a recommendation for substrate correction is
based on the criteria given below.
1. Is the estimated bias significantly different from zero and at least 0.1
mg/cm2 in absolute value at the 0.0 mg/cm2 and 1.0 mg/cm2 lead levels?
2. Is the correction value, when computed from archive testing results7,
significantly different from zero?
3. Does correction of XRF results on substrates meeting the first two
conditions result in a decrease in the magnitude of the estimated bias of
at least 0.1 mg/cm2?
Substrate correction is recommended if significant bias is detected on a
substrate and if evidence suggests that substrate correction can reduce the bias to a
significant extent. If substrate correction is not recommended, it is possible that
significant bias was detected, but that correction was not found to be an effective
remedy. Recommendations regarding substrate correction should not be used to draw
inferences about the bias of an XRF instrument.
If the PCS is based on repeated archive tests with the same type of instrument,
substrate correction is recommended on those substrates for which the above criteria
are met in at least one of the tests. For example, if the need for substrate correction
was indicated on metal and wood in the first round of archive testing of instrument XYZ,
and on metal and plaster in the second round of archive testing, the PCS for instrument
XYZ based on the combination of the two tests will recommend substrate correction on
metal, plaster and wood. The PCS for the Pb Analyzer, which does not recommend
using substrate correction, was the only PCS developed using both EPA/HUD field
study and archive testing data. Thus, substrate correction methodology was not
developed for PCSs derived from both EPA/HUD field study and archive testing data.
7A correction value, based on archive testing results, is the average of all XRF
measurements taken on the red NIST SRM covering the bare substrate areas of the
same substrate type, minus 1.02.
46
-------
5.3.2 Measurement Range Subject to Substrate Correction
A recommendation of substrate correction applies only to XRF results less than
4.0 mg/cm2. This is because XRF results of 4.0 mg/cm2 or greater do reliably indicate
the presence of lead without the use of substrate correction across a wide range of
instruments. Table 5-2 presents results from the EPA/HUD field study that support this
finding. For each of the four K-shell XRF instruments evaluated in the field study, fewer
than 5% of XRF readings greater than or equal to 4.0 mg/cm2 were obtained on
samples with ICP-measured lead levels less than 1.0 mg/cm2. It is noteworthy that a
reduction of the cutoff value to 3.0 mg/cm2 would substantially increase the rate of false
positive results especially for the Microlead I and XK-3 instruments, both of which were
found to benefit from substrate correction. The 95% confidence intervals presented in
Table 5-2 demonstrate that the false positive rates for the Microlead I and XK-3 are
significantly greater than 5% with a 3.0 mg/cm2 cutoff value, and that none of the
instruments considered had false positive rates that are significantly greater than 5%
with a 4.0 mg/cm2 cutoff value. Thus, the use of 4.0 mg/cm2 as a cutoff value for
substrate correction can be expected to produce an acceptably small percentage of
false positive results due to uncorrected readings, with a minimal amount of
unnecessary correction.
An alternative approach would be to treat the substrate correction cutoff value as
an instrument-specific parameter to be estimated from the testing data. The advantage
to this approach is that instruments with cutoff values lower than 4.0 mg/cm2 would
require substrate correction less frequently than if a single 4.0 mg/cm2 cutoff value were
used for all instruments. The estimated cutoff value for an instrument would, however,
reflect the testing characteristics of only a small number of machines. And, the
sparseness of higher lead levels in the archive may make accurate estimation of the
cutoff value difficult. Whether or not the balance of arguments favors the use of
instrument-specific cutoff values for substrate correction is a topic that merits
exploration in future PCS development activity.
47
-------
Table 5-2. Frequencies and Percentages With 95% Confidence Intervals of False
Positive Results' for XRF Readings Obtained When the Uncorrected
XRF Reading is Greater Than or Equal to 2.0 mg/cm2, 3.0 mg/cm2, or 4.0
mg/cm2, Based on the EPA/HUD Field Study Data.
INSTRUMENT
Pb Analyzer
MAP-3
Microlead 1
XK-3
All Instruments
Combined
MEASURE
Frequency
Percentage
95% Conf. Int.*
Frequency
Percentage
95% Conf. Int.
Frequency
Percentage
95% Conf. Int.
Frequency
Percentage
95% Conf. Int.
Frequency
Percentage
95% Conf. Int.
UNCORRECTED XRF READINGS
Greater Than or Equal to
2.0
2/156
1.3%
0.2. 4.6
21 / 343
6.1%
3.9. 9.4
75 / 446
16.8%
13.5.20.7
261 /657
39.7%
36.0. 43.6
359/1602
22.4%
20.4, 24.6
3.0
0/106
0.0%
0.0. 3.4
5/240
2.1%
0.7.4.8
24 / 298
8.1%
5.3. 11.2
38 / 336
11.3%
8.2. 15.3
67 / 980
6.8%
5.3, 8.4
4.0
0/76
0.0%
0.0.4.7
1/161
0.6%
0.0. 3.4
9/214
4.2%
1.9.8.1
7/214
3.3%
1.3.6.9
17/665
2.6%
1.5,4.1
'A false positive result refers to an XRF reading of 1 .0 mg/cm2 or greater when the ICP-measured lead level
is less than 1 .0 mg/cm2.
*Upper and lower bounds for a 95% confidence interval for the percentage.
5.4 Inconclusive Ranges and Thresholds
The purpose of an inconclusive range or threshold is to give a rule by which an
XRF result, corrected for substrate bias as needed, can classify the lead level of a
painted surface relative to the 1.0 mg/cm2 federal standard for lead in paint. Negative,
positive, and inconclusive classifications are the possible outcomes when using an
48
-------
inconclusive range. A positive classification is an inference that the lead level is greater
than or equal to the 1.0 mg/cm2 federal standard, and a negative classification is an
inference that the lead level is less than the 1.0 mg/cm2 federal standard. An
inconclusive classification means that the XRF result cannot reliably distinguish
between positive and negative classifications, and that laboratory confirmation is
required to resolve the ambiguity. Negative and positive classifications are the only
possible outcomes when using a threshold; there is no inconclusive classification. The
procedures for classifying XRF results when using an inconclusive range or a threshold
are described in section 5.4.1.
Inconclusive ranges and thresholds are derived for an XRF instrument by
substrate. Inconclusive ranges and thresholds are designed to achieve, approximately:
A five percent (5%) rate of false positive results, over a representative range of
lead levels below the 1.0 mg/cm2 federal standard;
A five percent (5%) rate of false negative results, over a representative range of
lead levels at or above the 1.0 mg/cm2 federal standard;
A minimal rate of inconclusive results.
If substrate correction is recommended, the inconclusive range or threshold
reported is applicable to substrate corrected XRF results only. Bias and precision
values, which are based on uncorrected XRF results, may not appear to be consistent
with the inconclusive range or threshold reported if substrate correction is
recommended. Section 5.4.6 presents examples of how inconclusive ranges and
thresholds are calculated.
5.4.1 Classification Using Inconclusive Ranges and Thresholds
An inconclusive range is expressed as a lower (xj and an upper (x^) number.
With an inconclusive range, an XRF result is classified as:
Negative, if it is less than or equal to XL;
Positive, if it is greater than or equal to xu\
Inconclusive, if it is greater than XL and it is less than xu.
49
-------
For example, with an inconclusive range of 0.6 to 1.1 (mg/cm2), an XRF result of 0.6 is
classified as negative. If substrate correction is recommended for the XRF instrument
on a substrate being tested, the correction must be performed before making reference
to an inconclusive range.
A threshold is expressed as a single number (xr). With a threshold, an XRF
result is classified as:
Negative, if it is less than XT;
Positive, if it is greater than or equal to xr.
For example, with a threshold of 0.6 an XRF result of 0.6 mg/cm2 is classified as
positive. There is no inconclusive classification with a threshold. If substrate correction
is recommended for the XRF instrument on a substrate being tested, the XRF result
must be substrate corrected, as indicated, before reference to a threshold is made.
5.4.2 Data Used in Inconclusive Range and Threshold Derivation
Inconclusive ranges and thresholds are estimated for an XRF instrument using
the same testing data from which bias and precision estimates are derived. Data are
pooled across machines and/or substrates according to the conventions outlined in
sections 5.1.3 and 5.1.4 for bias and precision estimation. Since the variation in XRF
measurements for particular machines has been found to be substantially greater than
the variation across machines for the instruments tested to date, adjustments for
intermachine variation have not been made. The reported inconclusive ranges or
thresholds based on pooled results should provide adequate leeway for differences
between machines of the same instrument type, if the machines are similar to those
tested. In future testing, direct consideration of intermachine variation in the
computation of inconclusive ranges and thresholds may be necessary, depending on
the magnitude of intermachine variation relative to within-machine variation.
5.4.3 Model-Based Derivation of Inconclusive Ranges and Thresholds
Archive and/or EPA/HUD field study testing data are used to obtain estimates of
the parameters of the model described in section 5.1. The model, with its estimated
parameters, gives an approximate description of the probabilistic behavior of XRF
50
-------
measurements as a function of the lead level. Specifically, XRF measurements for a
single machine and at a fixed lead level Pb = p, are assumed to follow a normal
distribution, with bias B(p) and precision (standard deviation) SD(p), using the
terminology of section 5.1.1. These properties guide the construction of inconclusive
ranges and thresholds, according to the procedure explained in sections 5.4.5 through
5.4.9 below.
In principle, the upper and lower endpoints of an inconclusive range can be
estimated without a model. To estimate the lower endpoint, XRF results for samples
with ICP-measured lead levels greater than or equal to 1.0 mg/cm2 are examined. The
fifth percentile of these results constitutes an estimate of the lower endpoint. Similarly,
the 95th percentile of XRF results for samples with ICP less than 1.0 mg/cm2
constitutes an estimate of the upper endpoint. Although easy to derive, these
nonparametric estimates are highly variable, except in very large samples. The lower
endpoint of an inconclusive range based on a single archive test would in essence be
determined by the second smallest XRF result for these samples. Two unusually low
XRF results would be sufficient to create a wide inconclusive range. A model-based
approach avoids this difficulty by adopting reasonable assumptions about the
probabilistic nature of XRF measurements, based on results of the EPA/HUD field
study. Another factor in favor of a model-based approach is that the model takes into
account error in ICP-measured lead levels, which is not possible in the nonparametric
approach.
5.4.4 Number of Decimal Places Reported
Inconclusive ranges and thresholds are reported to one decimal place, by
rounding to the nearest tenth. Calculations used to derive these quantities are
performed on a computer using its full arithmetic accuracy. The use of rounded
quantities in calculations, such as bias and precision estimates reported on a PCS, may
produce slightly different results.
5.4.5 Criteria for Inconclusive Range and Threshold Derivation
The three objectives stated at the beginning of section 5.4 guide the derivation of
inconclusive ranges and thresholds. These objectives, which are to achieve a 5% rate
of false positives, a 5% rate of false negatives, and a minimal rate of inconclusive
classifications, are somewhat vague, because it was established in section 5.1 that the
51
-------
performance characteristics of an XRF instrument depend on the lead level of paint
being tested. It is therefore necessary to express these objectives in terms of criteria
that are achievable in specific terms.
The following six criteria are used to guide the derivation of inconclusive ranges
and thresholds that meet the objectives stated at the beginning of section 5.4:
1. The probability of a false negative classification cannot be greater than five
percent (5%) when the true lead level is 2.0 mg/cm2 or greater.
2. The probability of a false positive classification cannot be greater than five
percent (5%) when the true lead level is 0.5 mg/cm2 or smaller.
3. An XRF result of 1.0 mg/cm2 or greater cannot be classified as negative.
4. An XRF result of 0.0 mg/cm2 or less can only be classified as negative.
5. An XRF result that is greater than or equal to its expected value at the 1.0
mg/cm2 lead level cannot be classified as negative.
6. An XRF result that is less than its expected value at the 1.0 mg/cm2 lead level
cannot be classified as positive.
The lead levels 2.0 mg/cm2 and 0.5 mg/cm2 appearing in Criteria 1 and 2 are
called pivotal values. It is explained in section 5.4.6 below why, by obtaining 5% false
negative and 5% false positive classification rates at these pivotal lead levels, it is
possible to achieve, approximately, the same objectives over a range of lead levels
similar to that found in the EPA/HUD field study.
Criterion 3 mandates, for an inconclusive range, that the lower endpoint,
designated XL , be 0.9 or smaller, since larger values could lead to XRF results equal to
1.0 being classified as negative. For the same reason, thresholds, designated XT , must
be 1.0 or smaller. Criterion 4 mandates that XL cannot be negative, and that x7 cannot
be zero or less. Criteria 5 and 6 together state that if a threshold is obtained, it must
coincide with the expected value of an XRF result at the 1.0 mg/cm2 lead level. The
criteria are not symmetric with respect to the treatment of XRF results above and below
the 1.0 mg/cm2 federal standard. For instance, an XRF result of 1.1 classified as a
negative is not feasible, but an XRF result of 0.9 classified as a positive is feasible.
52
-------
5.4.6 Use of 0.5 mg/cm2 and 2.0 mg/cm2 as Pivotal Values
Criteria 1 and 2 in section 5.4.5 refer to the 0.5 mg/cm2 and 2.0 mg/cm2 lead
levels. These two pivotal values are used to represent lead levels below and above the
1.0 mg/cm2 federal standard for the purpose of calculating inconclusive ranges and
thresholds. Relative to the distribution of lead levels of the 1,290 EPA/HUD field study
samples, 0.5 mg/cm2 is slightly larger than the median of lead levels below the federal
standard, and 2.0 mg/cm2 is slightly smaller than the median of lead levels above the
federal standard. This property is desirable for remaining within the five percent false
positive and five percent false negative rate targets across samples having variable
lead levels.
Use of the pivotal values 0.5 mg/cm2 and 2.0 mg/cm2 was found to produce
inconclusive ranges and thresholds similar to those obtained with a more direct, and
more complicated, procedure that does not require the use of pivots. The procedure in
reference uses the model introduced in section 5.1 to compute the integrated estimated
probability of a false negative across a range of lead levels greater than or equal to 1.0
mg/cm2, and the integrated estimated probability of a false positive across a range of
lead levels below 1.0 mg/cm2. The integrated estimated probabilities are calculated
using lognormal distributions, by substrate, for lead levels similar to the observed
distribution of measured lead levels8 in the EPA/HUD field study. The integrated
estimated probabilities are given by the following expressions:
fProb(XRF < XL Pb=p)f(p)dp
FN(xL) = J
[f(p)dp
1
1
(Prob(XRF * x(J Pb=p)f(p)dp
FP(xu) = -2-
"Measured lead levels from the EPA/HUD field study were found to be comparable
in distribution to a national survey of lead-based paint in housing conducted in 1990
(see section 4.8).
53
-------
The probabilities inside the integrals are approximated by normal distributions
using parameter estimates obtained from the XRF measurement model. The lognormal
density is given by f(p). The upper (x^) and lower (XL) bounds of the inconclusive range
are set so that both FP(x(7) and FN(xJ are equal to 5%, using a line-search procedure.
Inconclusive ranges and thresholds for instruments tested in the EPA/HUD field study
obtained with the direct procedure were compared to inconclusive ranges and
thresholds calculated using a variety of pivotal values. The choice of 0.5 mg/cm2 and
2.0 mg/cm2 as pivotal values gave the best approximation to the direct procedure for
the range of instruments that were tested.
5.4.7 Properties of Inconclusive Ranges and Thresholds
Section 5.4.5 lists criteria which inconclusive ranges and thresholds must satisfy.
The inconclusive range or threshold obtained for an instrument reflects the objectives
that motivated its derivation, and the constraints under which it was derived.
Interpretation of an inconclusive range or threshold in light of other performance
characteristics of the XRF instrument, such as bias and precision estimates, should
take this factor into account. For example, an XRF instrument that gives unbiased
results at all lead levels does not necessarily have an inconclusive range that is
symmetrical with respect to the 1.0 mg/cm2 federal standard. Additionally, positive and
negative bias at the 1.0 mg/cm2 lead level do not affect the process of obtaining an
inconclusive range or threshold in the same manner. The following is a review of some
of the properties of inconclusive ranges and thresholds to aid in their interpretation:
Asymmetrical inconclusive ranges. There are two reasons why an inconclusive
range may appear off-centered with respect to the bias exhibited at the 1.0 mg/cm2 lead
level: (1) the pivotal values 0.5 mg/cm2 and 2.0 mg/cm2 are not centered about 1.0
mg/cm2, and (2) most XRF instruments have precision (SD) estimates that increase with
the lead level. The choice of the indicated pivotal values, which is explained in section
5.4.6, reflects the skewed distribution of lead levels found in the EPA/HUD field study,
which in turn is a reasonable reflection of lead levels likely to be encountered in
practical lead-based paint testing. Inconclusive ranges and thresholds are designed to
be the smallest feasible solutions that give approximately 5% false positive and 5%
false negative classifications under practical testing conditions. The imposition of a
symmetry requirement would lead to wider inconclusive ranges; thresholds replaced
with inconclusive ranges; and, ultimately, a greater proportion of inconclusive
classifications.
54
-------
Disparate treatment of bias. Criterion 3 mandates that an XRF result that is 1.0 or
greater cannot be classified as negative, regardless of the bias. To illustrate how this
affects the derivation of an inconclusive range or threshold, what would have been a
threshold of 1.3 is converted into an inconclusive range of 0.9 to 1.3. This is necessary
to avoid classifying an XRF result of 1.2, for example, as negative. Bias of the opposite
kind, however, is not treated in the same manner: a threshold of 0.7, for example, does
not violate Criterion 3. As a result, an XRF instruments that has high precision but
systematic, positive bias may report an inconclusive range on its PCS, while a different
instrument with inferior precision and systematic, negative bias of the same magnitude
may report a threshold. This discussion underscores the invalidity of using an
inconclusive range or threshold as a stand-alone measure of the accuracy or precision
of an XRF instrument.
Meaning of the 5% targets. The 5% targets for false positive and false negative
classifications are across ranges of lead levels less than 1.0 mg/cm2 and greater than
or equal to 1.0 mg/cm2, respectively. The probability of a false positive classification at
a fixed lead level slightly below 1.0 mg/cm2 will be greater than 5%, as will the
probability of a false negative classification at a fixed lead level slightly above 1.0
mg/cm2.
5.4.8 Calculation of Inconclusive Ranges and Thresholds
The process leading to the calculation of inconclusive ranges and thresholds is
central to PCS development:
1. XRF testing data are obtained for a particular instrument, from the EPA/HUD
field study and/or the archive.
2. Testing data are used to estimate parameters of the model described in section
5.1. Results from several tests are combined in the manner described in section
5.1.4.
3. Bias and precision estimates are obtained, using the estimated model
parameters, at the following lead levels: 0.5, 1.0 and 2.0 mg/cm2. (Bias and
precision estimates were also obtained at 0.0 mg/cm2 lead, but, these estimates
were not used to develop PCS inconclusive ranges or thresholds.)
55
-------
4. Bias and precision estimates obtained in step 3 are used to derive an
inconclusive range or threshold that conforms to the six criteria outlined in
section 5.4.5.
The calculation of inconclusive ranges and thresholds is illustrated in this section
with three examples. In each example the estimated model parameters, which are
given, are assumed to have been obtained using maximum likelihood estimation.
Example 1: Archive testing of an XRF instrument results in the following estimated
model parameters, following the procedure outlined in section 5.1.3:
a = 0.15,b = 1.08, c = 0.12,d = 0.05.
Substrate correction is assumed not to have been recommended on any substrate.
The same bias and precision estimates are reported for all substrates on the PCS, and
a common inconclusive range or threshold is also reported. The expected value (EV) of
an XRF result at lead level p is estimated by EV(p) = 0.15 + 1.08-p, and the SD is
estimated by SD(p) = [0.12 + 0.05-p]1/2. The following table gives values for the three
lead levels used in the derivation of inconclusive ranges and thresholds.
ji_ EV(p) SDfp)
0.5 0.69 0.38
1.0 1.23 0.41
2.0 2.31 0.47
The first step is to use estimates from the table to derive a Lower Interval at the
0.5 mg/cm2 lead level, and an Upper Interval at the 2.0 mg/cm2 lead level:
Lower Interval = EV(0.5) ± 1.645- SD(0.5)
= 0.69 ±1.645-(0.38)
= 0.1 to 1.3
Upper Interval = EV(2.0) ± 1.645- SD(2.0)
= 2.31 ±1.645-(0.47)
= 1.5 to 3.1
The Lower Interval is chosen so that there is a five percent (approximate)
probability of obtaining an XRF result greater than 1.3 if the lead level is 0.5 mg/cm2.
Similarly, the Upper Interval is chosen so that there is about a five percent probability of
obtaining an XRF result less than 1.5 if the lead level is 2.0 mg/cm2. In order to meet
56
-------
criteria 1 and 2, any interval with XL less than or equal to 1.5 and xu greater than or
equal to 1.3 would work. Threshold values xr set anywhere between 1.3 and 1.5 would
also work.
The second, and final, step is to modify the inconclusive range or threshold as
needed to satisfy criteria 3 through 6. Since EV(1.0) - 1.23 is not between 1.3 and 1.5,
a threshold solution is not feasible (criterion 5), making 1.2 to 1.3 the narrowest
possible inconclusive range. Since the lower endpoint cannot be 1.0 or greater
(criterion 3), the narrowest possible inconclusive range is 0.9 to 1.3, which reflects
rounding to one decimal place. The process of determining this inconclusive range is
depicted graphically below.
Example 1
Lower Interval: Upper Interval:
0.7 ± 0.6 = 0.1 to 1.3 2.3 ± 0.8 = 1.5 to 3.1
°-1 1-3 2.0 3.0 3.5
I I
o.O 0.9 1-° 1.5
INCONCLUSIVE RANGE is the interval
0.9 to 1.3 mg/cm2, since it Is the narrowest
feasible interval that contains both 1.0
mg/cm2 and EV(1) = 1 .23 mg/cm2
3.1
Example 2: An XRF instrument is tested twice on the archive. On both occasions, the
procedure outlined in section 5.1.4 points to the use of a model with indicator variables
for drywall, metal and plaster. Since the same sample size applies to both tests,
pooling takes the form of averaging the model parameters. The following are the
estimated model parameters for the two tests.
Parameter
a (wood)
b
D (drywall)
M (metal)
P (plaster)
c
d
Testl
-0.116
1.126
0.098
-0.302
0.004
0.066
0.104
Test 2
-0.052
0.965
0.149
-0.186
0.014
0.048
0.112
Pooled
-0.084
1.046
0.124
-0.244
0.009
0.057
0.108
Suppose that the criteria for substrate correction outlined in section 5.3.1 are met
for metal in Test 1, but not in Test 2. As explained in section 5.3.1, metal is
recommended for substrate correction using the pooled results, because it is
57
-------
recommended at least once. Suppose that the pooled correction factor for metal,
based on averaging correction factors for Tests 1 and 2, is -0.28. Brick and concrete
results are assumed to be reported using parameters (a,b,c,d) of the pooled model, as
explained in section 5.1.4, which therefore coincide with results for wood.
The following illustrates the derivation of the inconclusive range or threshold for
metal. With substrate correction, the expected value for an XRF result at lead level p is
estimated by
EV(p) = a + M + b-p - {metalcorrection}
= -0.048+1.046-p
The SD is estimated by SDfp; = [0.057 + 0.108-p]1*, the same for all substrates. The
following table gives values for the three lead levels used in the derivation.
M- EV(p) SDfp)
0.5 0.48 0.33
1.0 1.00 0.41
2.0 2.04 0.52
As in Example 1, the first step is to obtain a Lower Interval and an Upper Interval, as
shown below.
Lower Interval = EV(0.5) ± 1.645- SD(0.5)
= 0.48 ±1.645-(0.33)
= -0.1 to 1.0
Upper Interval = EV(2.0) ± 1.645- SD(2.0)
= 2.04 ±1.645-(0.52)
= 1.2 to 2.9
Since the two intervals do not overlap, a threshold xr chosen between 1.0 and 1.2
meets criteria 1 and 2, but only 1.0 is also feasible under criterion 3. Since the
expected value at 1.0 mg/cm2 is 1.00, a threshold solution xr = 1.0 is reported for
substrate corrected metal on the PCS. The process of determining this threshold
solution is depicted graphically below.
58
-------
Example 2
Lower Interval:
0.48 ±0.54 = -0.1 to 1.0
-0.1
-0.5
0.0
1.0
Upper Interval:
2.04 ± 0.86 = 1.2 to 2.9
2.0
3.0
mg/cm!
1.0
1.2
2.9
THRESHOLD is 1.0 mg/cmz, since
1.0 mg/cm2 lies between the
Lower and Upper Intervals
Example 3: Data from archive testing of an XRF instrument yield the following
estimated model parameters:
a = 0.1, b = 0.8,c = 0.1, d = 0.2.
These estimates apply to all substrates, and substrate correction is not recommended.
The following table gives values for the three lead levels needed in the derivation of an
inconclusive range or threshold.
M.
0.5
1.0
2.0
0.50
0.90
1.70
SD(p)
0.45
0.55
0.71
From these values the following Lower and Upper Intervals are obtained:
Lower Interval
Upper Interval
= EVY0.5;±1.645-SD(0.5)
= 0.50 ±1.645-(0.45)
= -0.2 to 1.2
= EV(2.0) ± 1.645- SD(2.0)
= 1.70 ±1.645-(0.71)
= 0.5 to 2.9
Unlike the first two examples, the Lower and Upper Intervals overlap in the range 0.5 to
1.2. Because this overlap interval contains both 1.0 and EV(1.0) = 0.90, it is not
necessary to widen it further in order to satisfy the 6 criteria for inconclusive ranges
given in section 5.4.5. The inconclusive range for this example is therefore 0.5 to 1.2,
as illustrated below.
59
-------
Example 3
Lower Interval: Upper Interval:
0.5 ±0.7 = -0.2 to 1.2 1.7 ± 1.2 =0.5 to 2.9
. - . •
-0'5 °-° 0.5 2.9
INCONCLUSIVE RANGE
is the Overlap Interval:
0.5 to 1.2 mg/cm2
mg/cmj
5.4.9 Classification Performance of Inconclusive Ranges and Thresholds
Inconclusive ranges and thresholds are derived from model estimates. The
observed (empirical) rates of false positive and false negative classifications based on
testing data, and reported in appendix D, are subject to variation from the five percent
targets that are set for these rates. Small sample sizes, especially from archive testing,
should be taken into account when comparing the empirical rates to the targeted five
percent rate. For instance, there are 45 archive samples with ICP-measured lead
levels greater than 1.0 mg/cm2. If only three false negative results are observed, the
empirical false negative rate is 6.7 percent. With four false negative classifications the
rate becomes 8.9 percent, seemingly in violation of the five percent target. In fact, four
out of 45 false negative results is an outcome compatible (at the 95 percent confidence
level) with a true false negative rate of five percent.
The inconclusive ranges and thresholds reported on a PCS are estimates
obtained from data on one, or perhaps several, machines of an XRF instrument type.
They are designed for machines similar to those that were tested. The extent to which
the machine or machines that were tested are representative of a larger population of
machines cannot be formally determined. Consequently, a different machine of the
same instrument type may have false positive or false negative rates that are higher
than the five percent targets. The use of substrate correction where indicated was
found to significantly reduce variation between machines, and should broaden the
applicability of the inconclusive rates and thresholds reported on a PCS beyond the
machines that were tested.
5.5 XRF Calibration Check Methodology
The XRF calibration check reported on a PCS is performed using the average of
three consecutive XRF results obtained on red NIST SRM film placed over a wood
control block, minus the 1.02 mg/cm2 lead level of red NIST SRM film. Failure of the
60
-------
calibration check occurs if this quantity is less than the "minus value" or greater than the
"plus value" reported in the PCS. In the event of failure, the PCS recommends that the
manufacturer's instructions be followed to bring the machine into control. The minus
and plus values are derived so that there is approximately a 1 in 200 (one-half of one
percent) chance of failing a properly calibrated instrument.
The minus and plus values are estimates derived from testing one or more
machines of an XRF instrument type on red NIST SRM film placed over wood control
blocks. For each machine that was tested, bias and variance estimates are obtained
from averages of three successive wood control readings. If more than one machine
was tested, the separate machine estimates are combined to give pooled bias and
within-machines variance estimates. These combined estimates are used to calculate
the minus and plus values for the XRF calibration check. In two-way analysis of
variance (ANOVA) terminology, the pooled bias is the grand mean; the within-machines
variance is a weighted average of the separate machine variance estimates. The
within-machines variance does not reflect variation between machines. If only one
machine was tested, the bias and variance estimates for that machine are used.
If 6 represents the bias estimate and SD the square root of the variance
estimate, the minus and plus values are computed as follows:
minus value = 6 - tk 0025- SD
plus value = B + tk 0025- SD,
where k is the number of degrees of freedom (total sample size minus the number of
machines tested), and tk m25 is the upper .0025 probability cutoff for a Student's t
distribution with k degrees of freedom. For k greater than 30, tk 0025 is approximately
equal to 3. For k equal to 10, tk,0025 is approximately equal to 4. Instruments tested in
the EPA/HUD field study had greater than 30 triplicate wood control readings. In
archive testing, fewer triplicate wood control readings are made, with 10 or 11 such
triplicate readings being typical.
5.6 XRF Re-evaluation Test Methodology
Chapter 7 of the HUD Guidelines, as implemented in XRF Performance
Characteristic Sheets, provides a method to be used for evaluating XRF testing by
means of a re-test of selected components followed by comparison of the original and
re-test results. Although the details differ in single- and multifamily units, the basic idea
is to re-test ten randomly selected testing combinations (defined as locations on painted
surfaces) and then compare the average of the ten re-tests to the original average. If
61
-------
the difference of the two averages exceeds a calculated tolerance, a potential problem
is indicated and an additional ten re-tests are performed at random. If the second re-
test is also out of tolerance, a failure of the re-test is declared. This creates a prima
facie case for deficiencies in the inspection, ranging from sloppy procedures to outright
fraud, and indicates the need for a more thorough examination of the testing that has
been performed. The re-test procedure is calibrated to result in spurious failures in
approximately 1% of inspections. That is, the re-test procedure will indicate the need
for further examination of the testing in about 1 out of 100 cases where there is actually
no problem, and the re-test failure is merely a random event.
This section describes the statistical methodology underlying the re-test method,
for both single-family and multifamily applications. Examples are presented to illustrate
the calculations needed for the re-test. Finally, an error in some of the previously
published versions of the re-test method is discussed, and its impact illustrated through
an example.
5.6.1 Re-Testing in Multifamily Housing
In multifamily housing, an XRF result has been defined as a single reading taken
on a testing combination, defined as a location on the painted surface of a building
component such as a wall, door or baseboard. To implement the re-test in a multifamily
development, select two units at random from those already tested in the development.
Then select 5 testing combinations in each unit which are known to have been
previously tested, although the exact location of the test will usually be unavailable. For
example, the re-tester will generally know only that a wall in a certain room was tested.
He will usually not know the exact tested spot on the wall, and may not even know
which wall in the room was previously selected. The re-test selection procedure gives a
total often testing combinations to be re-tested. Where possible, the re-testing should
be done using the same instrument and inspector as the original test, that is, the re-
testing should be carried out while the original inspection team is still on-site and
available.
Let X, 1f I = 1,...,10, denote the original XRF result on the ith testing combination
selected for the re-test. Similarly, let Xii2 denote the re-test result on the ith combination.
Assuming that both the original and the re-test are honest measurements, X;§1 and X, 2
have the same distribution, which can be represented by the sum of a normal and a
lognormal random variable having mean u, and variance a2. Here, u, represents the
true average lead level, in mg/cm2, on the testing combination. For example, u, might
62
-------
represent the average lead level on the walls in a room. The variance o2 represents the
combined effect of (normal) XRF measurement error and (lognormal) spatial variation in
the true lead level across the testing combination. Thus,
O2 = O,2 + 022,
where o,2 represents XRF measurement error and o22 represents spatial variation in
true lead across the testing combination. The technical report to the EPA/HUD field
study sheds light on the possible magnitude of a,2 and o22. Table 2-3 of the technical
report indicates a range of standard deviations for single K-shell XRF readings of from
0.03 mg/cm2 to 0.72 mg/cm2, depending on the instrument, substrate and lead level. In
the interests of simplicity of the re-test method, a value a., = 0.4 mg/cm2 (that is, a* =
0.16) was selected from this range as a reasonably representative value.
The technical report demonstrated that the magnitude of spatial variation in lead
levels across a testing combination was proportional to the true lead level. That is,
there was greater absolute variation in lead levels across more highly leaded
components. This leads to the equation
°22 = C2- Mi2,
where "C" is a constant factor. Table 4-21 of the technical report to the field study,
which shows the results of analysis conducted on a logarithmic scale, indicates a value
C = 0.27 based on field duplicate samples taken a distance 9 inches apart in multifamily
dwellings in Philadelphia. The EPA/HUD field study sheds no light on spatial variation
between samples taken greater distances apart. However, such variation is likely to be
greater than that reported in the field study. To allow for this fact, a value C = 0.6 was
chosen to represent spatial variation in lead levels across an entire testing combination.
This leads to the following statistical model for re-testing data:
2- u2) = NLN(u,0.16+0.36u2),
and
X12 ~ NLNdJ.o^+C2- u,2) = NLN(u,0.16+0.36u,2),
where u is the true (unknown) average lead level on the testing combination, and NLN
refers to the fact that the underlying distribution is that of the sum of a normal and a
lognormal random variable.
Let XBAR, and XBAR2 denote the average of the ten original XRF results and
the ten re-test XRF results, respectively. That is,
63
-------
= (Xu+...+X10,k)-MO,k=1,2.
Then, assuming statistical independence of the original and re-test XRF measurements,
it follows that, approximately,
XBAR, - XBAR2 - N(0,0.032+0.0072(|J12 + ... + Mio2))-
where again u, is the true but unknown average lead level in mg/cm2 on the ith testing
combination, I = 1,...,10. Observe that the underlying distribution is now normal,
instead of the normal-lognormal combination used for the individual XRF results. A
remark on the use of a normal approximation for XBAR, - XBAR2 can be found at the
end of this section. The best available estimate of u( is the average
Mi = (X,, + Xii2) + 2.
This leads to the approximate equation
XBARi - XBAR2 ~ N(0,0.032+0.0072- (M,2 + ... + M102)).
The approximate tolerance value for the re-test is then defined as
XBAR1 - XBAR2 > 1.645- (0.032+0.0072- (M,2 + ... + M102))y*.
Assuming no true difference between the original and re-test results, the
probability of exceeding the critical value is approximately 10%, from the standard
normal theory. Thus, the probability of spurious failure of two re-tests is 10%- 10% =
1%.
The statement that XBAR, - XBAR2 approximately follows a normal distribution,
in spite of the fact that its averaged constituents do not, is a consequence of the Central
Limit Theorem. To illustrate the effectiveness of the approximation, 10,000 values of
XBAR1 - XBAR2 were randomly simulated on a computer, based on X;f1 and X, 2
generated as NLN(u,0.16+0.36u2) for u = 1, and i ranging from 1 to 10. The following
table gives various percentiles of the distribution obtained from the 10,000 simulated
values (actual) and from the normal approximation (theory):
64
-------
Percentile Theory
5th -0.53
25th -0.22
50th 0.00
75th 0.22
95th 0.53
It is clear that the normal approximation agrees closely with the actual
distribution of XBAF^ - XBAR2, and that its use in testing the difference between the
original and re-test results is justified.
5.6.2 Re-Testing in Single-Family Housing
In single-family housing, an XRF result has been defined as the average of three
readings taken at randomly selected locations on a testing combination. This is
because spatial variation in lead levels across housing components was found, in the
above-referenced EPA/HUD field study, to be larger in single-family housing than in
multifamily housing. The re-test procedure is similar. Select ten testing combinations
at random in the single-family dwelling and re-test them with the same instrument and
operator. Apply the same tolerance calculation that was used in the multifamily case,
except that three-reading averages are used as the XRF results in all cases instead of
single readings.
The statistical properties of the re-test are similar to those for the multifamily
case, despite the difference between single and triple readings. There are two reasons
for this. First, the EPA/HUD field study found that XRF measurement variability was
generally very similar whether a single reading or the average of three readings was
used to define the XRF result. Thus, the XRF measurement component of variability is
similar in the multifamily and single-family re-tests. With regard to the spatial variation
in lead levels across a testing combination, Table 4-21 of the technical report shows a
standard deviation of 0.47 on a logarithmic scale for single-family housing in Denver as
compared to a standard deviation of 0.27 for multifamily housing in Philadelphia. Thus,
the standard deviation for spatial variation (on the measurement scale) is estimated as
0.47- u for single-family housing as compared to 0.27- u for multifamily housing. Thus,
the standard deviation for the average of three readings in single family dwellings is
estimated as 0.47- u+V3 = 0.27- u, which is the same as the standard deviation for a
65
-------
single reading in the multifamily case. Thus, the spatial variability for the average of
three readings in the single-family case is estimated to be the same as the spatial
variability for a single reading in the multifamily case. The bottom line is that the
statistical properties of a single reading in the multifamily case are similar to those of
the average of three readings at randomly selected locations in the single-family case.
Hence, the same re-test formulas apply to the single-family and multifamily cases,
bearing in mind the different definitions of XRF result.
5.6.3 Examples of Re-Test Calculations
Example 1: Suppose the following original and re-test results were obtained (mg/cm2):
Original: 1.2, 2.1, 3.0, 1.5, 1.6, 4.5, 2.0, 0.2, 0.1, 0.0.
Re-Test: 1.6, 1.8, 2.5, 1.9, 1.3, 3.7, 3.2, 0.0, 0.7, 0.5.
The average of the ten original XRF results is XBARi = 1.62, while the average
of the re-tests is XBAR2 = 1.72. Thus, XBARi - XBAR2 = 0.1. The critical value is
computed as follows. First, the averages (M,, i =1,...10) of the original and re-test
results by testing combination are computed to be:
1.4, 1.95, 2.75, 1.7, 1.45, 4.1, 2.6, 0.1, 0.4, and 0.25.
Next, the sum of the squared averages is computed to be:
M,2 + ... + M102 = 1.42 + ... + 0.252 = 42.12.
Thus, the critical value is
1.645- (0.032 + 0.0072 • 42.12)y* = 0.95.
Therefore, since 0.1 < 0.95, the re-test is passed in this case.
Example 2: Suppose the following original and re-test results were obtained (mg/cm2):
Original: 1.2, 2.1, 3.0, 1.5, 1.6, 4.5, 2.0, 0.2, 0.1, 0.0.
Re-Test: 4.0, 1.8, 2.5, 1.9, 1.3, 3.7, 3.2, 4.0, 0.7, 5.0.
66
-------
Clearly, several of the re-test results differ markedly from the original results in
this case. The difference of the averages is XBAR., - XBAR2| =1.19. In this case, the
critical value is 1.10, so the re-test fails, necessitating a second re-test. For a second
re-test, another, different set often randomly selected testing combinations is selected.
Suppose the following second original and second re-test results were obtained
(mg/cm2):
2nd Original: 0.9, 1.2, 1.1, 1.5, 2.4, 0.1, 2.2, 0.1, 0.1, 0.2.
2nd Re-Test: 2.6, 2.7, 3.2, 3.8, 5.2, 1.3, 3.1, 4.0, 1.7, 1.4.
The average of the second ten original XRF results is XBAR., = 0.98, while the
average of the second re-tests is XBAR2 = 3.16. Thus, |XBAR1 - XBAR2| = 2.18. The
critical value is computed by first finding the averages of the second original and
second re-test results by testing combination. Next, the sum of the squared averages
is computed. The critical value was computed to be 0.99. As with the first re-test,
several of the second re-test results differ markedly from the second original results. In
this case, the critical value is 0.99, so the second re-test fails since the 2.18 > 0.99.
Whenever both re-tests fail, the inspection should be considered deficient.
5.6.4 Error in Re-Test Formula
The re-test formula is presented in words in the 1995 HUD Guidelines and in
some of the previously published XRF Performance Characteristic Sheets is incorrect.
The wording has been corrected on all of the XRF Performance Characteristic Sheets
in Appendix D. The following wording (that appeared in some cases):
"Compute the square of each of the ten original and ten retest XRF results. Add
these squares of XRF results together. Call this quantity C."
is incorrect, and should be replaced by the following:
"Compute the average of the original and re-test result for each of the ten
testing combinations. Square the average for each testing combination.
Add the ten squared averages together. Call this quantity C."
The net effect of the erroneous wording is to make it more difficult for a re-test to
fail. This reduces the frequency of spurious failures, but also makes it more difficult to
67
-------
detect instances of fraud or other problems. As an example, consider the test/re-test
results in Example 2 of section 5.6.3. The use of the erroneous formula has the effect
in this example of increasing the critical value from 1.10 to 1.68, which would mean that
the re-test would pass instead of failing. The wording will be corrected in subsequent
versions of the XRF Performance Characteristic Sheets.
5.7 XRF Results That Do Not Consist of Fixed Reading Time Measurements
In the EPA/HUD field study, XRF instruments were evaluated relative to a fixed,
nominal 15-second reading time, or, in one case, a fixed, nominal 60-second reading
time. The XRF measurement model (section 5.1.1) was developed for XRF results
obtained with fixed reading times. Subsequent to the completion of the field study,
several XRF instruments emerged that do not report a fixed reading time measurement
as its outcome. In some cases, a random reading time is used, which may be
determined in various ways, such as (1) reading until the machine determines that a
target precision level (for example, 0.2 mg/cm2) has been reached, or (2) reading until
the machine determines that sufficient information is available to distinguish the lead
level from being above or below a certain target value, such as the 1.0 mg/cm2 federal
standard. In at least one other case, several nominally-fixed reading time K- and L-
shell readings are processed together in an algorithm to determine whether the lead
level is as above or below a target value.
An approach for analyzing variable time XRF measurements is described in
appendix B. Two PCSs with variable time modes are the LeadStar and MAP 4 PCSs.
These PCSs start on pages D-3 and D-60, respectively, in appendix D.
68
-------
6.0 REFERENCES
1. U.S. Environmental Protection Agency (1997), Archive Operations and Protocols,
EPA Report Number 747-R-97-004.
2. Siegmund, D., (1985), Sequential Analysis, Springer-Verlag: New York.
3. Stromberg, U, (1991), Computational Statistics and Data Analysis (11) 205-219.
4. Title X--Residential Lead-Based Paint Hazard Reduction Act of 1992, Public Law
102-550.
5. U.S. Environmental Protection Agency (1995), A Field Test of Lead-Based Paint
Testing Technologies: Technical Report, EPA Report No. 747-R-95-002b.
6. U.S. Environmental Protection Agency (1995), Report on the National Survey of
Lead-Based Paint in Housing, Appendix II: Analysis, EPA Report No. 747-R-95-005.
69
-------
APPENDICES
-------
APPENDIX A: XRF INSTRUMENT OPERATING PROTOCOLS
A.1 Introduction
PCS development applied XRF instrument results from testing samples from
either the EPA/HUD field study or the archive facility. Testing protocols used by XRF
instruments in the EPA/HUD field study can be found in the appendices of the report
entitled A Field Test of Lead-Based Paint Testing Technologies: Technical Report
(EPA 747-R-95-002b). This report can be ordered from the National Technical
Information Service (NTIS) at 703-487-4650 (NTIS reference number, PB96-125026).
A detailed description of the XRF testing methodologies used in the EPA/HUD field
study, some of which also apply to testing the archive samples, is provided in Chapter 3
of the technical report. A familiarity with these methodologies may be beneficial to the
reader. Provided in this appendix are descriptions of the testing protocols used by XRF
instruments to test the archive samples. The document entitled Archive Operations and
Protocols (EPA 747-R-97-004) provides detailed descriptions of the testing protocols.
A.2 Brief Description of Testing Areas
As many as four specific test areas are found on each archive sample and are
referred to as X1, X2, X3, and BARE. The protocols in this appendix refer to these
specific test areas. A brief description of each follows.
X1 is the primary painted test area on the sample.
X2 and X3 are the secondary and tertiary painted test areas. Test data taken from
the X2 and X3 test areas were not used during PCS development.
BARE is the test area from which all paint had been removed.
All archive samples have at least the X1 and BARE test areas. The archive
samples with the greatest surface area generally also have the X2 and X3 test areas.
Detailed descriptions of X1 and BARE may be found in Chapter 3 of the technical report
to the EPA/HUD field study referenced above.
A.3 Summary of Testing Protocols
All of the XRF testing protocols used for testing the archive samples have
common features. This commonality allows for describing these protocols using a
general flow chart. Figure A-1 is a general flow chart that provides a description of the
common protocol features for all XRF instruments that tested the archive samples.
A-1
-------
APPENDIX A: XRF INSTRUMENT OPERATING PROTOCOLS
Since each XRF instrument has its own unique operating characteristics, the
XRF testing protocols vary from the common protocol features to account for these
unique operating characteristics. Table A-1 describes the different operating modes
and number of readings used to perform quality control procedures, and Table A-2
provides the same information for readings taken on archive samples.
The operating mode shown in Table A-2 for the XL requires some additional
explanation since the mode indicated in this table is not specifically identified in the
manufacturer user manual. Collection of a longer measurement with this XRF,
according to the May 5,1994 version of the XL user manual, is based on the
manufacturer's manual and additional information supplied by the manufacturer.
A-2
-------
APPENDIX A: XRF INSTRUMENT OPERATING PROTOCOLS
FLOW CHART SUMMARY
DESCRIPTION DETAILS
Beginning of Day
(BOD) Instruction
Perform warm-up
and self-calibration
Complete new
INFORMATION
form
Perform BOD QC Checks (all six
substrates). Change to the
appropriate calibration before
taking any measurements.
Instructions to include: safety issues, field form use, archive
layout and testing order, sequence numbering for electronic
data capture, and voice-call back verification of data
recording by monitor.
Record all information that describes the XRF make and
model, operator, and other testing parameters to be used
during testing.
Perform all warm up and self calibration or calibration check
routines as required by the XRF manufacturer to conduct
testing for lead.
BOD (or EOD) QC checks: test a total of 6 pre-placed
blocks on a minimum of 12 inches of Styrofoam support.
Please do not move blocks.
Using a specified operating mode, take measurements on
each control block in the following order:
1. Yellow NIST
2. Red NIST
3. No NIST
Table A1 describes the variations of operating mode and
the number of readings for each XRF.
Figure A-1. General flow chart for XRF testing of archive samples.
A-3
-------
APPENDIX A: XRF INSTRUMENT OPERATING PROTOCOLS
FLOW CHART SUMMARY DESCRIPTION DETAILS
Using a specified operating mode, take measurements
on each of the following areas:
X1
BARE and RED MIST
BARE
X2
X3
Table A2 describes the variations of operating
mode and the number of readings for each XRF.
Move to first or next
location and take
readings (follow
TESTING ORDER list)
Have
'conditions bee
met that require
re-testing of a
sample
Complete reading,
disregard and re-test
Have
conditions be
met that require XRF
calibration or calibra-
tion verifica-
tion?
Perform calibration or
calibration verification
checks
IS the
testing day half
over?
Record temp &
humidity data
testing a
a continuing QC check.
Record temperature & humid
data, perform the EOD QC
checks, perform any manufactur
ers required validation checks
on the instalment as final
measurement at end
of da
IS the
testing day
within 30 minutes
of the end?
Have
15 locations
been tested since
last QC
Perform continuing
QC checks.
If XRF was powered down, allow a warmup period as
per manufacturer instructions and verify that all testing
parameters are correct.
EOD QC checks are the same as BOD QC checks;
follow same procedure as for BOD QC Checks.
Download any electronically stored data, verify
successful transfer before clearing memory.
Start next day at top of flow chart using the next
sample to be tested as the first location.
Continuing QC checks, total of 2 blocks tested on a
minimum of 12 inches of styrofoam support. Use blocks
matching same type as the last and next locations (if
same type for both, test same block twice).
If multiple calibration modes are being used for
measurements on different substrates, take additional
QC checks on other substrate blocks as needed to
ensure that all calibration modes are included.
Follow the same testing order, reading times and data
recording directives as described for BOD QC checks.
Figure A-1 (Continued). General flow chart for XRF testing of archive samples.
A-4
-------
APPENDIX A: XRF INSTRUMENT OPERATING PROTOCOLS
Table A-1. Variations in the Beginning and End of Day Quality Control Samples Used for XRF Testing of Archive
Samples.
XRF MODEL
Pb Analyzer
LPA-1
XL
LeadStar I
Microlead I
revision 4
MAP 4
XK-3
TESTING
DATES
01/18-20/95
09/14-15/95
03/12-14/95
06/13-26/95
07/17-19/95
09/11-12/95
03/15-17/95
06/13-26/95
06/06-08/95
08/14-16/95
08/17-19/96
09/04-06/96
09/13-14/95
02/05-07/96
02/19-21/96
02/08/96
OPERATOR1
PTF
NIST
NIST
NIST
PTF
NIST
NIST
NIST
NIST
PTF
PTF
PTF
NIST
NIST
PTF
NIST
OPERATING
MODE
Real time
Real time
Standard
Standard
Standard
Standard
Standard
Standard
Standard
Standard
Standard
Standard
Standard
Test
Test
Standard
NOMINAL
READING
TIME (SEC)
15
15
20
20
30
30
20
20
15
15
15
15
15
variable
variable
15
NO. OF READINGS
COLLECTED"
Y
1
1
1
0
1
1
1
0
1
1
1
1
1
1
1
1
R
3
1-3
1
3
1-3
1-3
1
3
1-3
1-3
1-3
1-3
1-3
1-3
1-3
1-3
L B
1
1
1
3
1
1
1
3
1
1
1
1
1
1
1
1
"PTF is an abbreviation for private testing firm, NIST is an abbreviation for National Institute of Standards and Technology.
"Readings on each control block: Y = Yellow NIST film SRM 2579 covering the control block.
R = Red NIST film SRM 2579 covering the control block.
B = Bare, not covered with any paint films.
A-5
-------
APPENDIX A: XRF INSTRUMENT OPERATING PROTOCOLS
Table A-2. Variations in Measurements at Test Areas of Archive Samples.
XRF
MODEL
Pb Analyzer
LPA-1
XL
LeadStar I
Microlead I
MAP 4
XK-3
TESTING
DATES
01/18-20/95
09/14-15/95
03/12-14/95
06/13-26/95"
07/17-19/95
09/11-12/95
03/15-17/95
06/13-26/95'
06/06-08/95
08/14-16/95
08/27-29/96
09/04-06/96
09/13-14/95
02/05-07/96
02/19-21/96
02/08/96
OPERATOR"
PTF
NIST
NIST
NIST
PTF
NIST
NIST
NIST
NIST
PTF
PTF
PTF
NIST
NIST
PTF
NIST
NO. OF
SAMPLES"
154"
158
154
4'
158
158
154
4f
158
158
158
158
158
158
158
158
OPERATING
MODE
Real Time
Real Time
Quick
Standard
Quick
Standard
Quick
Standard
Quick
Standard
variable9
variable9
Standard
Brief
Standard
Brief
Standard
Brief
Standard
Brief
Standard
Standard"1'
Unlimited1
Standard"'1
Unlimited1
Standard
NOMINAL
READING
TIME
(SEC)
15
15
variable
20
variable
20
variable
30
variable
30
variable
variable
15
variable
15
variable
15
variable
15
variable
15
variable
variable
variable
variable
15
NO. OF READINGS1
X1
3
1
1
1
L 1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
R
3
1
1
1
1
1
1
1
1
1
1
1
1
0
1
0
1
0
1
0
1
1
0
1
0
1
B
1
1
1
1
1
1
1
1
1
1
1
1
1
0
1
0
1
0
1
0
1
1
0
1
0
1
X2
1
0
0
1
1
1
1
1
1
1
1
1
1
0
1
0
1
0
1
0
0
1
0
1
0
0
X3
1
0
0
1
1
1
1
1
1
1
1
1
1
0
1
0
1
0
1
0
0
1
0
1
0
0
•PTF is an abbreviation for private testing firm, NIST is an abbreviation for National Institute of Standards and Technology.
"Four additional plaster samples from Denver were added to the Archives in June 1 995 for a total number of 1 58.
'Readings at sample areas: X1 = Primary XRF test area, R = Red NIST film SRM 2579 covering the bare area, B = Bare area, not
covered with any paint films, X2 = Secondary XRF test area, and X3 = Tertiary XRF test area.
"154 samples were tested twice, see section 4.5 of this document.
•Testing on these dates was performed to capture wood control block readings as well for PCS development.
The four additional plaster samples under note b above were tested.
"The testing mode varied depending on previous results as described under "How to Classify Readings" on the XL PCS. Refer to
the Archive Operations and Protocols report for additional information.
"For X1 areas, mode of testing was performed using both "Screen" and "Test" followed by "Confirm" mode if the "Test" mode
could not classify the result as either positive or negative. For R, B, X2, and X3 sample areas, only "Test" mode was used.
'A different internal calibration was used for non-aluminum metal substrates than for all other substrates.
A-6
-------
APPENDIX B: METHODOLOGICAL ISSUES IN THE ANALYSIS OF XRF
MEASUREMENTS WITH VARIABLE READING TIMES
B.1 Introduction
The statistical methodology that has been adopted for analyzing XRF
measurements in the development of Performance Characteristic Sheets (PCSs) is
based on fixed reading times. This designation was motivated by the EPA/HUD field
study, in which fixed 15-second measurements were used to evaluate the performance
of XRF instruments. Subsequent to the field study, however, several XRF instruments
have emerged that give variable reading time measurements. Instead of reading until a
fixed amount of time has elapsed, such instruments read until the conditions of a
stopping rule are met. The stopping rules are of two types: (1) to continue reading until
a fixed level of precision (SD) is achieved; (2) to continue reading until the machine can
reliably classify the lead level as above or below a certain target, such as the 1.0
mg/cm2 federal standard. In addition, the operator may terminate the reading if a
maximum time allotment has been exhausted.
The purpose of this appendix is to describe methodological issues that arise in
the analysis of variable reading time XRF measurements, and to describe analyses
related to other features of these instruments that are appropriate for inclusion in a
PCS. The following terminology will be used to refer to the variable reading time
measurement modalities considered in this appendix:
• Precision mode, in which reading continues until the instrument determines that
a target precision level has been achieved;
• Unlimited mode, in which reading continues until the instrument determines that
the lead level can be classified as above or below a target value;
• Sequential mode, in which a sequence of precision mode measurements is
made, with progressively higher levels of precision, until the lead level can be
classified as above or below a target value.
These modes may, for a particular XRF instrument, be subdivided further. For
example, one instrument designates separate precision modes corresponding to
precision levels of 0.4 mg/cm2, 0.2 mg/cm2, and 0.1 mg/cm2. An XRF instrument
typically adopts its own nomenclature for its measurement modalities, which may not
agree in all details with the terminology adopted in this appendix. Because sequential
mode is essentially a discretized version of unlimited mode, it is not treated separately
in this appendix.
B-1
-------
APPENDIX B: METHODOLOGICAL ISSUES IN THE ANALYSIS OF XRF
MEASUREMENTS WITH VARIABLE READING TIMES
In archive testing, measurements may be taken in all or only a subset of the
available precision modes at each sample. It is anticipated, however, that
measurements in at least one of the precision modes would be obtained on every
sample. If available, an unlimited mode measurement would be obtained on every
sample, as would ancillary information reported by the machine, including the reading
time. Because the reading time is an important element of performance, it would be
collected manually if the instrument does not provide this information directly, or if it
does not do so reliably.
B.2 Variable Reading Time Measurements
In precision mode, the machine continues to read until it determines that a
specified precision level has been reached. It is assumed that precision refers to the
standard deviation (SD), or to some fixed multiple of the SD. Ostensibly, the resulting
measurements have a constant precision that does not depend on the lead level, in
contrast to the observed performance of most XRF instruments that use fixed reading
times. In actuality, however, increasing the time of readings made by a machine at a
fixed location can only diminish instrumental variation, leaving non-instrumental
variation essentially untouched. Non-instrumental variation, due to characteristics of
painted building components other than the lead level that may affect the performance
of an XRF instrument, was found to be a substantial factor in the EPA/HUD field study.
This finding was confirmed with one of the XRF instruments tested at the archive in its
precision modes, for which evidence of heteroscedasticity remained on all substrates
except metal.
Unlimited mode also uses a variable reading time, but the stopping rule is
designed to detect whether or not the lead level is above an action level. With a 1.0
mg/cm2 action level, for instance, the instrument continues to test the painted surface
until the resulting measurement is outside of the range 1.0 ± p, where p is a "precision
bound" that diminishes with time. Measurements that are well above or below the
action level are likely to require shorter reading times than those that are close to the
action level.
B.3 A Model for Variable Reading Time Measurements
For fixed reading times, the XRF measurement model described in section 5.1.1
states that XRF measurements are normally distributed, with mean and variance
related in a simple way to the lead level. A fixed reading time model does not
B-2
-------
APPENDIX B: METHODOLOGICAL ISSUES IN THE ANALYSIS OF XRF
MEASUREMENTS WITH VARIABLE READING TIMES
necessarily provide an accurate description of XRF measurements obtained with
variable reading times, because the reading times may contain information about the
lead level. Readings made in real time, which depend on counts received by the
instrument, may be related to a Poisson process. Since the number of counts is
typically large, the increment in counts over a short time interval is approximately
normally distributed. This observation is the basis of a conceptual framework that is
applicable to variable reading time measurements.
Let Y(t) denote the XRF reading obtained on a given sample at time t. Although
it is likely that the instrument processes data in fixed time increments, it will be
convenient to regard t as a continuous time parameter. Suppose that a fixed,
one-second reading Y(1) is distributed as normal with mean y and variance o2. If the
reading process is unbiased, JL/ is the level of lead in the paint sample being tested. Let
W(t) denote a normal diffusion process with parameters f/v,o). Another way of stating
this is that, if f, and t2 are reading times with f, less than t2, then the difference
W(tJ - WftJ is distributed as normal with mean (t2 - tj/j, and variance (t2 - tjo2. Define
W(0) = 0. Accordingly,
constitutes a model for the XRF readings obtained in real time.
This model is consistent with the treatment of fixed reading times that was
described in chapter 5, because Y(t) is a normal random variable with mean fj and
variance o2/t if t is a fixed reading time. In other words, longer reading times diminish
the variance by the familiar rule for independent sampling. As noted above, this model
only accounts for instrumental variation, which is the only kind of precision that an XRF
instrument can discern, and therefore use, in deciding when to stop reading. The
reading times required for both precision and unlimited mode measurements are
random, because they depend on data that the instrument has received up to that point
in time.
B.3.1 Precision Mode Readings
In precision mode, testing continues until the estimated precision reaches its
target value. If the precision is estimated using a sample standard deviation, which is a
reasonable assumption, then the stopping rule is independent of the value of W(t) or
B-3
-------
APPENDIX B: METHODOLOGICAL ISSUES IN THE ANALYSIS OF XRF
MEASUREMENTS WITH VARIABLE READING TIMES
Y(t) at any time t. It then follows that the resulting XRF measurement remains an
unbiased estimate of/;. If reading times for the precision modes do not fluctuate
greatly, the resulting XRF measurements are approximately normally distributed, and
the XRF measurement model presented in section 5.1.1 may continue to be used. Bias
and precision estimates, and inconclusive ranges/thresholds, can be derived in the
same manner as in the fixed reading time case.
If the performance of an XRF instrument is affected by recognizable substrate
attributes, it is appropriate to group its testing data accordingly for model estimation and
other PCS development purposes. This applies in particular to instruments that have
special substrate calibration settings, and those for which reading times depend on the
density of the substrate, or on other substrate-related features.
B.3.2 Unlimited Mode Readings
The situation is different for unlimited mode readings, because the stopping rule
depends on the value of the measurement process. Using a 1.0 mg/cm2 action level for
illustration, it is plausible that the stopping rule is similar to the following: keep reading
until the null hypothesis that {j = 1 is rejected in a two-tailed test. The stopping time is
the smallest value of t such that the following inequality is satisfied:
| Y(t) -
f*
where s(t) is an estimate of the standard deviation of a one-second reading, based on
information available at time t. As t becomes larger, s(t) is approximately the same as
a
The resulting XRF measurement has several noteworthy properties:
it is a biased estimate of jtv;
it has a standard deviation that is smallest when the true lead level is
equal to the action level, and it increases as the lead level either
increases or decreases away from the action level;
it is not normally distributed, especially for lead levels that are close to the
action level.
B-4
-------
APPENDIX B: METHODOLOGICAL ISSUES IN THE ANALYSIS OF XRF
MEASUREMENTS WITH VARIABLE READING TIMES
These properties suggest that the XRF measurement model is not an appropriate
device for describing unlimited mode readings. In particular, the meaning of bias in this
context is difficult to formulate in a useful way, because it is designed to improve the
classification of painted surfaces relative to the action level.
A proper understanding of the statistical properties of unlimited mode
measurements resides in their diffusion process foundation. That this aspect is also
embedded in a regression problem, which in turn is embedded in an ICP measurement
error problem, makes PCS development of unlimited mode measurements a
challenging task that requires a sustained effort. As more instruments adopt this
modality of measurement, it will be worthwhile to make a concerted effort in this
direction.
B.4 Nonparametric Inconclusive Ranges for Unlimited Mode Readings
An inconclusive range is derived from estimates of the 95th percentile of XRF
readings at the 0.5 mg/cm2 lead level, and the 5th percentile of XRF readings at the 2.0
mg/cm2 lead level. When the underlying distribution is normal, these estimates are
quickly obtained from the XRF measurement model. When the underlying distributions
are unknown and suspected to be nonnormal, this approach does not work.
Nonparametric estimates of the 5th and 95th percentiles, as functions of the lead level,
can be derived under an assumption that these percentiles are increasing functions of
the lead level. These are called isotonic (or monotonic) estimates, and are derived in a
manner similar to monotone regression. The assumption needed to justify the use of
this procedure is valid for unlimited mode measurements modeled in the manner
described above.
The archive data used to derive the nonparametric percentile estimates are
those for which the laboratory-measured (ICP) lead level is less than 4.0 mg/cm2, and
excluding any outliers that are designated in precision mode analyses. All substrates
are grouped together, because sample sizes are too small to give separate estimates
for each substrate type. This is a drawback to the approach, along with the inability to
account for ICP measurement error. The nonparametric estimates are step functions,
because they are based on grouping data in a way that the sample 5th and 95th
percentiles are nondecreasing in the laboratory-measured lead level. Linear
interpolation is used to obtain estimates at the 0.5 mg/cm2 and 2.0 mg/cm2 lead levels.
If several archive tests were made, estimates are obtained for each test separately, and
pooled by averaging.
B-5
-------
APPENDIX B: METHODOLOGICAL ISSUES IN THE ANALYSIS OF XRF
MEASUREMENTS WITH VARIABLE READING TIMES
Pooling is illustrated below with hypothetical numbers:
5th percentile: first = 0.82, second = 1.54, pooled = 1.18
at 2.0 mg/cm2
95th percentile: first = 0.92, second = 1.26, pooled = 1.09
at 0.5 mg/cm2
Inconclusive range: initial threshold of 1.09, rounded to 1.1, which is converted
to an interval that covers the 1.0 action level: XL = 0.9, Xy =
1.1
B.5 Substrate Correction
The method for determining the need for substrate correction, and for applying
substrate correction where recommended, is the same for precision mode
measurements as it is for fixed reading time measurements. This is because precision
mode measurements exhibit bias in the same manner as fixed reading time
measurements, as explained in section E.3.1. Thus, the methodology described in
section 5.3 remains applicable to this mode of variable reading time measurement.
Substrate correction recommendations do not apply to unlimited mode
measurements. The development of a methodology for substrate correction that is
applicable to unlimited mode readings remains an unresolved issue. Substrate
correction that is recommended for precision mode measurements should not be used
with unlimited mode measurements.
B-6
-------
APPENDIX C: XRF PERFORMANCE CHARACTERISTIC SHEETS ERRATA
C.1 Introduction
A few errors have been identified in previously released XRF Performance
Characteristic Sheets (PCSs). The errors and corrections are described in this
appendix. The PCSs that are reproduced and provided in appendix D of this document
have been corrected of any known errors.
C.2 Errors Appearing in the "Instructions for Evaluating XRF Testing" Section
Some of the PCSs were released with errors in the "Instructions for Evaluating
XRF Testing" section of the PCS. The following wording was incorrect:
"Compute the square of each of the ten original and ten retest XRF results. Add
these squares of XRF results together. Call this quantity C."
This wording was replaced with the following:
"Compute the average of the original and re-test result for each of the ten testing
combinations. Square the average for each testing combination. Add the ten
squared averages together. Call this quantity C."
The net effect of the erroneous wording was to make it more difficult for a re-test
to fail. This reduces the frequency of spurious failures, but also makes it more difficult
to detect instances of fraud or other problems. As an example, consider the test/re-test
effect in example 2 of section 5.6.3. The use of the erroneous formula has the effect in
this example of increasing the critical value from 1.10 to 1.68, which would mean that
the re-test would pass instead of failing.
C.3 Errors Appearing in the MAP-3 PCS and Corrections
Errors were found for reported estimates for 60-second readings in the MAP-3
PCS. At 0.0 mg/cm2, the bias for plaster was incorrectly reported as -0.8. The correct
value is -0.9. At 0.5 mg/cm2, the bias for plaster was incorrectly reported as -0.6. The
correct value is -0.7. At 2.0 mg/cm2, the bias for metal was incorrectly reported as 0.9.
The correct value is 0.8. At 1.0 mg/cm2, the precision for wood was incorrectly reported
as 0.4. The correct value is 0.5. At 2.0 mg/cm2, the precision for wood was incorrectly
reported as 0.4. The correct value is 0.6. At 2.0 mg/cm2, the precision for metal was
incorrectly reported as 0.5. The correct value is 0.6.
C-1
-------
APPENDIX C (continued): XRF PERFORMANCE CHARACTERISTIC SHEETS
ERRATA
C.4 Errors Appearing in the LeadStar and MAP 4 PCSs and Corrections
The source strengths were incorrectly reported in the MAP 4 PCS. The strength
of the source installed in July 1994 was incorrectly reported as 9.4 mCi. The correct
value is 10.1 mCi. The strength of the source installed in September 1994 was
incorrectly reported as 10.6 mCi. The correct value is 11.5 mCi.
Grammatical errors were found in the LeadStar and MAP 4 PCS. The changes
involved spelling corrections and additional wording for clarification and improved flow.
The LeadStar and MAP 4 PCSs in appendix D have been corrected for these errors.
C.5 Errors Appearing in the LPA-1 PCS and Corrections
Two errors were found in the LPA-1 PCS. On page 9 of 12, at 0.0 mg/cm2, the
bias for metal was incorrectly reported as -0.4. The correct value is -0.5. On page 10
of 12, at 2.0 mg/cm2, the lower bound of the bias range for metal was incorrectly
reported as -0.8. The correct value is -0.7.
C.6 Errors Appearing in the Microlead I PCS and Corrections
Four estimates were omitted from page 5 of 5 in the Microlead I PCS. At 0.5
mg/cm2, the bias for brick was omitted. The correct value is 0.4. At 2.0 mg/cm2, the
bias for brick was omitted. The correct value is 0.7. At 0.5 mg/cm2, the precision for
brick was omitted. The correct value is 0.5. At 2.0 mg/cm2, the precision for brick was
omitted. The correct value is 0.5.
C.7 Error Appearing in the Title of the PCSs and Correction
PCSs were incorrectly titled with the plural word "Characteristics". The correct
title uses a singular "Characteristic". Thus, the correct title for all Edition 1 versions is
"XRF Performance Characteristic Sheet".
C-2
-------
APPENDIX D: XRF PERFORMANCE CHARACTERISTIC SHEETS AND
RELATED RESULTS
D.1 Introduction
The XRF Performance Characteristic Sheets (PCSs) that have been released as
of August 31, 1997, with all corrections made as necessary, are provided in this
appendix. See Appendix C for a summary of the corrections to the PCSs.
Following each PCS are associated results such as inconclusive ranges and
thresholds for cases where substrate correction is recommended but not performed,
bias and precision estimates, the standard errors of these estimates, and classification
results based on empirical data. In general, following each PCS will be the following
tables: tables for XRF inconclusive ranges and thresholds for cases where substrate
correction is recommended but not performed, tables of bias and their standard errors,
tables of precision and their standard errors, and tables of classification rates.
An important note about the tables presenting results for cases where substrate
correction is recommended but not performed follows. IT SHOULD BE EMPHASIZED
THAT THESE RESULTS ARE PRESENTED FOR ILLUSTRATIVE PURPOSES
ONLY, AND SHOULD NOT BE USED AS SUBSTITUTES FOR PERFORMING
SUBSTRATE CORRECTION. If substrate correction is recommended, testing of the
XRF instrument has shown a reduction in bias through the use of substrate correction.
Moreover, substrate correction tends to diminish the variation between machines, which
in turn broadens the applicability of inconclusive ranges and thresholds for substrate
corrected readings beyond the machines that were tested. A similar property does not
hold for inconclusive ranges and thresholds developed with non-corrected readings
when a benefit from performing substrate correction was observed.
D-1
-------
APPENDIX D: XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
D.2 XRF Performance Characteristic Sheet for the Advanced Detectors
LeadStar and Related Results
D-2
-------
APPENDIX D: XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
Advanced Detectors; LeadStar
EFFECTIVE DATE: October?, 1996 EDITION NO.. 1
MANUFACTURER AND MODEL:
Make: Advanced Detectors, Inc.
Model: LeadStar
Source: Co"
Note: This sheet supersedes all previous sheets for the XRF instrument of
the make, model, and source shown above.
EVALUATION DATA SOURCE AND DATE:
This sheet is supplemental information to be used in conjunction with Chapter 7 of the HUD
Guidelines for the Evaluation and Control of Lead-Based Paint Hazards in Housing ("HUD Guidelines").
Performance parameters shown on this sheet are calculated from the EPA/HUD evaluation using archived
building components. Testing was conducted on approximately 150 test locations. All of the test locations
were tested twice in 1995, in June and in August. Two instruments (both installed with software version 4.05)
were used in the June 1995 testing and one instrument (with software version 4.08) was used in the August
1995 testing. All three instruments had a June 1995 source at 15 mCi initial strength. The same test
locations were tested twice again in 1996, in August and in September. One instrument (with software
version 4.30) was used in the August 1996 testing, with a July 1996 source at 15 mCi initial strength. One
instrument (also with software version 4.30) was used in the September 1996 testing, with an August 1996
source at 15 mCi initial strength. Each instrument had a distinct serial number.
FIELD OPERATION GUIDANCE
OPERATING PARAMETERS:
Performance parameters shown in this sheet are applicable only when operating the instrument
under the same conditions as the evaluation testing and using the procedures described in Chapter 7 of the
HUD Guidelines. Operating parameters include:
Manufacturer-recommended warm-up and quality control procedures
• Use the Multifamily Decision Flowchart for determining the presence of lead on a
component type in multifamily housing
Take readings on three locations per component for single-family housing and one
location per component for multifamily housing
Calibration checks are taken using the red (1.02 mg/cm2) NIST Standard Reference
Material (SRM No. 2579) paint film
Readings for determining the substrate correction values are taken on bare substrate
covered with red (1.02 mg/cm2) NIST SRM paint film
Lead-based paint is defined as paint with lead equal to or in excess of 1.0 mg/cm2.
1of8
Figure D-1. XRF Performance Characteristic Sheet for the Advanced Detectors
LeadStar.
D-3
-------
APPENDIX D: XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
Advanced Detectors; LeadStar
XRF CALIBRATION CHECK:
Chapter 7 of the HUD Guidelines recommends using a calibration check procedure to determine
the operating condition of the XRF instrument. For this instrument, calibration check readings should be
taken in Fixed Mode. If the observed calibration check average minus 1.02 mg/cm2 is greater than the
positive (plus) calibration check tolerance value, or less than the negative (minus) calibration check tolerance
value, then the instructions provided by the manufacturer should be followed in order to bring the instrument
back into control before any more XRF testing is done. This calibration check is estimated to produce an
incorrect result (that is, a finding that the instrument is out of calibration) very infrequently - once out of every
200 times this procedure is followed.
Use the following calibration check tolerance values for Fixed Mode readings for those instruments
with software versions 4.1 to 4.30. (This guidance may be used for software versions higher than 4.30 if the
higher software version incorporates the same signal processing and data treatment algorithms that are in
software version 4.30).
minus value = -0.2 mg/cm2
plus value = +0.0 mg/cm2
Use the following calibration check tolerance values for Fixed Mode readings for those instruments
with software versions earlier than version 4.1.
minus value = -0.2 mg/cm2
plus value = +0.1 mg/cm2
(Operators may choose to use limits in the manufacturer's operations manual for this calibration
check. The rate of an incorrect result if the limits in the manufacturer's operations manual are followed may
be different from the rate of an incorrect result stated here.)
FOR XRF RESULTS BELOW 4.0 mg/cm2, SUBSTRATE CORRECTION RECOMMENDED FOR:
For those instruments with software versions 4.1 to 4.30 (This guidance may be used for software
versions higher than 4.30 if the higher software version incorporates the same signal processing
and data treatment algorithms that are in software version 4.30).
none
For those instruments with software versions earlier than version 4.1
Metal
SUBSTRATE CORRECTION NOT RECOMMENDED FOR:
For those instruments with software versions 4.1 to 4.30.
Brick, Concrete, Drywall, Metal, Plaster, and Wood
For those instruments with software versions earlier than version 4.1.
Brick, Concrete, Drywall, Plaster, and Wood
2 of 8
Figure D-1 continued. XRF Performance Characteristic Sheet for the Advanced
Detectors LeadStar.
D-4
-------
APPENDIX D: XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
Advanced Detectors; LeadStar
SUBSTRATE CORRECTION VALUE COMPUTATION:
Chapter 7 of the HUD Guidelines provides guidance on correcting XRF results for substrate bias.
Supplemental guidance for using the red (1.02 mg/cm2) NIST SRM paint film for substrate correction is
provided below.
XRF results are corrected for substrate bias by subtracting from each XRF result a correction value
determined separately in each house for single-family housing or in each development for multifamily
housing, for each substrate. The correction value is an average of XRF readings taken over red NIST SRM
(1.02 mg/cm2) paint films at test locations that had been scraped clean of their paint covering. Compute the
correction values as follows:
Using the same XRF instrument, take three readings on a bare substrate area covered with the red
NIST SRM (1.02 mg/cm2) paint film. Repeat this procedure by taking three more readings on a
second bare substrate area of the same substrate covered with the red NIST SRM (1.02 mg/cm2)
paint film.
Compute the correction value for each substrate type by computing the average of all six readings
as shown below.
For each substrate type recommended for substrate correction:
Correction
Value
1 ,- + 2nd + 3'" + 4th * 5th + 6'" Reading . ., . 2
> = - - 1.02mg/cm2
J 6
INCONCLUSIVE RANGE OR THRESHOLD:
XRF results are classified using either the threshold or the inconclusive range. In single-family
housing, an XRF result is the average of three readings taken on a testing combination. (A testing
combination is a location on a painted surface as defined in Chapter 7 of the HUD Guidelines). In multifamily
housing, an XRF result is a single reading taken on a testing combination. For computing the XRF result,
use all digits that are reported by the instrument. For the threshold, results are classified as positive if they
are greater than or equal to the threshold, and negative if they are less than the threshold. There is no
inconclusive classification when using the threshold. For the inconclusive range, results are classified as
positive if they are greater than or equal to the upper limit of the inconclusive range, and negative if they are
less than or equal to the lower limit of the inconclusive range. Thresholds and inconclusive ranges reported
here were determined for comparing results to the 1.0 mg/cm2 standard. For a listing of laboratories
recommended by the EPA National Lead Laboratory Accreditation Program (NLLAP) for the analysis of
samples to resolve an inconclusive XRF result or additional confirmational analysis, call the National Lead
Information Center Clearinghouse at 1-800-424-LEAD.
For those instruments with software versions 4.1 to 4.30. (This guidance may be used for software
versions higher than 4.30 if the higher software version incorporates the same signal processing and data
treatment algorithms that are in software version 4.30).
3 of 8
Figure D-1 continued. XRF Performance Characteristic Sheet for the Advanced
Detectors LeadStar.
D-5
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
Advanced Detectors; LeadStar
1 5-SECOND FIXED MODE
READING DESCRIPTION
Results not corrected for substrate bias
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
THRESHOLD
(mg/cm2)
None
None
None
None
1.0
None
INCONCLUSIVE
RANGE
(mg/cm2)
0.9 to 1.1
0.9 to 1.1
0.9 to 1.1
0.9 to 1.2
None
0.9 to 1.1
BRIEF MODE
READING DESCRIPTION
Results not corrected for substrate bias
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
TmSg/Hcm'D
1.0
1.0
1.0
1.0
1.0
1.0
For those instruments with software versions earlier than version 4.1
1 5-SECOND FIXED MODE
READING DESCRIPTION
Results corrected for substrate bias for
readings on metal substrates only
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
INCONCLUSIVE RANGE
in mg/cm2
0.9 to 1.3
0.9 to 1.3
0.9 to 1.1
0.9 to 1.1
0.9 to 1.1
0.9 to 1.1
INSTRUCTIONS FOR EVALUATING XRF TESTING:
Chapter 7 of the HUD Guidelines recommends several options for evaluating XRF testing. Among
those options is the following procedure which may be used after XRF testing has been completed. In
single-family housing, an XRF result is the average of three readings taken on a testing combination. (A
testing combination is a location on a painted surface as defined in Chapter 7 of the HUD Guidelines). In
multifamily housing, an XRF result is a single reading taken on a testing combination. If a multifamily
housing development is being retested, randomly select two units from within the development from which
the ten testing combinations should be randomly selected.
Randomly select ten testing combinations for retesting from each house or from the two selected units.
Conduct XRF retesting at the ten testing combinations selected for retesting.
4 of 8
Figure D-1 continued.
XRF Performance Characteristic Sheet for the Advanced
Detectors LeadStar.
D-6
-------
APPENDIX D: XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
Advanced Detectors; LeadStar
Determine if the XRF testing in the units or house passed or failed the test by applying the steps below.
Compute the Retest Tolerance Limit by the following steps:
Determine XRF results for the original and retest XRF readings. Do not
correct the original or retest results for substrate bias. In single-family housing
a result is defined as the average of three readings. In multifamily housing, a
result is a single reading. Therefore, there will be ten original and ten retest
XRF results for each house or for the two selected units.
Compute the average of the original and re-test result for each of the ten testing combinations.
Square the average for each testing combination.
Add the ten squared averages together. Call this quantity C.
Multiply the number C by 0.0072. Call this quantity D.
Add the number 0.032 to D. Call this quantity E.
Take the square root of E. Call this quantity F.
Multiply F by 1.645. The result is the Retest Tolerance Limit.
Compute the overall average of all ten retest XRF results over all ten testing combination
selected for retesting.
Take the difference of the overall average of the ten original XRF results and the overall
average of the ten retest XRF results. If the difference is negative, drop the negative
sign
If the difference of the overall averages is less than the Retest Tolerance Limit, the
inspection has passed the retest. If the difference of the overall averages equals or
exceeds the Retest Tolerance Limit, this procedure should be repeated with ten new
testing combinations. If the difference of the overall averages is equal to or greater than
the Retest Tolerance Limit a second time, then the inspection should be considered
deficient.
Use of this procedure is estimated to produce a spurious result approximately 1 % of the time. That is,
results of this procedure will call for further examination when no examination is warranted in approximately 1
out of 100 dwelling units tested.
TESTING TIMES:
For Fixed Mode, the LeadStar instrument tests for a set length of time before a result is obtained
and displayed. For Brief Mode, the LeadStar instrument tests until a reading is obtained relative to an
operator set Action Level. The following table provides a summary of testing times for readings taken in Brief
Mode with an Action Level set to 1.0 mg/cm2. All times have been scaled relative to an initial source strength
of 15 mCi. Note that source strength and factors such as substrate may affect testing times.
5pf8
Figure D-1 continued. XRF Performance Characteristic Sheet for the Advanced
Detectors LeadStar.
D-7
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
Advanced Detectors; LeadStar
Results from testing in August 1996 and September 1996
BRIEF MODE TESTING TIMES' (Seconds)
SUBSTRATE
Wood
Drywall
Metal
Brick
Concrete
Plaster
ALL DATA
25th
Percentile
7
7
8
Median
7
7
8
75th
Percentile
8
8
9
MEDIAN FOR LABORATORY-M
LEAD LEVELS (mg/cm2
Pb < 0.25
7
7
8
0.25 s Pb < 1.0
8
8
8
EASURED
1 .0 < Pb
7
7
8
"Testing times are based on readings obtained relative to a 1 .0 mg/cm2 Action Level.
BIAS AND PRECISION:
Do not use these bias and precision data to correct for substrate bias. These bias and precision
data were computed without substrate correction from samples with reported laboratory results less than 4.0
mg/cm2 lead. There were 15 test locations taken in Fixed Mode with a laboratory reported result equal to or
greater than 4.0 mg/cm2 lead. The fifteen test locations were each tested four times in Fixed Mode, once
under software version 4.05, once under software version 4.08, and twice under software version 4 30 Of
the 15 test locations tested under software version 4.05, one case resulted in an XRF reading less than
1.0 mg/cm2. Of the 45 test locations tested under software versions 4.08 and 4.30, there were no instances
in which an XRF reading was less than 1.0 mg/cm2. Each of the fifteen test locations was tested in Brief
Mode twice, both under software version 4.30 Out of the 30 Brief Mode testing cases, there were no
instances in which an XRF reading was less than 1.0 mg/cm2. The following data are for illustrative purposes
only. Actual bias must be determined on-site. Inconclusive ranges provided above already account for bias
and precision. Units are in mg/cm2.
6 of 8
Figure D-1 continued.
XRF Performance Characteristic Sheet for the Advanced
Detectors LeadStar.
D-8
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
Advanced Detectors; LeadStar
For those instruments with software versions 4.1 to 4.30.
FIXED MODE READINGS
MEASURED AT
0.0 mg/cm2
0.5 mg/cm2
1 .0 mg/cm2
2.0 mg/cm2
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
BIAS
(mg/cm2)
0.0
0.0
0.0
0.1
-0.1
0.0
0.1
0.1
0.0
0.1
0.0
0.1
0.1
0.1
0.1
0.2
0.0
0.1
0.2
0.2
0.2
0.3
0.1
0.2
PRECISION*
(mg/cm2)
0.1
0.1
0.1
0.1
0.1
0.1
0.2
0.2
0.2
0.2
0.2
0.2
0.3
0.3
0.3
0.3
0.3
0.3
0.4
0.4
0.4
0.4
0.4
0.4
'Precision at 1 standard deviation
7 of 8
Figure D-1 continued.
XRF Performance Characteristic Sheet for the Advanced
Detectors LeadStar.
D-9
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
Advanced Detectors; LeadStar
For those instruments with software versions earlier than version 4.1.
FIXED MODE READINGS
MEASURED AT
0.0 mg/cm2
0.5 mg/cm2
1 .0 mg/cm2
2.0 mg/cm2
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
BIAS
(mg/cm2)
0.1
0.1
0.0
0.1
0.0
0.0
0.2
0.2
0.1
0.2
0.1
0.1
0.3
0.3
0.1
0.2
0.1
0.1
0.4
0.4
0.3
0.4
0.3
0.3
PRECISION*
(mg/cm2)
0.1
0.1
0.1
0.1
0.1
0.1
0.2
0.2
0.2
0.2
0.2
0.2
0.3
0.3
0.3
0.3
0.3
0.3
0.5
0.5
0.5
0.5
0.5
0.5
"Precision at 1 standard deviation
A document titled Methodology for XRF Performance Characteristic Sheets provides an
explanation of the statistical methodology used to construct the data in the sheets, and provides empirical
results from using the recommended inconclusive ranges or thresholds for specific XRF instruments. For a
:opy of this document call the National Lead Information Center Clearinghouse at 1 -800-424-LEAD
This XRF Performance Characterist c Sheet is a joint product of the U S. Environmental Protection Agency (EPA) and the U.S.
Department of Housing and Urban Development (HUD). The issuance of this sheet does not constitute rulemakmg The
information provided here is intended solely as guidance to be used in conjunction with Chapter 7 of the Guidelines for the
Evaluation and Control of Lead-Based Paint Hazards in Housing EPA and HUD reserve the right to revise this guidance
Please address questions and comments on this sheet to Director, Office of Lead-Based Paint Abatement and Poisoning
Prevention, U S Department of Housing and Urban Development, Room B-133, 451 Seventh St, S W , Washington, DC 20410
8 Of 8
Figure D-1 continued.
XRF Performance Characteristic Sheet for the Advanced
Detectors LeadStar.
D-10
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
Table D-1. Bias Estimates, and Their Standard Errors, of Advanced Detectors LeadStar 15-Second
Fixed Mode Readings For Those Instruments With Software Versions 4.1 to 4.30.
15-SECOND FIXED MODE
READING MEASURED AT
0.0 mg/cm2
0.5 mg/cm2
1 .0 mg/cm2
2.0 mg/cm2
CORRECTED?
No
No
No
No
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
BIAS (mg/cm2)
0.0
0.0
0.0
01
-0.1
0.0
0.1
0.1
0.0
0 1
0.0
0.1
01
0.1
01
0.2
00
0.1
0.2
02
0.2
0.3
0.1
0.2
STANDARD
ERROR
0.02
0.02
002
002
0.02
0.02
0.03
0.03
0.03
0.03
0.03
0.03
005
0.05
0.05
0.05
005
0.05
009
009
0.09
009
009
0.09
D-11
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
Table D-2. Precision Estimates, and Their Standard Errors, of Advanced Detectors LeadStar
15-Second Fixed Mode Readings For Those Instruments With Software Versions 4.1 to
4.30.
15-SECOND FIXED MODE
READING MEASURED AT
0 0 mg/cm2
0.5 mg/cm2
1.0 mg/cm2
2.0 mg/cm2
CORRECTED?
No
No
No
No
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
PRECISION
(mg/cm2)
0.1
0.1
0.1
0.1
0.1
0.1
0.2
0.2
0.2
0.2
0.2
0.2
0.3
0.3
0.3
0.3
0.3
0.3
0.4
0.4
0.4
0.4
0.4
0.4
STANDARD
ERROR
0.01
0.01
0.01
0.01
0.01
0.01
0.02
0.02
0.02
0.02
0.02
0.02
0.03
0.03
0.03
0.03
0.03
0.03
0.05
005
0.05
0.05
0.05
0.05
D-12
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
Table D-3. Inconclusive Range For Advanced Detectors LeadStar 15-Second Fixed Mode Readings
For Results Where Substrate Correction Is Not Performed, But Substrate Correction is
Recommended For Those Instruments With Software Versions Earlier Than Version 4.1.
15-SECOND FIXED MODE READING DESCRIPTION
Readings not corrected for substrate bias
SUBSTRATE
Metal
INCONCLUSIVE
RANGE (mg/cm2)
0.9 to 1.2
Table D-4. Bias Estimates, and Their Standard Errors, of Advanced Detectors LeadStar 15-Second
Fixed Mode Readings For Those Instruments With Software Versions Earlier Than
Version 4.1.
15-SECOND
FIXED MODE READING
MEASURED AT
0 0 mg/cm2
0.5 mg/cm2
1 0 mg/cm2
2.0 mg/cm2
CORRECTED?
Mn
Yes
No
Yes
No
Yes
No
Yes
SUBSTRATE
Brick
Concrete
Dry wall
Metal
Plaster
Wood
Metal
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
BIAS (mg/cm2)
01
01
0.0
0.1
0.0
0.0
-0.1
0.2
0.2
0.1
0.2
0.1
0.1
0.0
0.3
03
0.1
0.2
0.1
0.1
0.1
0.4
0.4
0.3
0.4
0.3
0.3
0.2
STANDARD
ERROR
0.06
0.06
0.02
003
0.02
0.02
0.04
0.06
0.06
0.03
003
0.03
0.03
0.04
007
0.07
0.05
006
0.05
0.05
0.06
0.11
0.11
0.11
011
0.11
0.11
0.11
D-13
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
Table D-5. Precision Estimates, and Their Standard Errors, of Advanced Detectors LeadStar
15-Second Fixed Mode Readings For Those Instruments With Software Versions Earlier
Than Version 4.1.
15-SECOND
FIXED MODE READING
MEASURED AT
0.0 mg/cm2
0.5 mg/cm2
1 .0 mg/cm2
2.0 mg/cm2
CORRECTED?
No
Yes
No
Yes
No
Yes
No
Yes
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
PRECISION
(mg/cm2)
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.2
0.2
0.2
0.2
0.2
0.2
0.3
0.3
0.3
0.3
0.3
0.3
0.3
0.4
0.5
0.5
0.5
0.5
0.5
0.5
0.5
STANDARD
ERROR
0.01
0.01
0.01
0.01
0.01
0.01
0.02
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.04
0.04
004
0.04
0.04
0.04
0.05
005
0.05
0.05
0.05
0.05
0.05
0.07
D-14
-------
APPENDIX D: XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
Table D-6. Classification Results For Advanced Detectors LeadStar Brief Mode (Uncorrected),
Classified Using Threshold Values Reported in the XRF Performance Characteristic
Sheet For Instruments With Software Versions 4.1 to 4.30.
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
TOTAL
THRESHOLD
1.0
1.0
1.0
1.0
1.0
1.0
FALSE POSITIVE
RATE
0.0% (0/2)
0.0% (0/4)
0.0% (0/28)
5.6% (3/54)
3.4% (2/58)
7.5% (6/80)
4.9% (11/226)
FALSE NEGATIVE
RATE
0.0% (0/4)
(0/0)
(0/0)
0 0% (0/22)
0.0% (0/1 8)
6.5% (3/46)
3.3% (3/90)
Table D-7. Classification Results For Advanced Detectors LeadStar 15-Second Fixed Mode
Readings (Uncorrected), Classified Using Inconclusive Ranges or Threshold Values
Reported in the XRF Performance Characteristic Sheet For Instruments With Software
Versions 4.1 to 4.30.
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
TOTAL
INCONCLUSIVE
RANGE OR
THRESHOLD
0.9 to 1.1
0.9 to 1.1
0.9 to 1.1
0.9 to 1.2
1.0
0.9 to 1.1
FALSE POSITIVE
RATE
0.0% (0/2)
0.0% (0/4)
0 0% (0/28)
3.7% (2/54)
3.4% (2/58)
6.3% (5/80)
4.0% (9/226)
FALSE NEGATIVE
RATE
0.0% (0/4)
(0/0)
(0/0)
0.0% (0/22)
0.0% (0/1 8)
0.0% (0/46)
0.0% (0/90)
INCONCLUSIVE
RATE
0.0% (0/6)
0.0% (0/4)
0 0% (0/28)
9.2% (7/76)
0.0% (0/76)
40% (5/1 26)
3.8% (12/31 6)
D-15
-------
APPENDIX D: XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
Table D-8. Classification Results For Advanced Detectors LeadStar 15-Second Fixed Mode
Readings (Metal Substrates Corrected and Uncorrected), Classified Using Inconclusive
Ranges Reported in the XRF Performance Characteristic Sheet For Instruments With
Software Versions Earlier Than Version 4.1.
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal (Uncorrected)
TOTAL'
INCONCLUSIVE
RANGE
0.9 to 1.3
0.9 to 1.3
0.9 to 1.1
0.9 to 1.1
0.9 to 1.1
0.9 to 1 1
0.9 to 1.2
FALSE POSITIVE
RATE
0.0% (0/2)
0.0% (0/4)
0.0% (0/28)
0.0% (0/54)
3.4% (2/58)
7.7% (6/78)
0.0% (0/54)
3.5% (8/226)
FALSE NEGATIVE
RATE
0.0% (0/4)
(0/0)
(0/0)
0.0% (0/22)
5.6% (1/1 8)
0.0% (0/46)
0.0% (0/22)
1.1% (1/90)
INCONCLUSIVE
RATE
0.0% (0/6)
0.0% (0/4)
0.0% (0/28)
3.9% (3/76)
0.0% (0/76)
5 6% (7/1 26)
10. 5% (8/76)
3.2% (10/31 6)
'Total results are for values reported in PCS.
D-16
-------
APPENDIX D: XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
D.3 XRF Performance Characteristic Sheet for the Radiation Monitoring Device
LPA-1 and Related Results
D-17
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
EFFECTIVE DATE:
XRF PERFORMANCE CHARACTERISTIC SHEET
Radiation Monitoring Devices; LPA-1
November 27, 1995
EDITION NO.: 1
MANUFACTURER AND MODEL:
Make' Radiation Monitoring Devices
Model: LPA-1
Source: Co"
Note: This sheet supersedes all previous sheets for the XRF instrument of the make, model, and
source shown above.
EVALUATION DATA SOURCE AND DATE:
This sheet is supplemental information to be used in conjunction with Chapter 7 of the HUD Guidelines
for the Evaluation and Control of Lead-Based Paint Hazards in Housing ("HUD Guidelines"). Performance
parameters shown on this sheet are calculated from the EPA/HUD evaluation using archived building
components. Testing was conducted on approximately 150 test locations. All of the test locations were tested
three times, once in March 1995, once in July 1995, and once in September 1995 using three distinct
instruments. The instrument that performed testing in March had a new source installed in January 1995 with
12 mCi initial strength. The instrument that performed testing in July had a new source installed in June 1995
with 12 mCi initial strength. The instrument that performed testing in September had a new source installed
in February 1995 with 12 mCi initial strength. LPA-1 instruments that were purchased before June 26, 1995
and have not been serviced since June 26, 1995 have a different version of firmware than those instruments
sold or serviced after June 26, 1995. Therefore, this sheet distinguishes between instruments sold prior to
June 26, 1995 and have not been serviced since June 26, 1995 from those instruments serviced or sold after
this date.
FIELD OPERATION GUIDANCE
OPERATING PARAMETERS:
Performance parameters shown in this sheet are applicable only when operating the instrument under
the same conditions as the evaluation testing and using the procedures described in Chapter 7 of the HUD
Guidelines. Operating parameters include:
• Manufacturer-recommended warm-up and quality control procedures
• Use the Multifamily Decision Flowchart for determining the presence of lead on a component type
in multifamily housing
• Quick mode, nominal 20-second standard mode, or nominal 30-second standard mode readings on
three locations per component for single-family housing and one location per component for
multifamily housing
• The nominal reading time for standard mode readings must be adjusted to account for source decay
• Calibration checks are taken using nominal 30-second standard mode readings and the red (1.02
mg/cm2) NIST Standard Reference Material (SRM No. 2579) paint film
1 of 12
Figure D-2. XRF Performance Characteristic Sheet for the Radiation Monitoring
Devices LPA-1.
D-18
-------
APPENDIX D: XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
Radiation Monitoring Devices; LPA-1
• Readings for determining the substrate correction values are taken on bare substrate covered with
red (1.02 mg/cm2) NIST SRM paint film
• Lead-based paint is defined as paint with lead equal to or in excess of 1.0 mg/cm2.
XRF CALIBRATION CHECK:
Chapter 7 of the HUD Guidelines recommends using a calibration check procedure to determine the
operating condition of the XRF instrument. For this instrument, calibration check readings should be taken
with nominal 30-second standard mode readings regardless of the date of purchase or servicing. If the
observed calibration check average minus 1.02 mg/cm2 is greater than the positive (plus) calibration check
tolerance value, or less than the negative (minus) calibration check tolerance value, then the instructions
provided by the manufacturer should be followed in order to bring the instrument back into control before any
more XRF testing is done. This calibration check is estimated to produce an incorrect result (that is, a finding
that the instrument is out of calibration) very infrequently - once out of every 200 times this procedure is
followed.
For those instruments sold prior to June 26. 1995 and have not been serviced since June 26. 1995 use
the following calibration check tolerance values:
minus value = -0.3 mg/cm2
plus value = +0.1 mg/cm2
For those instruments sold or serviced after June 26. 1995 use the following calibration check tolerance
values:
minus value = -0.3 mg/cm2
plus value = +0.3 mg/cm2
XRF RESULTS BELOW 4.0 mg/cm2, SUBSTRATE CORRECTION RECOMMENDED FOR:
For those instruments sold prior to June 26. 1995 and have not been serviced since June 26. 1995:
• Metal and wood using quick mode or either 20-second or 30-second standard mode readings
For those instruments sold or serviced after June 26.1995:
Metal using 30-second standard mode readings
• None using quick mode readings
SUBSTRATE CORRECTION NOT RECOMMENDED FOR:
For those instruments sold prior to June 26.1995 and have not been serviced since June 26. 1995:
Brick, Concrete, Drywall, and Plaster using quick mode or either 20-second or 30-second standard
mode readings
2 of 12
Figure D-2 continued. XRF Performance Characteristic Sheet for the Radiation
Monitoring Devices LPA-1.
D-19
-------
APPENDIX D: XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
Radiation Monitoring Devices; LPA-1
For those instruments sold or serviced after June 26. 1995:
Brick, Concrete, Drywall, Plaster, and Wood using 30-second standard mode readings
Brick, Concrete, Drywall, Metal, Plaster, and Wood using quick mode readings
SUBSTRATE CORRECTION VALUE COMPUTATION:
Chapter 7 of the HUD Guidelines provides guidance on correcting XRF results for substrate bias.
Supplemental guidance for using the red (1.02 mg/cm2) NISI SRM paint film for substrate correction is
provided below.
XRF results are corrected for substrate bias by subtracting from each XRF result a correction value
determined separately in each house for single-family housing or in each development for multifamily housing,
for each substrate. The correction value is an average of XRF readings taken over red NIST SRM (1.02
mg/cm2) paint films at test locations that had been scraped clean of their paint covering. Compute the
correction values as follows:
Using the same XRF instrument, take three readings on a bare substrate area covered with the red NIST
SRM (1.02 mg/cm2) paint film. Repeat this procedure by taking three more readings on a second bare
substrate area of the same substrate.
Compute the correction value for each substrate type by computing the average of all six readings as
shown below.
For each substrate type:
Correction
erection 1 = 1" + 2"" + 3rd + 4th + 5'" + 6'" Reading 1 Q2mg/cm;
Value J e
Repeat this procedure for each substrate tested in the house or housing development as needed.
CLASSIFICATION OF RESULTS USING THRESHOLD VALUES:
XRF results are classified using either the threshold or the inconclusive range. In single-family housing,
an XRF result is the average of three readings taken on a testing combination. (A testing combination is a
location on a painted surface as defined in Chapter 7 of the HUD Guidelines). In multifamily housing, an XRF
result is a single reading taken on a testing combination. For computing the XRF result, use all digits that are
reported by the instrument. For the threshold, results are classified as positive if they are greater than or
equal to the threshold, and negative if they are less than the threshold. There is no inconclusive classification
when using the threshold. For the inconclusive range, results are classified as positive if they are greater than
or equal to the upper limit of the inconclusive range, and negative if they are less than or equal to the lower
limit of the inconclusive range. Thresholds and inconclusive ranges were determined for comparing results
to the 1.0 mg/cm2 standard. For a listing of laboratories recommended by the EPA National Lead Laboratory
Accreditation Program (NLLAP) for the analysis of samples to resolve an inconclusive XRF result or additional
confirmational analysis, call the National Lead Information Center Clearinghouse at 1-800-424-LEAD.
3 of 12
Figure D-2 continued. XRF Performance Characteristic Sheet for the Radiation
Monitoring Devices LPA-1.
D-20
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
Radiation Monitoring Devices; LPA-1
For those instruments sold oriorto June 26. 1995 and have not been serviced since June 26. 1995
30-SECOND STANDARD MODE
READING DESCRIPTION
Results corrected for substrate bias on metal and wood
substrates only
SUBSTRATE
Brick
Concrete
Dry wall
Metal
Plaster
Wood
THRESHOLD
(mg/cm2)
0.8
0.8
0.7
0.8
0.8
09
For those instruments sold oriorto June 26. 1995 and have not been serviced since June 26. 1995.
20-SECOND STANDARD MODE
READING DESCRIPTION
Results corrected for substrate bias on metal and wood
substrates only
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
THRESHOLD
(mg/cm2)
0.7
07
07
09
0.8
0.8
For those instruments sold prior to June 26. 1995 and have not been serviced since June 26. 1995'
QUICK MODE CMBCTRATP INCONCLUSIVE
READING DESCRIPTION S>UB& ' "* ' t RANGE (mg/cm2)
Brick 07-08
Concrete 07-08
Results corrected for substrate bias on metal and Dry wall 06-08
wood substrates only Metal 09-10
Plaster 07-08
Wood 0.7 - 0.8
4 of 12
Figure D-2 continued.
XRF Performance Characteristic Sheet for the Radiation
Monitoring Devices LPA-1.
D-21
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
Radiation Monitoring Devices; LPA-1
For those instruments sold or serviced after June 26.1995:
30-SECOND STANDARD MODE
READING DESCRIPTION
Results corrected for substrate bias on metal substrate
only
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
THRESHOLD
(mg/cm2)
1.0
1.0
1.0
0.9
1.0
1.0
For those instruments sold or serviced after June 26. 1995:
QUICK MODE
READING DESCRIPTION
Readings not corrected for substrate
bias on any substrate
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
THRESHOLD
(mg/cm2)
1.0
1.0
1.0
None
None
1.0
INCONCLUSIVE
RANGE
(mg/cm2)
None
None
None
0.9 to 1.3
0.9 to 1.0
None
INSTRUCTIONS FOR EVALUATING XRF TESTING:
Chapter 7 of the HUD Guidelines recommends several options for evaluating XRF testing Among those
options is the following procedure which may be used after XRF testing has been completed. In single-family
housing, an XRF result is the average of three readings taken on a testing combination. (A testing combination
is a location on a painted surface as defined in Chapter 7 of the HUD Guidelines). In multifamily housing, an
XRF result is a single reading taken on a testing combination. If a multifamily housing development is being
5 of 12
Figure D-2 continued.
XRF Performance Characteristic Sheet for the Radiation
Monitoring Devices LPA-1.
D-22
-------
APPENDIX D: XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
Radiation Monitoring Devices; LPA-1
retested, randomly select two units from within the development from which the ten testing combinations
should be randomly selected.
Randomly select ten testing combinations for retesting from each house or from the two selected units.
Conduct XRF retesting at the ten testing combinations selected for retesting
Determine if the XRF testing in the units or house passed or failed the test by applying the steps below
Compute the Retest Tolerance Limit by the following steps:
Determine XRF results for the original and retest XRF readings. Do not correct the original or
retest results for substrate bias. In single-family housing a result is defined as the average of
three readings. In multifamily housing, a result is a single reading. Therefore, there will be ten
original and ten retest XRF results for each house or for the two selected units.
Compute the average of the original and re-test result for each of the ten testing combinations.
Square the average for each testing combination.
Add the ten squared averages together. Call this quantity C.
Multiply the number C by 0.0072. Call this quantity D.
Add the number 0.032 to D. Call this quantity E.
Take the square root of E. Call this quantity F.
Multiply F by 1.645. The result is the Retest Tolerance Limit.
Compute the overall average of all ten retest XRF results over all ten testing combination selected
for retesting.
Take the difference of the overall average of the ten original XRF results and the overall average of
the ten retest XRF results. If the difference is negative, drop the negative sign.
If the difference of the overall averages is less than the Retest Tolerance Limit, the inspection has
passed the retest. If the difference of the overall averages equals or exceeds the Retest Tolerance
Limit, this procedure should be repeated with ten new testing combinations. If the difference of the
overall averages is equal to or greater than the Retest Tolerance Limit a second time, then the
inspection should be considered deficient.
Use of this procedure is estimated to produce a spurious result approximately 1 % of the time. That is, results
of this procedure will call for further examination when no examination is warranted in approximately 1 out of
100 dwelling units tested.
6 of 12
Figure D-2 continued. XRF Performance Characteristic Sheet for the Radiation
Monitoring Devices LPA-1.
D-23
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
Radiation Monitoring Devices; LPA-1
BIAS AND PRECISION.
Do not use these bias and precision data to correct for substrate bias. These bias and precision data
were computed without substrate correction from samples with reported laboratory results less than 4.0
mg/cm2 lead. The data which were used to determine the bias and precision estimates given in the three
tables above have the following properties. During the March testing, there were 11 test locations with a
laboratory reported result equal to or greater than 4.0 mg/cm2 lead. Of these, one 20-second standard mode
reading was less than 1.0 mg/cm2 and none of the quick mode readings were less than 1.0 mg/cm2. During
the July testing, there were 15 test locations with a laboratory reported result equal to or greater than 4.0
mg/cm2 lead. Of these, one 30-second standard mode reading was less than 1.0 mg/cm2 and none of the
quick mode readings were less than 1.0 mg/cm2. During the September testing, there were 15 test locations
with a laboratory reported result equal to or greater than 4.0 mg/cm2 lead. Of these, two 20-second and one
30-second standard mode readings were less than 1.0 mg/cm2, and one quick mode reading was less than
1.0 mg/cm2. The two instruments that tested in March and September are representative of instruments sold
prior to June 26, 1995 and have not been serviced since June 26, 1995 and the instrument that tested in July
is representative of instruments sold or serviced after June 26,1995. These data are for illustrative purposes
only. Actual bias must be determined on the site. Inconclusive ranges provided above already account for
bias and precision. Bias and precision ranges are provided to show the variability that was found between
machines of the same model. Units are in mg/cm2.
7 of 12
Figure D-2 continued.
XRF Performance Characteristic Sheet for the Radiation
Monitoring Devices LPA-1.
D-24
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
Radiation Monitoring Devices; LPA-1
For those instruments sold Drier to June 26. 1995 and have not been serviced since June 26. 1995:
20-SECOND READING
MEASURED AT
0.0 mg/cm2
0.5 mg/cm2
1 .0 mg/cm2
2.0 mg/cm2
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
BIAS
(mg/cm2)
-0.2
-0.2
-0.2
-0.4
-0.1
-0.2
-0.2
-0.2
-0.3
-0.5
-0.2
-0.2
-0.3
-0.3
-0.3
-0.6
-0.2
-0.3
-0.4
-0.4
-0.5
-0.7
-0.3
-0.4
BIAS
RANGE*
(mg/cm2)
(-0.1, -0.2)
(-0.1, -0.2)
(-0.2.-0.3)
(-0.4.-0.5)
(-01, -0.1)
(-0.1, -0.2)
(-0.2.-0.3)
(-0.2.-0.3)
(-0.2.-0.3)
(-0.4.-0.6)
(-0.2.-0.2)
(-0.2.-0.3)
(-0.2.-0.4)
(-0.2.-0.4)
(-0.3.-0.4)
(-0.5.-0.6)
(-0.2.-0.2)
(-0.2.-0.4)
(-0.3.-0.5)
(-0.3.-0.5)
(-0.4.-0.5)
(-0.6.-0.7)
(-0.3.-0.4)
(-0.3.-0.5)
PRECISION*
(mg/cm2)
0.2
0.2
0.2
0.2
0.2
0.2
0.3
0.3
0.3
0.3
0.3
0.3
0.3
0.3
0.3
0.3
0.3
0.3
0.5
0.5
0.5
0.5
0.5
0.5
PRECISION
RANGE'
(mg/cm2)
(0.1,0.2)
(0.1,0.2)
(0.1,0.2)
(0.1,0.2)
(0.1,0.2)
(0.1,02)
(0.2, 0.3)
(0.2, 0.3)
(0.2, 0.3)
(0.2, 0.3)
(0.2, 0.3)
(0.2, 0.3)
(0.3, 0.4)
(0.3, 0.4)
(0.3, 0.4)
(0.3, 0.4)
(0.3, 0.4)
(0.3, 0.4)
(04,0.5)
(0.4, 0.5)
(0.4, 0.5)
(0.4, 0 5)
(0.4, 0.5)
(0.4, 0.5)
'Ranges are provided to show the variability between machines of the same model.
Precision at 1 standard deviation.
8 of 12
Figure D-2 continued.
XRF Performance Characteristic Sheet for the Radiation
Monitoring Devices LPA-1.
D-25
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
Radiation Monitoring Devices; LPA-1
For those instruments sold orior to June 26. 1 995 and have not been serviced since June 26. 1 995:
30-SECOND READING MEASURED AT
0.0 mg/cm2
0.5 mg/cm2
1.0 mg/cm2
2.0 mg/cm2
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
BIAS
(mg/cm2)
-0.1
-0.1
-0.2
-0.5
-0.1
-0.1
-0.2
-0.2
-0.3
-0.5
-0.1
-0.2
-0.2
-0.2
-0.3
-0.6
-0.2
-0.2
-0.4
-0.4
-0.5
-0.7
-0.3
-0.4
PRECISION'
(mg/cm2)
0.1
0.1
0.1
0.1
0.1
0.1
0.2
0.2
0.2
0.2
0.2
0.2
0.3
0.3
0.3
0.3
0.3
0.3
0.4
0.4
0.4
0.4
0.4
0.4
'Precision at 1 standard deviation.
9 of 12
Figure D-2 continued.
XRF Performance Characteristic Sheet for the Radiation
Monitoring Devices LPA-1.
D-26
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
Radiation Monitoring Devices; LPA-1
For those instruments sold oriorto June 26. 1995 and have not been serviced since June 26. 1995.
QUICK MODE
MEASURED AT
0.0 mg/cm2
0.5 mg/cm2
1 .0 mg/cm2
2 0 mg/cm2
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
BIAS
(mg/cm2)
-0.2
-0.2
-0.3
-0.6
-0.2
-0.2
-0.3
-0.3
-0.3
-0.7
-0.2
-0.3
-03
-0.3
-0.4
-0.7
-0.3
-0.3
-0.4
-0.4
-0.5
-0.8
-0.4
-0.4
BIAS RANGE*
(mg/cm2)
(-0.2.-0.3)
(-0 2,-0 3)
(-0.2.-0 3)
(-0.6.-0.7)
(-0.2.-0.2)
(-0.2.-0.3)
(-0 2.-0.3)
(-0.2.-0.3)
(-0.3.-0.4)
(-0.6.-0 8)
(-0.2.-0.2)
(-0 2,-O.S)
(-0.3.-0.4)
(-0.3.-0.4)
(-0.3.-0.4)
(-0.7.-0.8)
(-0 3,-O.S)
(-0 3,-0 4)
(-0.4.-0.5)
(-0.4.-0 5)
(-0.4.-0.6)
(-07.-1.0)
(-0.4.-0.4)
(-0.4.-0.5)
PRECISION'
(mg/cm2)
0.3
03
0.3
0.3
0.3
0.3
04
0.4
04
0.4
0.4
04
0.4
0.4
0.4
0.4
04
0.4
0.5
0.5
0.5
0.5
0.5
05
PRECISION
RANGE+
(mg/cm2)
(0.3, 0.3)
(0.3, 0.3)
(0.3, 0.3)
(0.3, 0.3)
(0.3, 0.3)
(0.3, 0.3)
(0.3, 0.4)
(0.3, 0.4)
(0.3, 0.4)
(0.3, 0 4)
(0 3, 0.4)
(0.3, 0 4)
(0.4, 0.5)
(0.4, 0.5)
(0.4, 0.5)
(0.4, 0.5)
(0.4, 0.5)
(0.4, 0.5)
(0.4, 0.6)
(0.4, 0.6)
(0.4, 0.6)
(0.4, 0.6)
(0.4, 0.6)
(0.4, 0.6)
^Ranges are provided to show the variability between machines of the same model
'Precision at 1 standard deviation.
10 of 12
Figure D-2 continued.
XRF Performance Characteristic Sheet for the Radiation
Monitoring Devices LPA-1.
D-27
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
Radiation Monitoring Devices; LPA-1
For those instruments sold or serviced after June 26. 1995:
30-SECOND STANDARD MODE
READING MEASURED AT
0.0 mg/cm2
0.5 mg/cm2
1 .0 mg/cm2
2.0 mg/cm2
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
BIAS
(mg/cm2)
0.0
0.0
0.1
0.3
0.1
0.0
0.0
0.0
0.0
0.2
0.0
0.0
0.0
0.0
0.0
0.2
0.0
0.0
-0.1
-0.1
-0.1
0.1
-0.1
-0.1
PRECISION'
(mg/cm2)
0.1
01
0.1
0.1
0.1
0.1
0.2
0.2
0.2
0.2
0.2
0.2
0.3
0.3
0.3
0.3
0.3
0.3
0.4
0.4
0.4
0.4
0.4
0.4
'Precision at 1 standard deviation.
11 of 12
Figure D-2 continued.
XRF Performance Characteristic Sheet for the Radiation
Monitoring Devices LPA-1.
D-28
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
Radiation Monitoring Devices; LPA-1
For those instruments sold or serviced after June 26. 1995:
QUICK MODE READING
MEASURED AT
0.0 mg/cm2
0.5 mg/cm2
1 .0 mg/cm2
2.0 mg/cm2
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
BIAS
(mg/cm2)
0.0
0.0
0.1
0.2
0.0
0.0
00
0.0
0.0
0.2
0.0
0.0
0.0
0.0
0.0
0.1
-0.1
0.0
-0.1
-0.1
-0.1
0.1
-0.1
-0.1
PRECISION*
(mg/cm2)
0.2
0.2
0.2
0.2
0.2
0.2
03
0.3
0.3
0.3
0.3
0.3
0.4
0.4
0.4
0.4
0.4
0.4
05
0.5
0.5
0.5
0.5
0.5
'Precision at 1 standard deviation.
A document titled Methodology for XRF Performance Characteristic Sheets provides an explanation of
the statistical methodology used to construct the data in the sheets and provides emp rical results from using
the recommended inconclusive ranges or thresholds for specific XRF instruments. For a copy of this
document call the National Lead Information Center Clearinghouse at 1-800-424-LEAD.
This XRF Performance Characteristic Sheet is a joint product of the U S. Environmental Protection Agency (EPA) and the U S
Department of Housing and Urban Development (HUD). The issuance of this sheet does not constitute rulemakmg. The
information provided here is intended solely as guidance to be used in conjunction with Chapter 7 of the Guidelines for the
Evaluation and Control of Lead-Based Paint Hazards in Housing. EPA and HUD reserve the right to revise this guidance
Please address questions and comments on this sheet to: Director, Office of Lead-Based Paint Abatement and Poisoning
Prevention, U S Department of Housing and Urban Development, Room B-133, 451 Seventh St, S.W , Washington, DC 20410
12 of 12
Figure D-2 continued.
XRF Performance Characteristic Sheet for the Radiation
Monitoring Devices LPA-1.
D-29
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
Table D-9. Threshold Results For Radiation Monitoring Device LPA-1 30-Second Standard Mode Readings
For Results Where Substrate Correction Is Not Performed, But Substrate Correction Is
Recommended For Instruments For Those Instruments Sold Prior to June 26, 1995 and Have
Not Been Serviced Since June 26,1995.
30-SECOND STANDARD MODE READING DESCRIPTION
Readings not corrected for substrate bias on any substrate
SUBSTRATE
Metal
Wood
THRESHOLD (mg/cm2)
0.4
0.8
Table D-10. Threshold Results For Radiation Monitoring Device LPA-1 20-Second Standard Mode
Readings For Results Where Substrate Correction Is Not Performed, But Substrate
Correction Is Recommended For Instruments For Those Instruments Sold Prior to June 26,
1995 and Have Not Been Serviced Since June 26, 1995.
20-SECOND STANDARD MODE READING DESCRIPTION
Readings not corrected for substrate bias on any substrate
SUBSTRATE
Metal
Wood
THRESHOLD (mg/cm2)
0.4
0.7
Table D-11. Inconclusive Range Results For Radiation Monitoring Device LPA-1 Quick Mode Readings
For Results Where Substrate Correction Is Not Performed, But Substrate Correction Is
Recommended For Instruments For Those Instruments Sold Prior to June 26, 1995 and
Have Not Been Serviced Since June 26, 1995.
QUICK MODE READING DESCRIPTION
Readings not corrected for substrate bias on any substrate
SUBSTRATE
Metal
Wood
INCONCLUSIVE RANGE
(mg/cm2)
0.3 to 0.4
0.7 to 0.8
Table D-12. Threshold Results For Radiation Monitoring Device LPA-1 30-Second Standard Mode
Readings For Results Where Substrate Correction Is Not Performed, But Substrate
Correction Is Recommended For Instruments For Those Instruments Sold or Serviced After
June 26,1995.
30-SECOND STANDARD MODE READING DESCRIPTION
Readings not corrected for substrate bias on any substrate
SUBSTRATE
Metal
THRESHOLD (mg/cm2)
0.9to1.2
D-30
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
Table D-13. Bias Estimates, and Their Standard Errors, of Radiation Monitoring Device LPA-1 20-Second
Standard Mode Readings For Those Instruments Sold Prior to June 26, 1995 and Have Not
Been Serviced Since June 26,1995.
20-SECOND
STANDARD MODE READING
MEASURED AT
0.0 mg/cm2
0.5 mg/cm2
1 .0 mg/cm2
2.0 mg/cm2
CORRECTED?
Yes
No
Yes
No
Yes
No
Yes
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
Wood
BIAS (mg/cm2)
-0.2
-0.2
-0.2
-0.4
-0.1
-0.2
0.0
0.0
-0.2
-0.2
-0.3
-0.5
-0.2
-0.2
-0.2
-0.1
-0.3
-0.3
-0.3
-0.6
-0.2
-0.3
-04
-0.2
-0.4
-0.4
-0.5
-0.7
-0.3
-0.4
-0.8
-0.3
STANDARD
ERROR FOR
BIAS
0.03
0.03
0.05
0.04
0.04
0.03
0.03
003
0.04
0.04
0.06
0.04
004
0.04
0.04
0.04
0.05
0.05
0.07
0.06
0.06
0.05
0.05
0.05
010
0.10
0.12
0.10
0.10
0.10
0.10
0.10
D-31
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
Table D-14. Bias Estimates, and Their Standard Errors, of Radiation Monitoring Device LPA-1 30-Second
Standard Mode Readings For Those Instruments Sold Prior to June 26, 1995 and Have Not
Been Serviced Since June 26, 1995.
30-SECOND
STANDARD MODE READING
MEASURED AT
0.0 mg/cm2
0.5 mg/cm2
1 .0 mg/cm2
2.0 mg/cm2
CORRECTED?
Nn
liU
Vac
I 65
Kin
I'lu
Yes
Nn
I'lU
Yes
Nln
INU
Yes
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
Wood
BIAS
(mg/cm2)
-0.1
-0.1
-0.2
-0.5
-0.1
-0.1
0.0
0.0
-0.2
-0.2
-0.3
-0.5
-0.1
-0.2
-0.1
0.0
-0.2
-0.2
-0.3
-0.6
-0.2
-0.2
-0.2
-0.1
-0.4
-0.4
-0.5
-0.7
-0.3
-0.4
-0.3
-0.2
STANDARD
ERROR FOR
BIAS
0.03
0.03
0.04
0.03
0.03
0.03
005
0.03
0.03
0.03
0.04
0.03
0.03
0.03
0.05
0.03
0.05
0.05
0.06
0.05
0.05
0.05
0.06
0.05
0.09
0.09
0.10
0.10
0.09
0.09
0.10
0.09
D-32
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
Table D-15. Bias Estimates, and Their Standard Errors, of Radiation Monitoring Device LPA-1 Quick
Mode Readings For Those Instruments Sold Prior to June 26, 1995 and Have Not Been
Serviced Since June 26,1995.
QUICK MODE
MEASURED AT
0.0 mg/cm2
0.5 mg/cm2
1 .0 mg/cm2
2.0 mg/cm2
CORRECTED?
No
Yes
No
Yes
No
Yes
No
Yes
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
Wood
BIAS
(mg/cm2)
-0.2
-0.2
-0.3
-0.6
-0.2
-0.2
0.0
-0.1
-0.3
-03
-0.3
-0.7
-0.2
-0.3
0.0
-0.2
-0.3
-0.3
-0.4
-0.7
-0.3
-0.3
-0.1
-0.3
-0.4
-0.4
-0.5
-0.8
-0.4
-0.4
-0.2
-0.4
STANDARD ERROR FOR
BIAS
0.05
0.05
0.08
0.06
0.06
0.05
0.06
0.05
0.05
0.05
0.09
006
0.06
0.05
0.06
0.05
0.07
0.07
0.10
0.07
0.07
0.07
0.07
0.07
0.12
0.12
0.14
0.12
0.12
0.12
0.12
0.12
D-33
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
Table D-16. Bias Estimates, and Their Standard Errors, of Radiation Monitoring Device LPA-1 30-Second
Standard Mode Readings For Those Instruments Sold or Serviced after June 26, 1995.
30-SECOND
STANDARD MODE READING
MEASURED AT
0.0 mg/cm2
0.5 mg/cm2
1 .0 mg/cm2
2.0 mg/cm2
CORRECTED?
Nln
Yes
Mn
Yes
Mn
Yes
Nn
Yes
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
BIAS (mg/cm2)
0.0
0.0
0.1
0.3
0.1
0.0
0.0
0.0
0.0
0.0
0.2
0.0
0.0
-0.1
0.0
0.0
0.0
0.2
00
0.0
-01
-0.1
-0.1
-0.1
0.1
-0.1
-0.1
-0.2
STANDARD
ERROR FOR BIAS
003
0.03
0.04
0.03
0.03
0.03
0.03
0.03
0.03
0.05
0.03
0.03
0.03
0.03
0.05
0.05
0.06
0.05
0.05
0.05
0.05
0.09
0.09
0.10
0.09
0.09
0.09
0.09
D-34
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
Table D-17. Bias Estimates, and Their Standard Errors, of Radiation Monitoring Device LPA-1 Quick
Mode Readings For Those Instruments Sold or Serviced after June 26, 1995.
QUICK MODE
READING MEASURED AT
0.0 mg/cm2
0.5 mg/cm2
1.0 mg/cm2
2.0 mg/cm2
CORRECTED?
No
No
No
No
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
BIAS (mg/cm2)
0.0
0.0
0.1
02
0.0
0.0
0.0
0.0
0.0
0.2
0.0
0.0
0.0
00
0.0
0.1
-0.1
0.0
-01
-01
-0.1
0.1
-0.1
-0.1
STANDARD
ERROR FOR BIAS
0.04
0.04
0.06
0.05
0.05
0.04
0.04
0.04
0.07
005
0.05
0.04
0.06
0.06
008
007
0.07
0.06
012
012
0.13
0.11
0.12
0.12
D-35
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
Table D-18. Precision Estimates, and Their Standard Errors, of Radiation Monitoring Device LPA-1
20-Second Standard Mode Readings For Those Instruments Sold Prior to June 26,1995 and
Have Not Been Serviced Since June 26, 1995.
20-SECOND
STANDARD MODE READING
MEASURED AT
0.0 mg/cm2
0.5 mg/cm2
1 .0 mg/cm2
2.0 mg/cm2
CORRECTED?
No
Yes
No
Yes
No
Yes
No
Yes
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
Wood
Brick
Concrete
' Drywall
Metal
Plaster
Wood
Metal
Wood
PRECISION'
(mg/cm2)
0.2
0.2
0.2
0.2
0.2
0.2
0.2
0.2
0.3
0.3
0.3
0.3
0.3
0.3
0.3
0.3
0.3
0.3
0.3
0.3
03
0.3
0.4
0.3
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
STANDARD
ERROR FOR
PRECISION
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.03
0.03
0.03
0.03
0.03
0.03
0.02
0.03
0.04
0.04
0.04
0.04
0.04
0.04
0.04
0.04
0.07
0.07
0.07
0.07
0.07
0.07
0.07
007
'Precision at 1 standard deviation.
D-36
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
Table D-19. Precision Estimates, and Their Standard Errors, of Radiation Monitoring Device LPA-1
30-Second Standard Mode Readings For Those Instruments Sold Prior to June 26, 1995 and
Have Not Been Serviced Since June 26, 1995.
30-SECOND
STANDARD MODE
READING MEASURED AT
0.0 mg/cm2
0.5 mg/cm2
1 .0 mg/cm2
2.0 mg/cm2
CORRECTED?
No
Yes
No
Yes
Kin
Yes
Nn
Yes
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
Wood
PRECISION'
(mg/cm2)
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.2
0.2
0.2
02
0.2
0.2
0.2
0.2
0.3
0.3
0.3
0.3
0.3
03
0.3
03
04
0.4
0.4
0.4
04
04
0.4
0.4
STANDARD
ERROR FOR
PRECISION
0.01
0.01
0.01
0.01
0.01
0.01
0.02
0.01
0.02
0.02
002
002
002
0.02
0.02
0.02
004
0.04
004
0.04
0.04
0.04
0.04
003
0.05
0.05
0.05
0.05
005
0.05
0.05
0.05
'Precision at 1 standard deviation.
D-37
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
Table D-20. Precision Estimates, and Their Standard Errors, of Radiation Monitoring Device LPA-1 Quick
Mode Readings For Those Instruments Sold Prior to June 26, 1995 and Have Not Been
Serviced Since June 26,1995.
QUICK MODE
MEASURED AT
0.0 mg/cm2
0.5 mg/cm2
1 .0 mg/cm2
2.0 mg/cm2
CORRECTED?
Nn
Yes
Nn
Yes
No
Yes
No
Yes
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
Wood
PRECISION' (mg/cm2)
0.3
0.3
0.3
0.3
0.3
0.3
0.3
0.3
0.4
0.4
0.4
0.4
0.4
0.4
0.4
0.4
0.4
0.4
0.4
0.4
0.4
0.4
0.5
0.4
0.5
0.5
0.5
0.5
0.5
0.5
0.6
0.5
STANDARD ERROR FOR
PRECISION
003
0.03
0.03
0.03
0.03
003
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.05
005
0.05
0.05
0.05
0.05
0.05
0.05
0.09
0.09
0.09
0.09
009
0.09
0.09
0.09
'Precision at 1 standard deviation.
D-38
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
Table D-21. Precision Estimates, and Their Standard Errors of Radiation Monitoring Device LPA-1
30-Second Standard Mode Readings For Those Instruments Sold or Serviced After June 26,
1995.
30-SECOND
STANDARD MODE
READING MEASURED AT
0.0 mg/cm2
0.5 mg/cm2
1 .0 mg/cm2
2.0 mg/cm2
CORRECTED?
Nn
Yes
Nn
Yes
No
Yes
No
Yes
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
PRECISION'
(mg/cm2)
01
0.1
0.1
0.1
0.1
0.1
0.2
0.2
02
0.2
0.2
0.2
0.2
0.2
0.3
0.3
0.3
0.3
0.3
0.3
0.3
0.4
0.4
0.4
0.4
0.4
0.4
0.4
STANDARD
ERROR FOR
PRECISION
0.01
0.01
001
0.01
001
0.01
0.01
0.02
0.02
0.02
0.02
0.02
0.02
0.02
004
0.04
004
0.04
0.04
0.04
0.03
005
0.05
0.05
0.05
0.05
0.05
0.05
'Precision at 1 standard deviation.
D-39
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
Table D-22. Precision Estimates, and Their Standard Errors, of Radiation Monitoring Device LPA-1 Quick
Mode Readings For Those Instruments Sold or Serviced After June 26, 1995.
QUICK MODE
READING MEASURED AT
0.0 mg/cm2
0.5 mg/cm2
1 .0 mg/cm2
2.0 mg/cm2
CORRECTED?
No
No
No
No
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
PRECISION'
(mg/cm2)
0.2
0.2
0.2
0.2
0.2
0.2
0.3
03
0.3
0.3
0.3
0.3
0.4
0.4
0.4
0.4
0.4
0.4
0.5
0.5
0.5
0.5
0.5
0.5
STANDARD
ERROR FOR
PRECISION
0.02
0.02
0.02
0.02
0.02
0.02
0.03
0.03
0.03
0.03
0.03
0.03
0.05
0.05
0.05
005
0.05
0.05
0.07
0.07
0.07
0.07
0.07
0.07
'Precision at 1 standard deviation.
D-40
-------
APPENDIX D: XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
Table D-23. Classification Results For Radiation Monitoring Device LPA-1 30-Second Readings (Metal
Substrates Corrected and Uncorrected), Classified Using Threshold Values Reported in the
XRF Performance Characteristic Sheet For Instruments Sold Or Serviced After June 26,
1995 and Compared to Laboratory Results in mg/cm2 Lead Classified Using the 1.0 mg/cm2
Lead Federal Standard for Data Taken From Testing Archived Building Components in July
1995.
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal (Uncorrected)
TOTAL'
THRESHOLD
VALUES
1.0
1.0
1.0
0.9
1.0
1.0
0.9 to 1.2
FALSE POSITIVE
RATE
0.0% (0/1)
0.0% (0/2)
0.0% (0/14)
0.0% (0/27)
3.5% (1/29)
5.0% (2/40)
0.0% (0/27)
2.7% (3/1 13)
FALSE NEGATIVE
RATE
0.0% (0/2)
(0/0)
(0/0)
0.0% (0/11)
11.1% (1/9)
8.7% (2/23)
0.0% (0/11)
6.7% (3/45)
INCONCLUSIVE
RATE
0.0%
00%
0.0%
0.0%
0.0%
0.0%
10.5% (4/38)
0.0%
'Total results are for values reported in PCS.
Table D-24. Classification Results For Radiation Monitoring Device LPA-1 Quick Mode Readings
(Uncorrected), Classified Using Inconclusive Ranges and Threshold Values Reported in the
XRF Performance Characteristic Sheet For Instruments Sold Or Serviced After June 26,
1995 and Compared to Laboratory Results in mg/cm2 Lead Classified Using the 1.0 mg/cm2
Lead Federal Standard For Data Taken From Testing Archived Building Components in July
1995.
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
TOTAL
INCONCLUSIVE
RANGE OR
THRESHOLD
1.0
1.0
1.0
0.9 to 1.3
0.9 to 1.0
1.0
FALSE POSITIVE
RATE
0.0% (0/1)
0.0% (0/2)
0.0% (0/14)
3.7% (1/27)
3.5% (1/29)
5.0% (2/40)
3.5% (4/1 13)
FALSE NEGATIVE
RATE
0.0% (0/2)
(0/0)
(0/0)
0.0% (0/11)
0.0% (0/9)
8.7% (2/23)
4.4% (2/45)
INCONCLUSIVE
RATE
0.0% (0/3)
0.0% (0/2)
0.0% (0/14)
13.2% (5/38)
0.0% (0/38)
0.0% (0/63)
3.2% (5/1 58)
D-41
-------
APPENDIX D: XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
Table D-25. Classification Results For Radiation Monitoring Device LPA-1 30-Second Readings (Metal
and Wood Substrates Corrected and Uncorrected), Classified Using Threshold Values
Reported in the XRF Performance Characteristic Sheet For Instruments That Were Sold
Prior to June 26, 1995 and Have Not Been Serviced Since June 26, 1995 and Compared to
Laboratory Results in mg/cm2 Lead Classified Using the 1.0 mg/cm2 Lead Federal Standard
For Data Taken From Testing Archived Building Components in March and September 1995.
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal (Uncorrected)
Wood (Uncorrected)
TOTAL'
THRESHOLD
VALUES
0.8
0.8
0.7
0.8
0.8
0.9
0.4
0.8
FALSE POSITIVE
RATE
0.0% (0/1)
0.0% (0/2)
0.0% (0/14)
3.7% (1/27)
3.5% (1/29)
5.1% (2/39)
3.7% (1/27)
5.0% (2/40)
3. 5% (4/1 13)
FALSE NEGATIVE
RATE
0.0% (0/2)
(0/0)
(0/0)
0.0% (0/11)
11.1% (1/9)
8.7% (2/23)
0.0% (0/11)
8.7% (2/23)
6.7% (3/45)
INCONCLUSIVE
RATE
00%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0% (0/38)
0.0% (0/63)
0.0%
'Total results are for values reported in PCS.
D-42
-------
APPENDIX D: XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
Table D-26. Classification Results For Radiation Monitoring Device LPA-1 20-Second Readings (Metal
and Wood Substrates Corrected and Uncorrected), Classified Using Threshold Values
Reported in the XRF Performance Characteristic Sheet For Instruments That Were Sold
Prior to June 26,1995 and Have Not Been Serviced Since June 26, 1995 and Compared to
Laboratory Results in mg/cm2 Lead Classified Using the 1.0 mg/cm2 Lead Federal Standard
For Data Taken From Testing Archived Building Components in March and September 1995.
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal (Uncorrected)
Wood (Uncorrected)
TOTAL'
THRESHOLD
VALUES
0.7
0.7
0.7
0.9
0.8
0.8
0.4
0.7
FALSE POSITIVE
RATE
0.0% (0/2)
0.0% (0/4)
0.0% (0/28)
5.6% (3/54)
3.5% (2/58)
6.3% (5/80)
9.3% (5/54)
6.3% (5/80)
4.4% (10/226)
FALSE NEGATIVE
RATE
0.0% (0/4)
(0/0)
(0/0)
0 0% (0/22)
71% (1/14)
13.0% (6/46)
0.0% (0/22)
10. 9% (5/46)
8.1% (7/86)
INCONCLUSIVE
RATE
0.0%
0.0%
0.0%
0.0%
0.0%
00%
0.0% (0/76)
0.0% (0/1 26)
0.0%
'Total results are for values reported in PCS.
D-43
-------
APPENDIX D: XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
Table D-27. Classification Results For Radiation Monitoring Device LPA-1 Quick Mode Readings (Metal
and Wood Substrates Corrected and Uncorrected), Classified Using Inconclusive Ranges
Reported in the XRF Performance Characteristic Sheet For Instruments That Were Sold
Prior to June 26,1995 and Have Not Been Serviced Since June 26,1995 and Compared to
Laboratory Results in mg/cm2 Lead Classified Using the 1.0 mg/cm2 Lead Federal Standard
For Data Taken From Te
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal (Uncorrected)
Wood (Uncorrected)
TOTAL8
INCONCLUSIVE
RANGE
0.7 to 0.8
0.7 to 0.8
0.6 to 0.8
0.9 to 1.0
0.7 to 0.8
0.7 to 0.8
0.3 to 0.4
0.7 to 0.8
sting Archived Building
FALSE POSITIVE
RATE
0.0% (0/2)
0.0% (0/4)
0.0% (0/28)
1.9% (1/54)
3.5% (2/58)
3.8% (3/80)
0.0% (0/54)
3.9% (3/80)
2.7% (6/226)
Components in March and September 1995.
FALSE NEGATIVE
RATE
0.0% (0/4)
(0/0)
(0/0)
4.6% (1/22)
7.1% (1/14)
13.0% (6/46)
4.6% (1/22)
13.0% (6/46)
9.3% (8/86)
INCONCLUSIVE
RATE
0.0% (0/6)
0.0% (0/4)
0.0% (0/28)
0.0% (0/76)
0.0% (0/72)
0.8% (1/1 26)
0.0% (0/76)
0.0% (0/1 26)
0.3% (1/312)
Total results are for values reported in PCS.
D-44
-------
APPENDIX D: XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
D.4 XRF Performance Characteristic Sheet for the Scitec MAP-3 and Related
Results
D-45
-------
APPENDIX D: XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
Scitec Corporation; MAP-3
EFFECTIVE DATE: August 24, 1995 EDITION NO.: 1
MANUFACTURER AND MODEL:
Make: Scitec Corporation
Model: MAP-3
Source: Co"
Note: This sheet supersedes all previous sheets for the XRF instrument of the make, model, and source
shown above.
EVALUATION DATA SOURCE AND DATE:
This sheet is supplemental information to be used in conjunction with Chapter 7 of the HUD Guidelines
for the Evaluation and Control of Lead-Based Paint Hazards in Housing ("HUD Guidelines"). Performance
parameters shown on this sheet are calculated from evaluation data collected during the EPA/HUD field
evaluation study conducted from March through October 1993. The data were collected from four instruments
at approximately 1,200 15-second test locations and 300 60-second test locations. One instrument had a
January 1993 source and the other three instruments had July 1993 sources. All four instruments had
sources with 40 mCi initial strengths. The results of this study are reported in A Field Test of Lead-Based
Paint Testing Technologies: Technical Report, EPA 747-R-95-002b, May 1995.
FIELD OPERATION GUIDANCE
OPERATING PARAMETERS:
Performance parameters shown in this sheet are applicable only when operating the instrument under
the same conditions as the evaluation testing and using the procedures described in Chapter 7 of the HUD
Guidelines. Operating parameters include:
• Manufacturer-recommended warm-up and quality control procedures
• Use the Multifamily Decision Flowchart for determining the presence of lead on a component type in
multifamily housing
• Nominal 15-second or nominal 60-second readings on three locations per component for single-family
housing and one location per component for multifamily housing
• Calibration checks are taken using the red (1.02 mg/cm2) NIST Standard Reference Material (SRM
No. 2579) paint film
• Readings for determining the substrate correction values are taken on bare substrate covered with
red (1.02 mg/cm2) NIST SRM paint film
• Lead-based paint is defined as paint with lead equal to or in excess of 1.0 mg/cm2.
1of6
Figure D-3. XRF Performance Characteristic Sheet for the Scitec Corporation MAP-3.
D-46
-------
APPENDIX D: XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
Scitec Corporation; MAP-3
XRF CALIBRATION CHECK:
Chapter 7 of the HUD Guidelines recommends using a calibration check procedure to determine the
operating condition of the XRF instrument. If the observed calibration check average minus 1.02 mg/cm2 is
greater than the positive (plus) calibration check tolerance value, or less than the negative (minus) calibration
check tolerance value, then the instructions provided by the manufacturer should be followed in order to bring
the instrument back into control before any more XRF testing is done. This calibration check is estimated to
produce an incorrect result (that is, a finding that the instrument is out of calibration) very infrequently - once
out of every 200 times this procedure is followed.
15-SECOND READINGS
minus value = -0.6 mg/cm2
plus value = +0.3 mg/cm2
60-SECOND READINGS
minus value = -0.4 mg/cm2
plus value = +0.1 mg/cm2
FOR XRF RESULTS BELOW 4.0 mg/cm2, SUBSTRATE CORRECTION RECOMMENDED FOR:
Metal and Wood
SUBSTRATE CORRECTION NOT RECOMMENDED FOR:
Brick, Concrete, Drywall, and Plaster
SUBSTRATE CORRECTION VALUE COMPUTATION:
Chapter 7 of the HUD Guidelines provides guidance on correcting XRF results for substrate bias.
Supplemental guidance for using the red (1.02 mg/cm2) NIST SRM paint film for substrate correction is
provided below.
XRF results are corrected for substrate bias by subtracting from each XRF result a correction value
determined separately in each house for single-family housing or in each development for multifamily housing,
for each substrate. The correction value is an average of XRF readings taken over red NIST SRM (1.02
mg/cm2) paint films at test locations that had been scraped clean of their paint covering. Compute the
correction values as follows:
Using the same XRF instrument, take three readings on a bare substrate area covered with the red NIST
SRM (1.02 mg/cm2) paint film. Repeat this procedure by taking three more readings on a second bare
substrate area of the same substrate covered with the red NIST SRM (1.02 mg/cm2) paint film.
• Compute the correction value for each substrate type by computing the average of all six readings as
shown below.
2 of 6
Figure D-3 continued. XRF Performance Characteristic Sheet for the Scitec
Corporation MAP-3.
D-47
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
Scitec Corporation; MAP-3
For each substrate type:
Correction
Value
2"°
6" Reading
1.02mg/cm2
Repeat this procedure for each substrate tested in the house or housing development.
INCONCLUSIVE RANGE:
XRF results are classified as positive if they are greater than or equal to the upper limit of the inconclusive
range, and negative if they are less than or equal to the lower limit of the inconclusive range. In single-family
housing, an XRF result is the average of three readings taken on a testing combination. (A testing
combination is a location on a painted surface as defined in Chapter 7 of the HUD Guidelines). In multifamily
housing, an XRF result is a single reading taken on a testing combination. For computing the XRF result, use
all digits that are reported by the instrument. Inconclusive ranges were determined for comparing results to
the 1.0 mg/cm2 standard. For a listing of laboratories recommended by the EPA National Lead Laboratory
Accreditation Program (NLLAP) for the analysis of samples to resolve an inconclusive XRF result or additional
confirmational analysis, call the National Lead Information Center Clearinghouse at 1-800-424-LEAD.
15-SECOND READING DESCRIPTION
Results corrected for substrate bias for 1 5-
second readings on metal and wood
substrates only
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
INCONCLUSIVE
RANGE (mg/cm2)
0.0 to 1.5
0.0 to 1.5
0.9 to 1.0
0.9 to 1.3
0.3 to 1.3
0.9 to 1.3
60-SECOND READING DESCRIPTION
Readings corrected for substrate bias
for 60-second readings on metal
and wood substrates only
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
THRESHOLD
(mg/cm2)
None
None
None
None
None
1.0
INCONCLUSIVE
RANGE (mg/cm2)
0.3 to 0.9
0.3 to 0.9
0.6 to 0.8
0.9 to 1.2
0.2 to 0.9
None
INSTRUCTIONS FOR EVALUATING XRF TESTING:
Chapter 7 of the HUD Guidelines recommends several options for evaluating XRF testing. Among those
options is the following procedure which may be used after XRF testing has been completed. In single-family
housing, an XRF result is the average of three readings taken on a testing combination. (A testing
combination is a location on a painted surface as defined in Chapter 7 of the HUD Guidelines). In multifamily
3 of 6
Figure D-3 continued.
XRF Performance Characteristic Sheet for the Scitec
Corporation MAP-3.
D-48
-------
APPENDIX D: XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
Scitec Corporation; MAP-3
housing, an XRF result is a single reading taken on a testing combination. If a multifamily housing
development is being retested, randomly select two units from within the development from which the ten
testing combinations should be randomly selected.
Randomly select ten testing combinations for retesting from each house or from the two selected units.
Conduct XRF retesting at the ten testing combinations selected for retesting.
Determine if the XRF testing in the units or house passed or failed the test by applying the steps below.
Compute the Retest Tolerance Limit by the following steps:
Determine XRF results for the original and retest XRF readings. Do not correct the original or
retest results for substrate bias. In single-family housing a result is defined as the average of
three readings. In multifamily housing, a result is a single reading. Therefore, there will be ten
original and ten retest XRF results for each house or for the two selected units.
Compute the average of the original and re-test result for each of the ten testing combinations.
Square the average for each testing combination.
Add the ten squared averages together. Call this quantity C.
Multiply the number C by 0.0072. Call this quantity D.
Add the number 0.032 to D. Call this quantity E.
Take the square root of E. Call this quantity F.
Multiply F by 1.645. The result is the Retest Tolerance Limit.
Compute the overall average of all ten retest XRF results over all ten testing combinations selected
for retesting.
Take the difference of the overall average of the ten original XRF results and the overall average of
the ten retest XRF results. If the difference is negative, drop the negative sign.
If the difference of the overall averages is less than the Retest Tolerance Limit, the inspection has
passed the retest. If the difference of the overall averages equals or exceeds the Retest Tolerance
Limit, this procedure should be repeated with ten new testing combinations. If the difference of the
overall averages is equal to or greater than the Retest Tolerance Limit a second time, then the
inspection should be considered deficient.
Use of this procedure is estimated to produce a spurious result approximately 1 % of the time. That is, results
of this procedure will call for further examination when no examination is warranted in approximately 1 out of
100 dwelling units tested.
4 of 6
Figure D-3 continued. XRF Performance Characteristic Sheet for the Scitec
Corporation MAP-3.
D-49
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
Scitec Corporation; MAP-3
BIAS AND PRECISION:
Do not use these bias and precision data to correct for substrate bias. These bias and precision data
were computed without substrate correction from samples with reported laboratory results less than 4.0
mg/cm2 lead. There were 124 15-second testing locations with a laboratory reported result equal to or greater
than 4.0 mg/cm2 lead. Of these, none had XRF readings less than 1 .0 mg/cm2. For the 60-second testing
locations, 34 had laboratory reported results equal to or greater than 4.0 mg/cm2 lead, with 2 of those having
XRF readings less than 1.0 mg/cm2. These data are for illustrative purposes only. Actual bias must be
determined on the site. Inconclusive ranges provided above already account for bias and precision. Units
are in mg/cm2.
1 5-SECOND READING MEASURED
AT
0.0 mg/cm2
0.5 mg/cm2
1 .0 mg/cm2
2.0 mg/cm2
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Dry wall
Metal
Plaster
Wood
BIAS (mg/cm2)
-0.7
-0.7
0.0
0.3
-0.7
-0.1
-0.5
-0.5
-0.1
0.4
-0.6
0.2
-0.4
-0.4
-0.1
0.5
-0.4
0.4
-0.1
-0.1
-0.3
0.6
-0.2
0.8
PRECISION* (mg/cm2)
0.9
0.9
0.4
0.3
0.8
0.5
1.0
1.0
0.4
0.5
0.8
0.6
1.0
1.0
0.4
0.6
0.9
0.7
1.2
1.2
0.4
0.7
0.9
0.8
'Precision at 1 standard deviation.
5 Of 6
Figure D-3 continued.
XRF Performance Characteristic Sheet for the Scitec
Corporation MAP-3.
D-50
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
1
XRF PERFORMANCE CHARACTERISTIC SHEET
Scitec Corporation; MAP-3
60-SECOND READING MEASURED AT
0.0 mg/cm2
0.5 mg/cm2
1 .0 mg/cm2
2.0 mg/cm2
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
BIAS (mg/cm2)
-0.8
-0.8
0.0
0.3
-0.9
-0.2
-0.7
-0.7
-0.2
0.4
-0.7
0.1
-0.7
-0.7
-0.4
0.6
-0.5
0.3
-0.6
-0.6
-0.8
0.8
-0.1
0.8
PRECISION* (mg/cm2)
0.7
0.7
0.3
0.2
0.5
0.4
0.7
0.7
0.3
0.3
0.7
0.4
0.7
0.7
0.3
0.4
0.8
0.5
0.7
0.7
0.3
0.6
1.0
0.6
'Precision at 1 standard deviation.
A document titled Methodology for XRF Performance Characteristic Sheets provides an explanation of the
statistical methodology used to construct the data in the sheets and provides empirical results from using the
ecommended inconclusive ranges or thresholds for specific XRF instruments. For a copy of this document call
he National Lead Information Center Clearinghouse at 1-800-424-LEAD.
This XRF Performance Characteristic Sheet is a joint product of the U.S. Environmental Protection Agency (EPA) and the U.S.
Department of Housing and Urban Development (HUD). The issuance of this sheet does not constitute rulemakmg. The
information provided here is intended solely as guidance to be used in conjunction with Chapter 7 of the Guidelines lor the
Evaluation and Control of Lead-Based Paint Hazards in Housing. EPA and HUD reserve the right to revise this guidance.
Please address questions and comments on this sheet to- Director, Office of Lead-Based Paint Abatement and Poisoning
Prevention, U.S. Department of Housing and Urban Development, Room B-133, 451 Seventh St, S.W., Washington, DC 20410.
6 of 6
Figure D-3 continued.
XRF Performance Characteristic Sheet for the Scitec
Corporation MAP-3.
D-51
-------
APPENDIX D: XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
Table D-28. Inconclusive Ranges For Scitec MAP-3 15-Second K-Shell Readings Where Substrate
Correction Is Not Performed, But Substrate Correction Is Recommended.
15-SECOND READING DESCRIPTION
Readings not corrected for substrate bias on any substrate
SUBSTRATE
Metal
Wood
INCONCLUSIVE
RANGE (mg/cm2)
0.9 to 1.6
0.9 to 1.6
Table D-29. Inconclusive Ranges or Thresholds For Scitec MAP-3 K-Shell 60-Second Readings Where
Substrate Correction Is Not Performed, But Substrate Correction Is Recommended.
60-SECOND READING DESCRIPTION
Readings not corrected for substrate bias on any substrate
SUBSTRATE
Metal
Wood
INCONCLUSIVE
RANGE (mg/cm2)
09 to 1.6
0.9 to 1.3
D-52
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
Table D-30. Bias Estimates, and Their Standard Errors, of Scitec MAP-3 K-Shell 15-Second Readings.
15-SECOND READING MEASURED AT
0.0 mg/cm2
0.5 mg/cm2
1 .0 mg/cm2
2.0 mg/cm2
CORRECTED?
Nn
I'lU
Yes
No
Yes
No
Yes
No
Yes
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
Wood
BIAS (mg/cm2)
-0.7
-0.7
0.0
03
-0.7
-0.1
-0.1
-0.3
-0.5
-0.5
-0.1
0.4
-0.6
0.2
0.0
-01
-0.4
-0.4
-0.1
0.5
-0.4
0.4
0.1
0.1
-01
-0.1
-0.3
0.6
-0.2
0.8
0.3
0.5
STANDARD ERROR
FOR BIAS
0.06
0.06
0.04
0.04
0.07
0.04
004
0.04
0.07
007
0.09
0.04
0.07
0.04
0.04
0.04
0.12
0.12
0.20
0.06
0.14
0.06
0.06
0.06
0.25
0.25
0.41
0.12
0.30
0 13
0.12
0.12
D-53
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
Table D-31. Bias Estimates, and Their Standard Errors, of Scitec MAP-3 K-Shell 60-Second Readings.
60-SECOND READING MEASURED AT
0.0 mg/cm2
0.5 mg/cm2
1 .0 mg/cm2
2.0 mg/cm2
CORRECTED?
Nn
n\j
Vac
TcS
Nn
tw
VAC
T6S
Kin
INU
Vac
I Co
No
I'lU
VAC
Yes
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
Wood
BIAS (mg/cm2)
-0.8
-0.8
0.0
0.3
-0.9
-0.2
0.0
-0.3
-0.7
-0.7
-0.2
0.4
-0.7
0.1
0.1
-0.2
-0.7
-0.7
-0.4
0.6
-0.5
0.3
0.2
0.0
-0.6
-0.6
-0.8
0.8
-0.1
0.8
0.4
0.3
STANDARD ERROR
FOR BIAS
0.13
0.13
0.07
0.05
0.10
0.06
0.05
006
0.12
0.12
0.21
0.05
0.10
006
0.05
0.06
0.22
0.22
0.44
0.10
0.21
0.10
0.09
0.10
0.48
0.48
0.91
0.19
0.46
0.20
0.18
0.19
D-54
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
Table D-32. Precision Estimates, and Their Standard Errors, of Scitec MAP-3 K-Shell 15-Second Readings.
15-SECOND READING
MEASURED AT
0.0 mg/cm2
0.5 mg/cm2
1 .0 mg/cm2
2.0 mg/cm2
CORRECTED?
No
Vac
Nn
No
Voc
No
Yes
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
Wood
PRECISION'
(mg/cm2)
0.9
0.9
0.4
0.3
0.8
0.5
04
0.6
1.0
1.0
0.4
0.5
0.8
0.6
0.5
0.7
1.0
1.0
0.4
0.6
0.9
0.7
0.6
0.8
1.2
1.2
0.4
0.7
0.9
0.8
0.8
0.9
STANDARD ERROR
FOR PRECISION
0.04
0.04
0.03
0.03
0.05
0.03
0.03
0.04
0.06
0.06
0.03
0.03
0.06
0.04
0.03
0.04
0.10
0.10
0.03
0.05
0.10
0.06
0.05
0.06
0.17
017
0.03
0.08
0.19
0.10
0.07
0.10
'Precision at 1 standard deviation
D-55
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
Table D-33. Precision Estimates, and Their Standard Errors, of Scitec MAP-3 K-Shell 60-Second Readings.
60-SECOND READING
MEASURED AT
0.0 mg/cm2
0.5 mg/cm2
1 .0 mg/cm2
2.0 mg/cm2
CORRECTED?
No
Yes
No
No
Yes
Nn
Vac
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal
Wood
PRECISION'
(mg/cm2)
07
0.7
0.3
0.2
0.5
0.4
0.3
0.4
0.7
0.7
0.3
0.3
0.7
0.4
0.3
0.5
0.7
0.7
0.3
0.4
0.8
0.5
0.3
05
0.7
0.7
0.3
0.6
1.0
0.6
0.3
0.6
STANDARD ERROR
FOR PRECISION
0.09
0.09
0.05
0.04
0.07
0.05
004
004
0.09
009
0.05
0.06
0.10
0.06
0.05
0.05
0.09
0.09
0.05
0 10
0.16
0.11
0.10
0.09
0.09
0.09
0.05
0 16
027
0.20
019
0.15
'Precision at 1 standard deviation
D-56
-------
APPENDIX D: XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
Table D-34. Classification Results For Scitec MAP-3 K-Shell 15-Second Readings (Metal and Wood
Corrected and Uncorrected), Classified Using the Inconclusive Ranges Reported in the XRF
Performance Characteristic Sheet and Compared to Laboratory Results in mg/cm2 Lead
Classified Using the 1.0 mg/cm2 Lead Federal Standard For Data Taken During the EPA/HUD
Field Study.
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal (Uncorrected)
Wood (Uncorrected)
TOTAL'
INCONCLUSIVE
RANGE
0.0 to 1.5
0.0 to 1.5
0.9 to 1.0
0.9 to 1.3
0.3 to 1.3
0.9 to 1.3
0.9 to 1.6
0.9 to 1.6
FALSE POSITIVE
RATE
1.4% (2/1 43)
2.9% (11/382)
3.5% (8/226)
1 .7% (5/290)
2.0% (8/392)
3.2% (16/499)
1.7% (5/290)
4.0% (20/499)
2.6% (50/1 932)
FALSE NEGATIVE
RATE
0.0% (0/42)
1.9% (3/54)
(0/0)
1.1% (1/88)
11.5% (6/52)
7.5% (15/199)
1 1%(1/88)
5.5% (11/1 99)
5.8% (25/435)
INCONCLUSIVE
RATE
20.5% (38/1 85)
29.6% (129/436)
0.9% (2/226)
6.1% (23/378)
15.5% (69/444)
4.6% (32/698)
16.9% (64/378)
8.5% (59/698)
12.4% (293/2367)
'Total results are for values reported in PCS.
D-57
-------
APPENDIX D: XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
Table D-35. Classification Results For Scitec MAP-3 K-Shell 60-Second Readings (Metal and Wood
Corrected and Uncorrected), Classified Using the Inconclusive Ranges and Threshold Value
Reported in the XRF Performance Characteristic Sheet and Compared to Laboratory Results in
mg/cm2 Lead Classified Using the 1.0 mg/cm2 Lead Federal Standard For Data Taken During
the EPA/HUD Field Study.
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Metal (Uncorrected)
Wood (Uncorrected)
TOTAL"
INCONCLUSIVE
RANGE
OR THRESHOLD
0.3 to 0.9
0.3 to 0.9
0.6 to 0.8
0.9 to 1 2
0.2 to 0.9
1.0
0.9 to 1.6
0.9 to 1.3
FALSE POSITIVE
RATE
5.9% (2/34)
5.8% (6/1 03)
5.0% (3/60)
2.7% (2/74)
0.0% (0/1 07)
1.9% (2/1 06)
2.7% (2/74)
1.9% (2/1 07)
3 1% (15/484)
FALSE NEGATIVE
RATE
0.0% (0/8)
7.1% (1/14)
(0/0)
3.6% (1/28)
5.9% (1/17)
5.8% (3/52)
0.0% (0/28)
0.0% (0/51)
5.0% (6/1 19)
INCONCLUSIVE
RATE
4.8% (1/42)
12.0% (14/1 17)
0.0% (0/60)
5.9% (6/1 02)
8.1% (10/124)
0.0% (0/1 58)
15.7% (16/1 02)
7.0% (11/1 58)
5.1% (31/603)
'Total results are for values reported in PCS.
D-58
-------
APPENDIX D: XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
D.5 XRF Performance Characteristic Sheet for the Scitec MAP 4 and Related
Results
D-59
-------
APPENDIX D: XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
Scitec Corporation; MAP 4
EFFECTIVE DATE: June 26, 1996 EDITION NO.: 1
MANUFACTURER AND MODEL:
Make: Scitec Corporation
Model: MAP 4
Source: Co57
Note: This sheet supersedes all previous sheets for the XRF Instrument of the make, model, and
source shown above.
EVALUATION DATA SOURCE AND DATE:
This sheet is supplemental information to be used in conjunction with Chapter 7 of the HUD Guidelines
for the Evaluation and Control of Lead-Based Paint Hazards in Housing ("HUD Guidelines"). Performance
parameters shown on this sheet are calculated from an EPA/HUD evaluation using archived building
components. Testing was conducted on approximately 150 test locations. All of the test locations were tested
in February 1996 using two different instruments. One instrument had a new source installed in July 1994
and its strength at the time of testing was calculated as 10.1 mCi. The other instrument had a new source
installed in September 1994 and its strength at the time of testing was calculated as 11.5 mCi.
FIELD OPERATION GUIDANCE
OPERATING PARAMETERS:
Performance parameters shown in this sheet are applicable only when operating the instrument under
the same conditions as the evaluation testing and using the procedures described in Chapter 7 of the HUD
Guidelines. Operating parameters include:
• Manufacturer-recommended warm-up and quality control procedures.
• Use the Multifamily Decision Flowchart for determining the presence of lead on a component type in
multifamily housing.
• Take readings on three locations per component for single-family housing and one location per
component for multifamily housing.
• Calibration checks are taken in TEST mode while using the red (1.02 mg/cm2) NIST Standard
Reference Material (SRM No. 2579) paint film.
• Readings for determining the substrate correction values are taken on bare substrate covered with
red (1.02 mg/cm2) NIST SRM paint film.
• Lead-based paint is defined as paint with lead equal to or in excess of 1.0 mg/cm2.
1 of 8
Figure D-4. XRF Performance Characteristic Sheet for Scitec MAP 4.
D-60
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
Scitec Corporation; MAP 4
XRF CALIBRATION CHECK:
Chapter 7 of the HUD Guidelines recommends using a calibration check procedure to determine the
operating condition of the XRF instrument. For this instrument, calibration checks should be taken in TEST
mode. If the observed calibration check average minus 1.02 mg/cm2 is greater than the positive (plus)
calibration check tolerance value, or less than the negative (minus) calibration check tolerance value, then
the instructions provided by the manufacturer should be followed in order to bring the instrument back into
control before any more XRF testing is done. This calibration check is estimated to produce an incorrect result
(that is, a finding that the instrument is out of calibration) very infrequently - once out of every 200 times this
procedure is followed.
minus value = -0.4 mg/cm2
plus value = +0.2 mg/cm2
WHEN USING UNLIMITED MODE, SUBSTRATE CORRECTION RECOMMENDED FOR:
None
WHEN USING UNLIMITED MODE, SUBSTRATE CORRECTION NOT RECOMMENDED FOR:
Brick, Concrete, Drywall, Metal, Plaster, and Wood
WHEN USING SCREEN OR TEST MODE, FOR XRF RESULTS BELOW 4.0 mg/cm2, SUBSTRATE
CORRECTION RECOMMENDED FOR:
Drywall, Metal, and Wood
WHEN USING SCREEN OR TEST MODE, SUBSTRATE CORRECTION NOT RECOMMENDED FOR:
Brick, Concrete, and Plaster
SUBSTRATE CORRECTION VALUE COMPUTATION:
Chapter 7 of the HUD Guidelines provides guidance on correcting XRF results for substrate bias.
Supplemental guidance for using the red (1.02 mg/cm2) NIST SRM paint film for substrate correction is
provided below.
XRF results are corrected for substrate bias by subtracting from each XRF result a correction value
determined separately in each house for single-family housing or in each development for multifamily housing,
for each substrate. The correction value is an average of XRF readings taken over red NIST SRM (1.02
mg/cm2) paint films at test locations that had been scraped clean of their paint covering. Compute the
correction values as follows.
Using the same XRF instrument, take three readings on a bare substrate area covered with the red NIST
SRM (1.02 mg/cm2) paint film. Repeat this procedure by taking three more readings on a second bare
substrate area of the same substrate covered with the red NIST SRM (1.02 mg/cm2) paint film.
Compute the correction value for each substrate type by computing the average of all six readings as
shown below.
2 of 8
Figure D-4 continued. XRF Performance Characteristic Sheet for the Scitec Map 4.
D-61
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
Scitec Corporation; MAP 4
For each substrate type:
Correctfon
Value
\ 1" + 2"" + 3rd + 4th *• 5'" * 6m Reading , „ . 2
> = - 1.02mg/cm*
i 6
Repeat this procedure for each substrate tested in the house or housing development.
INCONCLUSIVE RANGE OR THRESHOLD.
XRF results are classified using either the threshold or the inconclusive range. In single-family housing,
an XRF result is the average of three readings taken on a testing combination. (A testing combination is a
location on a painted surface as defined in Chapter 7 of the HUD Guidelines). In multifamily housing, an XRF
result is a single reading taken on a testing combination. For computing the XRF result, use all digits that are
reported by the instrument. For the threshold, results are classified as positive if they are greater than or
equal to the threshold, and negative if they are less than the threshold. There is no inconclusive classification
when using the threshold. For the inconclusive range, results are classified as positive if they are greater than
or equal to the upper limit of the inconclusive range, and negative if they are less than or equal to the lower
limit of the inconclusive range. Thresholds and inconclusive ranges were determined for comparing results
to the 1.0 mg/cm2 standard. For a listing of laboratories recommended by the EPA National Lead Laboratory
Accreditation Program (NLLAP) for the analysis of samples to resolve an inconclusive XRF result or additional
confirmational analysis, call the National Lead Information Center Clearinghouse on 1-800-424-LEAD.
UNLIMITED MODE READING DESCRIPTION
Results not corrected for substrate bias for
unlimited mode readings
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
INCONCLUSIVE
RANGE (mg/cm2)
0.9 to 1.2
0.9 to 1.2
0.9 to 1.2
0.9 to 1.2
0.9 to 1.2
0.9 to 1.2
SCREEN MODE READING DESCRIPTION
Results corrected for substrate bias for screen
mode readings on drywall, metal, and wood
substrates only
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
INCONCLUSIVE
RANGE (mg/cm2)
0.9 to 1.1
0.9 to 1.1
0.9 to 1.4
0.9 to 1.2
0.9 to 1.1
0.9 to 1.3
3 of 8
Figure D-4 continued. XRF Performance Characteristic Sheet for the Scitec Map 4.
D-62
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
Scitec Corporation; MAP 4
TEST MODE READING
DESCRIPTION
Readings corrected for substrate
bias for test mode readings on
drywall, metal, and wood
substrates only
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
THRESHOLD
(mg/cm2)
0.9
0.9
None
None
0.9
None
INCONCLUSIVE RANGE
(mg/cm2)
None
None
0.9 to 1.4
0.9 to 1.1
None
0.9 to 1.3
INSTRUCTIONS FOR EVALUATING XRF TESTING:
Chapter 7 of the HUD Guidelines recommends several options for evaluating XRF testing. Among those
options is the following procedure which may be used after XRF testing has been completed. In single-family
housing, an XRF result is the average of three readings taken on a testing combination. (A testing
combination is a location on a painted surface as defined in Chapter 7 of the HUD Guidelines). In multifamily
housing, an XRF result is a single reading taken on a testing combination. If a multifamily housing
development is being retested, randomly select two units from within the development from which the ten
testing combinations should be randomly selected. Use either 15-second readings or 60-second readings.
Randomly select ten testing combinations for retesting from each house or from the two selected units.
Conduct XRF retesting at the ten testing combinations selected for retesting.
Determine if the XRF testing in the units or house passed or failed the test by applying the steps below.
Compute the Retest Tolerance Limit by the following steps:
Determine XRF results for the original and retest XRF readings. Do not correct the original or
retest results for substrate bias. In single-family housing a result is defined as the average of
three readings. In multifamily housing, a result is a single reading. Therefore, there will be ten
original and ten retest XRF results for each house or for the two selected units.
Compute the average of the original and re-test result for each of the ten testing combinations.
Square the average for each testing combination.
Add the ten squared averages together. Call this quantity C.
Multiply the number C by 0.0072. Call this quantity D.
Add the number 0.032 to D. Call this quantity E.
Take the square root of E. Call this quantity F.
Multiply F by 1.645. The result is the Retest Tolerance Limit.
4 of 8
Figure D-4 continued. XRF Performance Characteristic Sheet for the Scitec Map 4.
D-63
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
Scitec Corporation; MAP 4
Compute the overall average of all ten retest XRF results over all ten testing combinations selected
for retesting.
Take the difference of the overall average of the ten original XRF results and the overall average of
the ten retest XRF results. If the difference is negative, drop the negative sign.
If the difference of the overall averages is less than the Retest Tolerance Limit, the inspection has
passed the retest. If the difference of the overall averages equals or exceeds the Retest Tolerance
Limit, this procedure should be repeated with ten new testing combinations. If the difference of the
overall averages is equal to or greater than the Retest Tolerance Limit a second time, then the
inspection should be considered deficient.
Use of this procedure is estimated to produce a spurious result approximately 1% of the time. That is, results
of this procedure will call for further examination when no examination is warranted in approximately 1 out of
100 dwelling units tested.
TESTING TIMES
For screen, test, and confirm modes, the MAP 4 instrument tests until a K-shell result is obtained relative
to a level of precision. A result is "positive", "negative" or "retest" as displayed by indicator lights. For the
unlimited mode, the MAP 4 instrument tests until a K-shell result is indicated relative to an action level (1.0
mg/cm2 for archive testing) and the current precision, or until the reading is terminated by releasing the trigger.
A few unlimited mode readings were terminated because they exceeded the two-minute limit used for archive
testing. The following tables provide testing time information for three testing modes. Insufficient information
is available to provide this information for confirm mode. All times have been scaled to match an initial 12 mCi
source. Note that source strength and factors such as substrate may affect testing times.
UNLIMITED MODE TESTING TIMES (Seconds)
SUBSTRATE"
Wood
Drywall
Metal
Brick
Concrete
Plaster
ALL DATA
25th
Percentile
3
3
4
Median
4
4
5
75th
Percentile
6
8
8
MEDIAN FOR LABORATORY-MEASURED
LEAD LEVELS (mg/cm2)
Pb < 0.25
4
4
6
0.25 ±Pb < 1.0
13
9
6
1 .0 s Pb
3
3
3
The general calibration was used for wood, drywall, brick, concrete, plaster. Steel
calibration was used for metal. (There are no aluminum samples in the archive facility).
5 of 8
Figure D-4 continued. XRF Performance Characteristic Sheet for the Scitec Map 4.
D-64
-------
APPENDIX D: XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
Scitec Corporation; MAP 4
TESTING TIMES (continued):
SCREEN MODE TESTING TIMES (Seconds)
SUBSTRATE"
Wood
Dry wall
Metal
Brick
Concrete
Plaster
ALL DATA
25th
Percentile
4
4
11
Median
6
5
11
75th
Percentile
7
6
13
MEDIAN FOR LABORATORY-MEASURED
LEAD LEVELS (mg/cm2)
Pb < 0.25
5
5
11
0.25 s Pb < 1.0
6
5
11
1.0 s Pb
7
5
11
The general calibration was used for wood, drywall, brick, concrete, plaster. Steel
calibration was used for metal. (There are no aluminum samples in the archive facility).
TEST MODE TESTING TIMES (Seconds)
SUBSTRATE"
Wood
Drywall
Metal
Brick
Concrete
Plaster
ALL DATA
25th
Percentile
17
13
41
Median
22
20
42
75th
Percentile
27
23
52
MEDIAN FOR LABORATORY-MEASURED
LEAD LEVELS (mg/cm2)
Pb < 0.25
21
20
41
0.25 s Pb < 1.0
20
20
46
1 .0 s Pb
28
20
43
The general calibration was used for wood, drywall, brick, concrete, plaster. Steel
calibration was used for metal. (There are no aluminum samples in the archive facility).
6 of 8
Figure D-4 continued. XRF Performance Characteristic Sheet for the Scitec Map 4.
D-65
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
Scitec Corporation; MAP 4
BIAS AND PRECISION:
Do not use these bias and precision data to correct for substrate bias. These bias and precision data
were computed without substrate correction from samples with laboratory-measured lead levels less than 4 0
mg/cm2 lead. There were 15 testing locations taken in the screen mode with laboratory-measured lead
levels equal to or greater than 4.0 mg/cm2 lead. None of these had XRF readings less than 1.0 mg/cm2.
There were 15 testing locations taken in the test mode with laboratory-measured lead levels equal to or
greater than 4.0 mg/cm2 lead. None of these had XRF readings less than 1.0 mg/cm2. There were not any
testing locations taken in the confirm mode with a laboratory-measured lead levels equal to or greater than
4.0 mg/cm2 lead. There were 15 testing locations taken in the unlimited mode with laboratory-measured
lead levels equal to or greater than 4.0 mg/cm2 lead. None of these had XRF readings less than 1.0 mg/cm2
All testing was done in February 1996 with two different instruments. The following data are for illustrative
purposes only. Actual bias must be determined on the site. Inconclusive ranges provided above already
account for bias and precision. Units are in mg/cm2.
SCREEN MODE
READING MEASURED AT
0.0 mg/cm2
0.5 mg/cm2
1 .0 mg/cm2
2.0 mg/cm2
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
BIAS
(mg/cm2)
-0.1
-0.1
0.1
0.1
-01
0.0
0.0
0.0
0.3
0.2
0.0
02
0.1
0.1
0.5
0.3
0.1
0.4
0.4
0.4
0.9
05
0.4
0.7
PRECISION*
(mg/cm2)
0.3
0.3
0.2
0.3
0.3
0.2
0.3
0.3
0.4
0.3
0.3
0.4
0.4
0.4
0.6
0.3
0.4
0.6
0.5
0.5
0.8
0.3
0.5
0.8
'Precision at 1 standard deviation
7 of 8
Figure D-4 continued. XRF Performance Characteristic Sheet for the Scitec Map 4.
D-66
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
Scitec Corporation; MAP 4
BIAS AND PRECISION (continued):
t
1
TEST MODE
READING MEASURED AT
0.0 mg/cm2
0.5 mg/cm2
1 .0 mg/cm2
2.0 mg/cm2
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
BIAS
(mg/cm2)
-0.
-0.
0.
0.
-0.
0.0
-0.1
-0.1
0.3
0.2
-0.1
0.2
-0.1
-0.1
0.5
0.3
-0.1
0.4
0.0
0.0
1.0
0.5
0.0
0.8
PRECISION*
(mg/cm2)
0.2
0.2
0.1
0.2
0.2
0.1
0.3
0.3
0.4
0.2
0.3
0.4
0.3
0.3
0.6
0.2
0.3
0.6
0.4
0.4
0.8
0.2
0.4
0.8
'Precision at 1 standard deviation
A document titled Methodology for XRF Performance Characteristic Sheets provides an explanation of
he statistical methodology used to construct the data in the sheets and provides empirical results from using
he recommended inconclusive ranges or thresholds for specific XRF instruments. For a copy of this
document call the National Lead Information Center Clearinghouse at 1-800-424-LEAD.
This XRF Performance Characteristic Sheet is a joint product of the U S Environmental Protection Agency (EPA) and the U S
Department of Housing and Urban Development (HUD). The issuance of this sheet does not constitute rulemaking The
information provided here is intended solely as guidance to be used in conjunction with Chapter 7 of the Guidelines for the
Evaluation and Control of Lead-Based Paint Hazards in Housing. EPA and HUD reserve the right to revise this guidance
Please address questions and comments on this sheet to. Director. Office of Lead-Based Paint Abatement and Poisoning
Prevention, U.S. Department of Housing and Urban Development, Room B-133, 451 Seventh St, S.W., Washington, DC 20410
8 of 8
Figure D-4 continued.
XRF Performance Characteristic Sheet for the Scitec Map 4.
D-67
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
Table D-36. Bias Estimates, and Their Standard Errors, of Scitec MAP 4 K-Shell SCREEN Mode Readings.
15-SECOND SCREEN MODE
READING MEASURED AT
0.0 mg/cm2
0.5 mg/cm2
1 .0 mg/cm2
2.0 mg/cm2
CORRECTED?
No
INU
Yes
Nn
MU
Yes
Nn
MU
Yes
No
I1U
Yes
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Drywall
Metal
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Drywall
Metal
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Drywall
Metal
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Drywall
Metal
Wood
BIAS (mg/cm2)
-0.1
-0.1
0.1
0.1
-0.1
0.0
0.0
0.0
-0.1
0.0
00
0.3
0.2
0.0
0.2
0.1
0.1
0.0
0.1
0.1
0.5
0.3
0.1
0.4
0.3
0.2
0.2
0.4
0.4
0.9
0.5
0.4
0.7
0.7
0.3
0.6
STANDARD
ERROR
0.08
0.08
0.07
0.06
0.08
0.05
0.07
0.07
0.07
0.07
0.07
0.09
0.05
0.07
0.06
0.09
0.06
0.07
0.13
0.13
0.14
0.08
0.13
0.11
0.14
0.09
0.12
0.30
0.30
0.26
0.16
0.30
0.24
026
0.17
0.24
D-68
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
Table D-37. Bias Estimates, and Their Standard Errors, of Scitec MAP 4 K-Shell TEST Mode Readings.
15-SECOND TEST MODE
READING MEASURED AT
0.0 mg/cm2
0.5 mg/cm2
1.0 mg/cm2
2.0 mg/cm2
CORRECTED?
Mrt
INU
Yes
No
Yes
No
\W
Yes
No
Yes
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Drywall
Metal
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Drywall
Metal
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Drywall
Metal
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Drywall
Metal
Wood
BIAS (mg/cm2)
-0.1
-0.1
0.1
01
-0.1
00
0.0
-01
-0.2
-0.1
-0.1
03
02
-0.1
0.2
0.2
0.0
0.0
-0.1
-0.1
0.5
0.3
-0.1
0.4
0.4
0.1
0.2
0.0
0.0
1.0
0.5
0.0
0.8
0.8
03
0.6
STANDARD
ERROR
0.06
0.06
0.05
0.06
0.06
0.04
0.06
0.07
0.06
0.05
005
0.07
0.05
0.05
0.06
0.08
0.06
0.07
0.11
0.11
0.13
0.08
0.11
0.12
013
009
012
024
0.24
0.25
0.17
0.24
0.24
0.25
0.17
0.24
D-69
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
Table D-38. Precision Estimates, and Their Standard Errors, of Scitec MAP 4 K-Shell SCREEN Mode
Readings.
15-SECOND SCREEN MODE
READING MEASURED AT
0.0 mg/cm2
0.5 mg/cm2
1 .0 mg/cm2
2.0 mg/cm2
CORRECTED?
Nn
n\J
Yes
Nn
n\j
Yes
No
Yes
No
I1U
Yes
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Drywall
Metal
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Drywall
Metal
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Drywall
Metal
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Drywall
Metal
Wood
PRECISION*
(mg/cm2)
0.3
0.3
0.2
0.3
0.3
0.2
0.2
0.3
0.2
0.3
0.3
0.4
0.3
0.3
0.4
0.4
0.3
0.4
0.4
0.4
0.6
0.3
0.4
0.6
0.6
0.3
0.6
0.5
0.5
0.8
0.3
0.5
0.8
0.8
0.3
0.8
STANDARD
ERROR
0.05
0.05
0.04
0.04
0.05
0.04
0.04
003
0.04
0.06
0.06
0.07
0.04
0.06
0.07
0.07
0.03
0.07
0.10
0.10
0.10
0.04
0.10
0.10
0.10
003
0.10
0.16
0.16
0.15
0.04
0.16
0.15
0.15
0.03
0.15
'Precision at 1 standard deviation.
D-70
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
Table D-39. Precision Estimates, and Their Standard Errors, of Scitec MAP 4 K-Shell TEST Mode Readings.
15-SECOND TEST MODE
READING MEASURED AT
0.0 mg/cm2
0.5 mg/cm2
1 .0 mg/cm2
2.0 mg/cm2
CORRECTED?
No
n\J
Yes
No
Yes
Nn
I1U
Yes
No
Yes
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Drywall
Metal
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Drywall
Metal
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Drywall
Metal
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Drywall
Metal
Wood
PRECISION*
(mg/cm2)
0.2
0.2
0.1
0.2
0.2
0.1
0.1
0.3
0.1
0.3
0.3
0.4
0.2
0.3
0.4
0.4
0.3
0.4
0.3
0.3
0.6
0.2
0.3
0.6
0.6
0.3
0.6
0.4
0.4
0.8
0.2
0.4
0.8
0.8
0.3
0.8
STANDARD
ERROR
0.04
004
003
0.04
0.04
0.03
0.3
0.4
0.3
0.05
0.05
0.07
0.04
0.05
0.07
0.6
0.4
0.6
0.08
0.08
0.10
0.04
0.08
0.10
0.10
004
0.10
0.12
0.12
0.14
004
0.12
0.14
0.14
0.04
0.14
'Precision at 1 standard deviation.
D-71
-------
APPENDIX D: XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
Table D-40. Classification Results For Scitec MAP 4 K-Shell UNLIMITED Mode Readings (Uncorrected),
Classified Using the Inconclusive Ranges Reported in the XRF Performance Characteristic Sheet
and Compared to Laboratory-Measured Lead Levels in mg/cm2 Classified Using the 1.0 mg/cm2
Lead Federal Standard For Data Taken From Testing Archived Building Components in February
1996.
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
TOTAL
INCONCLUSIVE
RANGE
0.9 to 1.2
0.9 to 1.2
0.9 to 1.2
0.9 to 1.2
0.9 to 1.2
0.9 to 1.2
FALSE POSITIVE
RATE
0.0% (0/2)
0.0% (0/4)
3.6% (1/28)
3.7% (2/54)
3.4% (2/58)
1 1 .3% (9/80)
6.2% (14/226)
FALSE NEGATIVE
RATE
0.0% (0/4)
(0/0)
(0/0)
0.0% (0/22)
0.0% (0/1 8)
0.0% (0/46)
0.0% (0/90)
INCONCLUSIVE
RATE
0.0% (0/6)
0.0% (0/4)
14.3% (4/28)
5.3% (4/76)
1.3% (1/76)
5.6% (7/126)
5.1% (16/316)
D-72
-------
APPENDIX D: XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
Table D-41. Classification Results For Scitec MAP 4 K-Shell SCREEN Mode Readings (Drywall, Metal, and
Wood Corrected and Uncorrected), Classified Using the Inconclusive Ranges Reported in the
XRF Performance Characteristic Sheet and Compared to Laboratory-Measured Lead Levels in
mg/cm2 Classified Using the 1.0 mg/cm2 Lead Federal Standard For Data Taken From Testing
Archived Building Components in February 1996.
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Drywall (Uncorrected)
Metal (Uncorrected)
Wood (Uncorrected)
TOTAL*
INCONCLUSIVE
RANGE
0.9 to 1.1
0.9 to 1.1
0.9 to 1.4
0.9 to 1.2
0.9 to 1.1
0.9 to 1.3
0.9 to 1.5
0.9 to 1.3
0.9 to 1.4
FALSE POSITIVE
RATE
0.0% (0/2)
0.0% (0/4)
0.0% (0/28)
1.9% (1/54)
3.4% (2/58)
10.0% (8/80)
0.0% (0/28)
1.9% (1/54)
10.0% (8/80)
4.9% (11/226)
FALSE NEGATIVE
RATE
0.0% (0/4)
(0/0)
(0/0)
0.0% (0/22)
0.0% (0/18)
0.0% (0/46)
(0/0)
0.0% (0/22)
0.0% (0/46)
0.0% (0/90)
INCONCLUSIVE
RATE
0.0% (0/6)
0.0% (0/4)
7.1% (2/28)
3.9% (3/76)
0.0% (0/76)
4.8% (6/126)
17.9% (5/28)
7.9% (6/76)
7.9% (10/126)
3.5% (11/316)
'Total results are for values reported in PCS.
D-73
-------
APPENDIX D: XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
Table D-42. Classification Results For Scitec MAP 4 K-Shell TEST Mode Readings (Drywall, Metal, and
Wood Corrected), Classified Using the Inconclusive Ranges and Threshold Value Reported in
the XRF Performance Characteristic Sheet and Compared to Laboratory-Measured Lead Levels
in mg/cm2 Classified Using the 1.0 mg/cm2 Lead Federal Standard For Data Taken From Testing
Archived Building Components in February 1996.
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Drywall (Uncorrected)
Metal (Uncorrected)
Wood (Uncorrected)
TOTAL*
INCONCLUSIVE
RANGE
OR THRESHOLD
0.9
0.9
0.9 to 1.4
0.9 to 1.1
0.9
0.9 to 1.3
0.9 to 1.5
0.9 to 1.3
0.9 to 1.4
FALSE POSITIVE
RATE
0.0% (0/2)
0.0% (0/4)
0.0% (0/28)
1.9% (1/54)
3.4% (2/58)
8.8% (7/80)
0.0% (0/28)
1.9% (1/54)
8.8% (7/80)
4.4% (10/226)
FALSE NEGATIVE
RATE
0.0% (0/4)
(0/0)
(0/0)
0.0% (0/22)
0.0% (0/18)
2.2% (1/46)
(0/0)
0.0% (0/22)
0.0% (0/46)
1.1% (1/90)
INCONCLUSIVE
RATE
0.0% (0/6)
0.0% (0/4)
10.7% (3/28)
1.3% (1/76)
0.0% (0/76)
7.9% (10/126)
17.9% (5/28)
5.3% (4/76)
8.7% (11/126)
4.4% (14/316)
"Total results are for values reported in PCS.
D-74
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
Table D-43. Misclassification of Indicated Positive and Negative Results and Retest Percentages For Scitec
MAP 4 K-Shell Readings (Uncorrected), Taken in UNLIMITED, SCREEN, and TEST Modes and
SCREEN-TEST-CONFIRM Mode Sequence Compared to Laboratory-Measured Lead Levels in
mg/cm2 Classified Using the 1.0 mg/cm2 Lead Federal Standard For Data Taken From Testing
Archived Building Components in February 1996 Obtained From Archive Testing Without
Applying Substrate Correction Procedures, Inconclusive Ranges, or Thresholds. (These results
can be compared to the corresponding percentages in Tables D-40, D-41, and D-42).
TESTING MODE
Unlimited mode
Screen mode
Test mode
Screen-Test-Confirm mode
sequence"
FALSE
POSITIVE
PERCENTAGE
7.1%
5.1%
6.6%
7.1%
FALSE
NEGATIVE
PERCENTAGE
0.0%
0.0%
0.0%
0.0%
RETEST
PERCENTAGE
3.8%
15.8%
7.0%
3.2%
"As indicated by the Operations Manual for the instrument: First, a screen mode reading was taken. If the result of the
screen mode result was positive or negative, testing stopped and the screen mode result was the final reading. Otherwise,
a test mode reading was taken. If the test mode result was positive or negative, then testing stopped, and the test mode
result was the final reading. Otherwise, a confirm mode reading was taken. If the confirm mode result was positive or
negative, the confirm mode result was the final reading. If the confirm mode result was indicated as "RETEST", the testing
was terminated, and the final result was designated as "RETEST".
D-75
-------
APPENDIX D: XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
D.6 XRF Performance Characteristic Sheet for the Warrington Microlead I Revision
4 and Related Results
D-76
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
EFFECTIVE DATE:
XRF PERFORMANCE CHARACTERISTIC SHEET
Warrington, Inc.; Microlead I revision 4
September 25, 1995
EDITION NO.: 1
MANUFACTURER AND MODEL:
Make: Warrington, Inc.
Model: Microlead I revision 4
Source: Co"
Note: This sheet supersedes all previous sheets for the XRF instrument of the make, model, and
source shown above
EVALUATION DATA SOURCE AND DATE:
This sheet is supplemental information to be used in conjunction with Chapter 7 of the HUD Guidelines for
the Evaluation and Control of Lead-Based Paint Hazards in Housing ("HUD Guidelines"). Performance
parameters shown on this sheet are calculated from evaluation data collected during the EPA/HUD field
evaluation study conducted from March through October 1993. The data were collected from approximately
1,200 test locations using five instruments with source dates ranging from March 1993 to October 1993. All
five instruments had sources with 10 mCi initial strengths. The results of this study are reported in A Field
Test of Lead-Based Paint Testing Technologies: Technical Report, EPA 747-R-95-002b, May 1995.
FIELD OPERATION GUIDANCE
OPERATING PARAMETERS:
Performance parameters shown in this sheet are applicable only when operating the instrument under
the same conditions as the evaluation testing and using the procedures described in Chapter 7 of the HUD
Guidelines. Operating parameters include:
• Manufacturer-recommended warm-up and quality control procedures
• Use the Multifamily Decision Flowchart for determining the presence of lead on a component type in
multifamily housing
• Nominal 15-second readings on three locations per component for single-family housing and one
location per component for multifamily housing
• Calibration checks are taken using the red (1 02 mg/cm2) NIST Standard Reference Material (SRM
No. 2579) paint film
• Readings for determining the substrate correction values are taken on bare substrate covered with
red (1.02 mg/cm2) NIST SRM paint film
• Lead-based paint is defined as paint with lead equal to or in excess of 1.0 mg/cm2.
1 of 5
Figure D-5. XRF Performance Characteristic Sheet for the Warrington Microlead
revision 4.
D-77
-------
APPENDIX D: XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
Warrington, Inc.; Microlead I revision 4
XRF CALIBRATION CHECK:
Chapter 7 of the HUD Guidelines recommends using a calibration check procedure to determine the
operating condition of the XRF instrument. If the observed calibration check average minus 1.02 mg/cm2 is
greater than the positive (plus) calibration check tolerance value, or less than the negative (minus) calibration
check tolerance value, then the instructions provided by the manufacturer should be followed in order to bring
the instrument back into control before any more XRF testing is done. This calibration check is estimated to
produce an incorrect result (that is, a finding that the instrument is out of calibration) very infrequently - once
out of every 200 times this procedure is followed.
minus value = -0.6 mg/cm2
plus value = +0.6 mg/cm2
FOR XRF RESULTS BELOW 4.0 mg/cm2, SUBSTRATE CORRECTION RECOMMENDED FOR:
Brick, Concrete, Drywail, Metal, and Wood.
SUBSTRATE CORRECTION NOT RECOMMENDED FOR:
Plaster.
SUBSTRATE CORRECTION VALUE COMPUTATION:
Chapter 7 of the HUD Guidelines provides guidance on correcting XRF results for substrate bias.
Supplemental guidance for using the red (1.02 mg/cm2) NIST SRM paint film for substrate correction is
provided below.
XRF results are corrected for substrate bias by subtracting from each XRF result a correction value
determined separately in each house for single-family housing or in each development for multifamily housing,
for each substrate. The correction value is an average of XRF readings taken over red NIST SRM (1.02
mg/cm2) paint films at test locations that had been scraped clean of their paint covering. Compute the
correction values as follows:
• Using the same XRF instrument, take three readings on a bare substrate area covered with the red NIST
SRM (1.02 mg/cm2) paint film. Repeat this procedure by taking three more readings on a second bare
substrate area of the same substrate covered with the red NIST SRM (1.02 mg/cm2) paint film.
Compute the correction value for each substrate type by computing the average of all six readings as
shown below.
For each substrate type:
'orrectit
Value
Correction \ f" + 2"" + 3" + 4th •>• 5th + 6lh Reading , „„ . ,
l = ° - i.o2tng/cm2
Repeat this procedure for each substrate tested in the house or housing development.
2 of 5
Figure D-5 continued. XRF Performance Characteristic Sheet for the Warrington
Microlead I revision 4.
D-78
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
Warrington, Inc.; Microlead I revision 4
INCONCLUSIVE RANGE:
XRF results are classified as positive if they are greater than or equal to the upper limit of the inconclusive
range, and negative if they are less than or equal to the lower limit of the inconclusive range. In single-family
housing, an XRF result is the average of three readings taken on a testing combination. (A testing
combination is a location on a painted surface as defined in Chapter 7 of the HUD Guidelines). In multifamily
housing, an XRF result is a single reading taken on a testing combination. For computing the XRF result, use
all digits that are reported by the instrument. Inconclusive ranges were determined for comparing results to
the 1.0 mg/cm2 standard. For a listing of laboratories recommended by the EPA National Lead Laboratory
Accreditation Program (NLLAP) for the analysis of samples to resolve an inconclusive XRF result or additional
confirmational analysis, call the National Lead Information Center Clearinghouse at 1-800-424-LEAD.
DESCRIPTION
Results corrected for substrate bias on all substrates
except plaster
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
INCONCLUSIVE RANGE
in mg/cm2
0.8 to 1.3
0.5 to 1.4
0.9 to 1.1
0.9 to 1.4
0.7 to 1 .6
0.9 to 1.6
INSTRUCTIONS FOR EVALUATING XRF TESTING:
Chapter 7 of the HUD Guidelines recommends several options for evaluating XRF testing. Among those
options is the following procedure which may be used after XRF testing has been completed. In single-family
housing, an XRF result is the average of three readings taken on a testing combination. (A testing
combination is a location on a painted surface as defined in Chapter 7 of the HUD Guidelines). In multifamily
housing, an XRF result is a single reading taken on a testing combination. If a multifamily housing
development is being retested, randomly select two units from within the development from which the ten
testing combinations should be randomly selected.
Randomly select ten testing combinations for retesting from each house or from the two selected units.
Conduct XRF retesting at the ten testing combinations selected for retesting.
Determine if the XRF testing in the units or house passed or failed the test by applying the steps below.
Compute the Retest Tolerance Limit by the following steps:
Determine XRF results for the original and retest XRF readings. Do not correct the original or
retest results for substrate bias. In single-family housing a result is defined as the average of
three readings. In multifamily housing, a result is a single reading. Therefore, there will be ten
original and ten retest XRF results for each house or for the two selected units.
Compute the average of the original and re-test result for each of the ten testing combinations.
3 of 5
Figure D-5 continued.
XRF Performance Characteristic Sheet for the Warrington
Microlead I revision 4
D-79
-------
APPENDIX D: XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
Warrington, Inc.; Microlead I revision 4
Square the average for each testing combination.
Add the ten squared averages together. Call this quantity C.
Multiply the number C by 0.0072. Call this quantity D.
Add the number 0.032 to D. Call this quantity E.
Take the square root of E. Call this quantity F.
Multiply F by 1.645. The result is the Retest Tolerance Limit.
Compute the overall average of all ten retest XRF results over all ten testing combinations selected
for retesting.
Take the difference of the overall average of the ten original XRF results and the overall average of
the ten retest XRF results. If the difference is negative, drop the negative sign.
If the difference of the overall averages is less than the Retest Tolerance Limit, the inspection has
passed the retest. If the difference of the overall averages equals or exceeds the Retest Tolerance
Limit, this procedure should be repeated with ten new testing combinations. If the difference of the
overall averages is equal to or greater than the Retest Tolerance Limit a second time, then the
inspection should be considered deficient.
Use of this procedure is estimated to produce a spurious result approximately 1% of the time. That is, results
of this procedure will call for further examination when no examination is warranted in approximately 1 out of
100 dwelling units tested.
BIAS AND PRECISION:
Do not use these bias and precision data to correct for substrate bias. These bias and precision data
were computed without substrate correction from samples with reported laboratory results less than 4.0
mg/cm2 lead. There were 143 test locations with a laboratory reported result equal to or greater than 4.0
mg/cm2 lead. Of these, 1 had an XRF reading less than 1.0 mg/cm2. These data are for illustrative purposes
only. Actual bias must be determined on the site. Inconclusive ranges provided above already account for
bias and precision. Bias and precision ranges are provided whenever significant variability was found
between machines of the same model. Units are in mg/cm2.
4 of 5
Figure D-5 continued. XRF Performance Characteristic Sheet for the Warrington
Microlead I revision 4.
D-80
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
Warrington, Inc.; Microlead 1 revision 4
MEASURED
AT
0.0 mg/cm2
0.5 mg/cm2
1 .0 mg/cm2
2.0 mg/cm2
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
BIAS
(mg/cm2)
0.1
0.3
0.0
-0.3
0.1
0.4
0.4
0.3
0.1
-0.2
0.1
0.7
-0.3
0.3
0.2
-0.1
0.1
1.0
0.7
0.2
0.4
0.2
0.2
1.6
BIAS RANGE
(mg/cm2)
( 0.0, 0.9)
( 0.0, 0.7)
(-0.4,1.1)
(-0.3, 0.2)
( 0.0, 0.5)
(0.1,1.1)
(0.1,1.3)
(-0.3, 1.2)
(-0.3, 0.1)
( 0.2, 0.7)
(0.2,1.2)
(0.2,1.9)
(-0.1, 1.4)
(-0.3, 0.3)
(0.3,1.0)
(0.2, 1.5)
(0.1,3.1)
(0.1,1.7)
(-0.3, 0.7)
(0.6,1.7)
PRECISION'
(mg/cm2)
0.6
0.6
0.3
0.5
0.5
0.6
0.5
0.6
0.3
0.6
0.6
0.7
0.6
0.7
0.3
0.6
0.7
0.7
0.5
0.8
0.3
0.7
0.9
0.8
PRECISION
RANGE (mg/cm2)
(0.5,1.2)
( 0.3, 0.5)
( 0.3, 0.8)
( 0.3, 0.6)
( 0.5, 0.8)
(0.5,1.3)
( 0.3, 0.5)
( 0.5, 0.8)
( 0.4, 0.8)
( 0.6, 0.8)
(0.6,1.4)
( 0.3, 0.5)
( 0.5, 0.8)
(0.5, 1.0)
( 0.6, 0.8)
(0.7,1.7)
( 0.3, 0.5)
( 0.5, 0.8)
(0.6,1.2)
( 0.7, 0.8)
'Precision at 1 standard deviation
A document titled Methodology for XRF Performance Characteristic Sheets provides an explanation of
he statistical methodology used to construct the data in the sheets and provides empirical results from using
he recommended inconclusive ranges or thresholds for specific XRF instruments. For a copy of this
document call the National Lead Information Center Clearinghouse at 1-800-424-LEAD
This XRF Performance Characteristic Sheet is a joint product of the U.S. Environmental Protection Agency (EPA) and the U.S
Department of Housing and Urban Development (HUD). The issuance of this sheet does not constitute rulemaking. The
information provided here is intended solely as guidance to be used in conjunction with Chapter 7 of the Guidelines for the
Evaluation ami Control of Lead-Based Paint Hazards in Housing. EPA and HUD reserve the right to revise this guidance.
Please address questions and comments on this sheet to: Director, Office of Lead-Based Paint Abatement and Poisoning
Prevention, U S. Department of Housing and Urban Development, Room B-133, 451 Seventh St, S.W , Washington, DC 20410
5 of 5
Figure D-5 continued.
XRF Performance Characteristic Sheet for the Warrington
Microlead I revision 4.
D-81
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
Table D-44. Inconclusive Ranges For Warrington Microlead I Revision 4 Readings Where Substrate
Correction Is Not Performed, But Substrate Correction Is Recommended.
DESCRIPTION
Results not corrected for substrate bias on any substrate
SUBSTRATE
Brick
Concrete
Drywall
Metal
Wood
INCONCLUSIVE RANGE
(mg/cm2)
0.9 to 1 7
0.9 to 1.8
0.9 to 1.2
0.9 to 1.3
0.9 to 2.3
D-82
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
Table D-45. Bias Estimates, and Their Standard Errors, of Warrington Microlead I Revision 4 Readings.
MEASURED AT
0 0 mg/cm2
0.5 mg/cm2
1 .0 mg/cm2
2.0 mg/cm2
CORRECTED?
No
Yes
No
Yes
No
Yes
No
Yes
SUBSTRATE
Brick'
Concrete
Drywall
Metal
Plaster
Wood
Brick"
Concrete
Drywall
Metal
Wood
Brick'
Concrete
Drywall
Metal
Plaster
Wood
Brick'
Concrete
Drywall
Metal
Wood
Brick'
Concrete
Drywall
Metal
Plaster
Wood
Brick"
Concrete
Drywall
Metal
Wood
Brick'
Concrete
Drywall
Metal
Plaster
Wood
Brick'
Concrete
Drywall
Metal
Wood
BIAS (mg/cm2)
0.1
0.3
0.0
-0.3
0.1
04
_t
-0.1
-0.1
-0.2
-0.3
0.4
0.3
01
-0.2
0.1
07
-0.2
-0.1
0.0
-0.1
0.0
-0.3
0.3
0.2
-0.1
0.1
1 0
-0.2
-0.2
0.1
0.1
0.2
0.7
0.2
0.4
0.2
02
1.6
-02
-02
03
0.4
0.7
STANDARD ERRORS FOR BIAS
0.06
0.05
0.06
0.06
0.07
0.05
0.05
0.06
0.07
0.05
0.11
0.05
0.05
0.06
0.05
0.11
0.05
0.06
0.09
0.23
0.07
0.11
0.09
0.10
0.23
0.07
0.08
0.20
0.47
0.14
0.25
018
0.21
0.48
0.15
0.17
'Nonparametric estimation used.
*A reliable estimate at 0.0 mg/cm2 could not be obtained.
D-83
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
Table D-46. Precision Estimates, and Their Standard Errors, of Warrington Microlead I Revision 4 Readings.
MEASURED AT
0.0 mg/cm2
0.5 mg/cm2
CORRECTED?
No
Mw
Yes
No
I1U
Yes
1 .0 mg/cm2
2.0 mg/cm2
No
Yes
Kin
nu
Yes
SUBSTRATE
Brick*
Concrete
Drywall
Metal
Plaster
Wood
Brick*
Concrete
Drywall
Metal
Wood
Brick*
Concrete
Drywall
Metal
Plaster
Wood
Brickt
Concrete
Drywall
Metal
Wood
Brick*
Concrete
Drywall
Metal
Plaster
Wood
Brick*
Concrete
Drywall
Metal
Wood
Brick*
Concrete
Drywall
Metal
Plaster
Wood
Brickt
Concrete
Drywall
Metal
Wood
PRECISION' (mg/cm2)
0.6
0.6
0.3
0.5
0.5
0.6
."
0.8
0.5
0.7
0.9
0.5
0.6
0.3
0.6
0.6
0.7
0.7
0.8
0.5
0.7
0.9
0.6
0.7
0.3
0.6
0.7
0.7
0.7
0.9
0.5
0.8
0.9
0.5
0.8
0.3
0.7
0.9
0.8
0.7
0.9
0.5
0.8
0.9
STANDARD ERROR FOR
PRECISION
_
0.04
0.03
0.04
0.04
0.05
_
0.04
0.03
0.03
0.04
_
0.05
0.03
0.03
0.06
0.05
_
0.05
0.03
0.03
0.04
_
0.07
0.03
0.03
0.10
0.09
_
006
0.03
0.04
0.04
_
0.13
0.03
0.03
0.16
0.17
.
0.11
0.03
0.06
0.04
'Precision at 1 standard deviation.
tNonparametric estimation used.
"A reliable estimate at 0.0 mg/cm2 could not be obtained.
D-84
-------
APPENDIX D: XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
Table D-47. Classification Results For Warrington Microlead I Revision 4 Readings (All Substrate Corrected
and Uncorrected Except For Plaster Which is Uncorrected), Classified Using the Inconclusive
Ranges Reported in the XRF Performance Characteristic Sheet and Compared to Laboratory
Results in mg/cm2 Lead Classified Using the 1.0 mg/cm2 Lead Federal Standard For Data Taken
During the EPA/HUD Field Study.
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick (Uncorrected)
Concrete (Uncorrected)
Drywall (Uncorrected)
Metal (Uncorrected)
Wood (Uncorrected)
TOTAL'
INCONCLUSIVE
RANGE
0.8 to 1.3
0.5 to 1.4
0.9 to 1.1
0.9 to 1 .4
0.7 to 1.6
0.9 to 1.6
0.9 to 1 .7
0.9 to 1.8
0.9 to 1.2
0.9 to 1.3
0.9 to 2.3
FALSE POSITIVE
RATE
4.2% (6/1 44)
4.9% (19/388)
2.1% (5/237)
5.4% (17/314)
1 .5% (6/404)
4.6% (24/519)
2.8% (4/144)
6.4% (25/388)
1 1 4% (27/237)
11. 2% (35/31 4)
4.8% (25/51 9)
3.8% (77/2006)
FALSE NEGATIVE
RATE
0.0% (0/42)
5.4% (3/56)
(0/0)
3.3% (3/92)
5.1% (3/59)
8.6% (19/220)
2.4% (1/42)
1.8% (1/56)
(0/0)
2.2% (2/92)
3.6% (8/220)
6.6% (28/427)
INCONCLUSIVE
RATE
10.8% (20/1 86)
27.3% (121/444)
1.3% (3/237)
5.9% (24/406)
17.1% (79/463)
8.5% (63/739)
15.1% (28/1 86)
17.8% (79/444)
6.3% (15/237)
6.9% (28/406)
20.6% (152/739)
12.5% (31 0/2475)
'Total results are for values reported in PCS.
D-85
-------
APPENDIX D: XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
D.7 XRF Performance Characteristic Sheet for the TN Technologies Pb Analyzer
and Related Results
D-86
-------
APPENDIX D: XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
TN Technologies (TN Spectrace); Pb Analyzer
EFFECTIVE DATE: October 31, 1995 EDITION NO.: 1
MANUFACTURER AND MODEL
Manufacturer: TN Technologies, Inc. (TN Spectrace)
Make: Pb Analyzer
Model: 9292
Source: Cd109
Note: This sheet supersedes all previous sheets for the XRF instrument of the make, model,
and source shown above.
EVALUATION DATA SOURCE AND DATE:
This sheet is supplemental information to be used in conjunction with Chapter 7 of the HUD Guidelines
for the Evaluation and Control of Lead-Based Paint Hazards in Housing ("HUD Guidelines"). Performance
parameters shown on this sheet are calculated from evaluation data collected during the EPA/HUD field
evaluation study conducted from March through October 1993 and from data collected during testing on the
EPA/HUD archived building components in January 1995 and in September 1995. The field evaluation data
were collected from approximately 1,200 test locations using two instruments both with radiation sources
installed in April 1993. The results of this study are reported in A Field Test of Lead-Based Paint Testing
Technologies: Technical Report, EPA 747-R-95-002b, May 1995. The archive testing data were collected
from approximately 150 test locations using two instruments. The instrument that was used in January had
a radiation source installed in July 1994 and the instrument that was used in September 1995 had a radiation
source installed in January 1995. All of the aforementioned instruments had 30 mCi initial strengths.
FIELD OPERATION GUIDANCE
OPERATING PARAMETERS:
Performance parameters shown in this sheet are applicable only when operating the instrument under
the same conditions as the evaluation testing and using the procedures described in Chapter 7 of the HUD
Guidelines. Operating parameters include:
• Manufacturer-recommended warm-up and quality control procedures
• Use the Multifamily Decision Flowchart for determining the presence of lead on a component type in
multifamily housing
• Nominal 15-second readings on three locations per component for single-family housing and one
location per component for multifamily housing
• Calibration checks are taken using the red (1.02 mg/cm2) NIST Standard Reference Material (SRM
No. 2579) paint film
• Lead-based paint is defined as paint with lead equal to or in excess of 1.0 mg/cm2.
1 of 5
Figure D-6. XRF Performance Characteristic Sheet for the TN Technologies Pb
Analyzer.
D-87
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
TN Technologies (TN Spectrace); Pb Analyzer
XRF CALIBRATION CHECK:
Chapter 7 of the HUD Guidelines recommends using a calibration check procedure to determine the
operating condition of the XRF instrument. If the observed calibration check average minus 1.02 mg/cm2 is
greater than the positive (plus) calibration check tolerance value, or less than the negative (minus) calibration
check tolerance value, then the instructions provided by the manufacturer should be followed in order to bring
the instrument back into control before any more XRF testing is done. This calibration check is estimated to
produce an incorrect result (that is, a finding that the instrument is out of calibration) very infrequently - once
out of every 200 times this procedure is followed.
minus value = -0.3 mg/cm2
plus value = +0.4 mg/cm2
FOR XRF RESULTS BELOW 4.0 mg/cm2, SUBSTRATE CORRECTION RECOMMENDED FOR:
None
SUBSTRATE CORRECTION NOT RECOMMENDED FOR:
Brick, Concrete, Drywall, Metal, Plaster, and Wood
INCONCLUSIVE RANGE OR THRESHOLD:
XRF results are classified using the inconclusive range. In single-family housing, an XRF result is the
average of three readings taken on a testing combination. (A testing combination is a location on a painted
surface as defined in Chapter 7 of the HUD Guidelines). In multifamily housing, an XRF result is a single
reading taken on a testing combination. For computing the XRF result, use all digits that are reported by the
instrument. For the inconclusive range, results are classified as positive if they are greater than or equal to
the upper limit of the inconclusive range, and negative if they are less than or equal to the lower limit of the
inconclusive range. Inconclusive ranges were determined for comparing results to the 1.0 mg/cm2 standard.
For a listing of laboratories recommended by the EPA National Lead Laboratory Accreditation Program
(NLLAP) for the analysis of samples to resolve an inconclusive XRF result or additional confirmational
analysis, call the National Lead Information Center Clearinghouse on 1 -800-424-LEAD.
DESCRIPTION
Results not corrected for substrate bias
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
INCONCLUSIVE RANGE in mg/cm2
LOWER BOUND
0.9
0.9
0.9
0.9
0.9
0.9
UPPER BOUND
1.2
1.2
1.2
1.2
1.1
1.3
2 of 5
Figure D-6 continued.
XRF Performance Characteristic Sheet for the TN
Technologies Pb Analyzer.
D-88
-------
APPENDIX D: XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
TN Technologies (TN Spectrace); Pb Analyzer
INSTRUCTIONS FOR EVALUATING XRF TESTING:
Chapter 7 of the HUD Guidelines recommends several options for evaluating XRF testing. Among those
options is the following procedure which may be used after XRF testing has been completed. In single-family
housing, an XRF result is the average of three readings taken on a testing combination. (A testing
combination is a location on a painted surface as defined in Chapter 7 of the HUD Guidelines). In multifamily
housing, an XRF result is a single reading taken on a testing combination. If a multifamily housing
development is being retested, randomly select two units from within the development from which the ten
testing combinations should be randomly selected.
Randomly select ten testing combinations for retesting from each house or from the two selected units.
Conduct XRF retesting at the ten testing combinations selected for retesting.
Determine if the XRF testing in the units or house passed or failed the test by applying the steps below.
Compute the Retest Tolerance Limit by the following steps:
Determine XRF results for the original and retest XRF readings. Do not correct the original or
retest results for substrate bias. In single-family housing a result is defined as the average of
three readings. In multifamily housing, a result is a single reading. Therefore, there will be ten
original and ten retest XRF results for each house or for the two selected units.
Compute the average of the original and re-test result for each of the ten testing combinations.
Square the average for each testing combination.
Add the ten squared averages together. Call this quantity C.
Multiply the number C by 0.0072. Call this quantity D.
Add the number 0.032 to D. Call this quantity E.
Take the square root of E. Call this quantity F.
Multiply F by 1.645. The result is the Retest Tolerance Limit.
Compute the overall average of all ten retest XRF results over all ten testing combinations selected
for retesting.
Take the difference of the overall average of the ten original XRF results and the overall average of
the ten retest XRF results. If the difference is negative, drop the negative sign.
If the difference of the overall averages is less than the Retest Tolerance Limit, the inspection has
passed the retest. If the difference of the overall averages equals or exceeds the Retest Tolerance
Limit, this procedure should be repeated with ten new testing combinations. If the difference of the
overall averages is equal to or greater than the Retest Tolerance Limit a second time, then the
inspection should be considered deficient.
3 of 5
Figure D-6 continued. XRF Performance Characteristic Sheet for the TN
Technologies Pb Analyzer.
D-89
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
TN Technologies (TN Spectrace); Pb Analyzer
Use of this procedure is estimated to produce a spurious result approximately 1 % of the time. That is, results
of this procedure will call for further examination when no examination is warranted in approximately 1 out of
100 dwelling units tested.
BIAS AND PRECISION:
I
e
(
These bias and precision data were computed without substrate correction from samples with reported
aboratory results less than 4.0 mg/cm2 lead. There were 88 test locations with a laboratory reported result
jqual to or greater than 4.0 mg/cm2 lead. Of these, none had XRF readings less than 1.0 mg/cm2. These
teta are for illustrative purposes only. Substrate correction is not recommended for this XRF instrument. Bias
and precision ranges are provided to show the variability found between machines of the same model. Units
ire in mg/cm2.
MEASURED AT
0.0 mg/cm2
0.5 mg/cm2
1 .0 mg/cm2
2.0 mg/cm2
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
BIAS
(mg/cm2)
0.0
0.0
0.0
0.0
0.0
0.0
0.1
0.1
0.1
0.1
0.0
0.1
0.2
0.2
0.2
0.2
0.1
0.3
0.4
0.3
0.5
0.4
0.2
0.5
BIAS
RANGE*
(mg/cm2)
( 0.0, 0.0)
( 0.0, 0.0)
( 0.0, 0.0)
(-0.1,0.1)
(-0.1,0.0)
( 0.0, 0.0)
( 0.0, 0.2)
( 0.0, 0.2)
( 0.0, 0.2)
( 0.0, 0.3)
(-0.1,0.2)
(0.1,0.2)
( 0 0, 0.4)
( 0.0, 0.4)
(0.1, 0.4)
( 0.0, 0.5)
(-0.1,0.3)
(0.1,0.4)
( 0.0, 0.7)
( 0.0, 0.7)
( 0.3, 0 7)
( 0.0, 0.8)
(-0.3, 0.7)
( 0.3, 0.7)
PRECISION*
(mg/cm2)
0.1
0.1
0.1
0.1
0.1
0.1
0.3
0.3
0.3
0.3
0.3
0.3
0.4
0.4
04
0.4
0.4
04
0.6
0.5
0.5
0.6
0.5
0.6
PRECISION
RANGE'
(mg/cm2)
(0.1, 0.1)
(0.1,0.1)
(0.1,0.1)
(0.1,01)
(0.1,0.1)
(<0.1,0.1)
( 0.3, 0.3)
( 0.2, 0.3)
(0.1,0.3)
( 0.3, 0.3)
(0.1,0.3)
( 0.3, 0.3)
(04,0.5)
( 0.3, 0.5)
( 0.2, 0.5)
( 0.4, 0.5)
(0.1,0.5)
( 0.4, 0.5)
(0.5,0.6)
( 0.4, 0.6)
( 0.3, 0.6)
( 0.5, 0.6)
(0.1,0.6)
( 0.5, 0.6)
'Ranges are provided to show the variability between machines of the same model
Precision at 1 standard deviation.
4 of 5
Figure D-6 continued.
XRF Performance Characteristic Sheet for the TN
Technologies Pb Analyzer.
D-90
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
TN Technologies (TN Spectrace); Pb Analyzer
A document titled Methodology for XRF Performance Characteristic Sheets provides an explanation of
the statistical methodology used to construct the data in the sheets, and provides empirical results from using
the recommended inconclusive ranges or thresholds for specific XRF instruments. For a copy of this
document call the National Lead Information Center Clearinghouse at 1-800-424-LEAD.
This XRF Performance Characteristic Sheet is a joint product of the U.S Environmental Protection Agency (EPA) and the U.S.
Department of Housing and Urban Development (HUD). The issuance of this sheet does not constitute rulemaking The
information provided here is intended solely as guidance to be used in conjunction with Chapter 7 of the Guidelines for the
Evaluation and Control of Lead-Based Paint Hazards in Housing. EPA and HUD reserve the right to revise this guidance.
Please address questions and comments on this sheet to. Director, Office of Lead-Based Paint Abatement and Poisoning
Prevention, U S Department of Housing and Urban Development, Room B-133, 451 Seventh St, S W., Washington, DC 20410
5 Of 5
Figure D-6 continued.
XRF Performance Characteristic Sheet for the TN
Technologies Pb Analyzer.
D-91
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
Table D-48. Bias Estimates, and Their Standard Errors, of TN Technologies Pb Analyzer K-Shell 15-Second
Readings.
MEASURED AT
0.0 mg/cm2
0.5 mg/cm2
1 .0 mg/cm2
2.0 mg/cm2
CORRECTED?
No
No
No
No
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
BIAS (mg/cm2)
0.0
0.0
0.0
0.0
0.0
0.0
0.1
0.1
0.1
0.1
0.0
0.1
0.2
0.2
0.2
0.2
0.1
0.3
0.4
0.3
0.5
0.4
0.2
0.5
STANDARD ERROR FOR BIAS
0.03
0.03
0.03
0.03
0.04
0.03
0.05
0.03
0.05
0.04
0.04
0.03
0.10
0.06
0.09
0.06
0.06
0.06
0.20
0.12
0.17
0.12
0.11
0.12
D-92
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
Table D-49. Precision Estimates, and Their Standard Errors, of TN Technologies Pb Analyzer K-Shell
15-Second Readings.
MEASURED AT
0.0 mg/cm2
0.5 mg/cm2
1 .0 mg/cm2
2.0 mg/cm2
CORRECTED?
No
No
No
No
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
PRECISION' (mg/cm2)
0.1
0.1
01
0.1
0.1
0.1
0.3
0.3
0.3
0.3
0.3
0.3
0.4
0.4
0.4
0.4
0.4
0.4
0.6
0.5
0.5
0.6
0.5
0.6
STANDARD ERROR FOR
PRECISION
0.02
0.01
0.02
0.02
0.01
0.02
0.05
0.03
0.03
0.03
0.03
0.03
0.07
0.04
0.05
0.04
0.04
0.04
0.10
0.06
0.07
0.06
0.07
0.06
'Precision at 1 standard deviation.
D-93
-------
APPENDIX D: XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
Table D-50. Classification Results For TN Technologies Pb Analyzer K-Shell 15-Second Readings
(Uncorrected), Classified Using Inconclusive Ranges Reported in the XRF Performance
Characteristic Sheet and Compared to Laboratory Results in mg/cm2 Lead Classified Using the
1.0 mg/cm2 Lead Federal Standard For Data Taken During the EPA/HUD Field Study and From
Testing Archived Building Components in January 1995 and in September 1995.
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
TOTAL
INCONCLUSIVE
RANGE
0.9 to 1.2
0.9 to 1.2
0.9 to 1.2
0.9 to 1 .2
0.9 to 1.1
0.9 to 1.3
FALSE POSITIVE
RATE
0.0% (0/74)
1.0% (2/1 95)
0.7% (1/140)
3.5% (7/1 99)
1 .6% (4/254)
4.5% (15/335)
2.4% (29/1 197)
FALSE NEGATIVE
RATE
0.0% (0/25)
11.1% (3/27)
(0/0)
7.6% (5/66)
2.5% (1/40)
4.7% (7/148)
5.2% (16/306)
INCONCLUSIVE
RATE
2.0% (2/99)
1.4% (3/222)
0.0% (0/140)
3.0% (8/265)
0.3% (1/294)
4.6% (22/483)
2.4% (36/1 503)
D-94
-------
APPENDIX D: XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
D.8 XRF Performance Characteristic Sheet for the Princeton Gamma-Tech XK-3
and Related Results
D-95
-------
APPENDIX D: XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
Princeton Gamma-Tech, Inc.; XK-3
EFFECTIVE DATE: September 25, 1995 EDITION NO.: 1
MANUFACTURER AND MODEL:
Make: Princeton Gamma-Tech, Inc.
Model: XK-3
Source: Co57
Note: This sheet supersedes all previous sheets for the XRF instrument of the make, model, and
source shown above
EVALUATION DATA SOURCE AND DATE:
This sheet is supplemental information to be used in conjunction with Chapter 7 of the HUD Guidelines
for the Evaluation and Control of Lead-Based Paint Hazards in Housing ("HUD Guidelines"). Performance
parameters shown on this sheet are calculated from evaluation data collected during the EPA/HUD field
evaluation study conducted from March through October 1993. The data were collected from approximately
1,200 test locations using three instruments. One instrument had a March 1993 source and the other two
instruments had April 1993 sources. All three instruments had sources with 10 mCi initial strengths. The
results of this study are reported in A Field Test of Lead-Based Paint Testing Technologies: Technical Report,
EPA 747-R-95-002b, May 1995.
FIELD OPERATION GUIDANCE
OPERATING PARAMETERS:
Performance parameters shown in this sheet are applicable only when operating the instrument under
the same conditions as the evaluation testing and using the procedures described in Chapter 7 of the HUD
Guidelines. Operating parameters include:
• Manufacturer-recommended warm-up and quality control procedures
• Use the Multifamily Decision Flowchart for determining the presence of lead on a component type in
multifamily housing
• Nominal 15-second readings on three locations per component for single-family housing and one
location per component for multifamily housing
• Calibration checks are taken using the red (1.02 mg/cm2) NIST Standard Reference Material (SRM
No. 2579) paint film
• Readings for determining the substrate correction values are taken on bare substrate covered with
red (1.02 mg/cm2) NIST SRM paint film
• Lead-based paint is defined as paint with lead equal to or in excess of 1.0 mg/cm2.
1 of 5
Figure D-7. XRF Performance Characteristic Sheet for the Princeton Gamma-Tech
XK-3.
D-96
-------
APPENDIX D: XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
Princeton Gamma-Tech, Inc.; XK-3
XRF CALIBRATION CHECK:
Chapter 7 of the HUD Guidelines recommends using a calibration check procedure to determine the
operating condition of the XRF instrument. For this instrument, calibration check readings should be taken on
wood. If the observed calibration check average minus 1.02 mg/cm2 is greater than the positive (plus)
calibration check tolerance value, or less than the negative (minus) calibration check tolerance value, then
the instructions provided by the manufacturer should be followed in order to bring the instrument back into
control before any more XRF testing is done. Testing must cease for those instruments with readings that
exceed the calibration check tolerance limits in accordance with manufacturer's specifications. This calibration
check is estimated to produce an incorrect result (that is, a finding that the instrument is out of calibration) very
infrequently - once out of every 200 times this procedure is followed.
minus value = -0.5 mg/cm2
plus value = +1.3 mg/cm2
(Operators may choose to use the limits in the manufacturer's instruction manual for this calibration check.
The rate of an incorrect result if the limits in the manufacturer's instruction manual are followed may be
different from the rate of an incorrect result stated here).
FOR XRF RESULTS BELOW 4.0 mg/cm2, SUBSTRATE CORRECTION RECOMMENDED FOR:
Brick, Concrete, Drywall, Metal, Plaster, and Wood.
SUBSTRATE CORRECTION NOT RECOMMENDED FOR:
None.
SUBSTRATE CORRECTION VALUE COMPUTATION:
Chapter 7 of the HUD Guidelines provides guidance on correcting XRF results for substrate bias.
Supplemental guidance for using the red (1.02 mg/cm2) NIST SRM paint film for substrate correction is
provided below.
XRF results are corrected for substrate bias by subtracting from each XRF result a correction value
determined separately in each house for single-family housing or in each development for multifamily housing,
for each substrate. The correction value is an average of XRF readings taken over red NIST SRM (1.02
mg/cm2) paint films at test locations that had been scraped clean of their paint covering. Compute the
correction values as follows:
Using the same XRF instrument, take three readings on a bare substrate area covered with the red NIST
SRM (1.02 mg/cm2) paint film. Repeat this procedure by taking three more readings on a second bare
substrate area of the same substrate covered with the red NIST SRM (1.02 mg/cm2) paint film.
Compute the correction value for each substrate type by computing the average of all six readings as
shown below.
2 of 5
Figure D-7 continued. XRF Performance Characteristic Sheet for the Princeton
Gamma-Tech XK-3.
D-97
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
Princeton Gamma-Tech, Inc.; XK-3
For each substrate type:
Correction
Value
\ 1" + 2nd * 3rd + 4th + 5th + 6th Reading , „, . 2
> = =- - 1.02mg/cm2
i 6
Repeat this procedure for each substrate tested in the house or housing development.
INCONCLUSIVE RANGE OR THRESHOLD:
XRF results are classified using either the threshold or the inconclusive range. In single-family housing,
an XRF result is the average of three readings taken on a testing combination. (A testing combination is a
location on a painted surface as defined in Chapter 7 of the HUD Guidelines). In multifamily housing, an XRF
result is a single reading taken on a testing combination. For computing the XRF result, use all digits that are
reported by the instrument. For the threshold, results are classified as positive if they are greater than or
equal to the threshold, and negative if they are less than the threshold. There is no inconclusive classification
when using the threshold. For the inconclusive range, results are classified as positive if they are greater than
or equal to the upper limit of the inconclusive range, and negative if they are less than or equal to the lower
limit of the inconclusive range. Thresholds and inconclusive ranges were determined for comparing results
to the 1.0 mg/cm2 standard. For a listing of laboratories recommended by the EPA National Lead Laboratory
Accreditation Program (NLLAP) for the analysis of samples to resolve an inconclusive XRF result or additional
confirmational analysis, call the National Lead Information Center Clearinghouse at 1-800-424-LEAD.
DESCRIPTION
Readings corrected for substrate
bias on all substrates
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
THRESHOLD
(mg/cm2)
None
None
1.0
None
None
None
INCONCLUSIVE
RANGE
{mg/cm2)
0.9 to 1.3
0.8 to 1.7
None
0.4 to 1.8
0.7 to 1.4
0.9 to 1.4
INSTRUCTIONS FOR EVALUATING XRF TESTING:
Chapter 7 of the HUD Guidelines recommends several options for evaluating XRF testing. Among those
options is the following procedure which may be used after XRF testing has been completed. In single-family
housing, an XRF result is the average of three readings taken on a testing combination. (A testing
combination is a location on a painted surface as defined in Chapter 7 of the HUD Guidelines). In multifamily
housing, an XRF result is a single reading taken on a testing combination. If a multifamily housing
development is being retested, randomly select two units from within the development from which the ten
testing combinations should be randomly selected.
Randomly select ten testing combinations for retesting from each house or from the two selected units.
Conduct XRF retesting at the ten testing combinations selected for retesting.
3 of 5
Figure D-7 continued.
XRF Performance Characteristic Sheet for the Princeton
Gamma-Tech XK-3.
D-98
-------
APPENDIX D: XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
Princeton Gamma-Tech, Inc.; XK-3
Determine if the XRF testing in the units or house passed or failed the test by applying the steps below.
Compute the Retest Tolerance Limit by the following steps:
Determine XRF results for the original and retest XRF readings. Do not correct the original or
retest results for substrate bias. In single-family housing a result is defined as the average of
three readings. In multifamily housing, a result is a single reading. Therefore, there will be ten
original and ten retest XRF results for each house or for the two selected units.
Compute the average of the original and re-test result for each of the ten testing combinations.
Square the average for each testing combination.
Add the ten squared averages together. Call this quantity C.
Multiply the number C by 0.0072. Call this quantity D.
Add the number 0.032 to D. Call this quantity E.
Take the square root of E. Call this quantity F.
Multiply F by 1.645. The result is the Retest Tolerance Limit.
Compute the overall average of all ten retest XRF results over all ten testing combinations selected
for retesting.
Take the difference of the overall average of the ten original XRF results and the overall average of
the ten retest XRF results. If the difference is negative, drop the negative sign.
If the difference of the overall averages is less than the Retest Tolerance Limit, the inspection has
passed the retest. If the difference of the overall averages equals or exceeds the Retest Tolerance
Limit, this procedure should be repeated with ten new testing combinations. If the difference of the
overall averages is equal to or greater than the Retest Tolerance Limit a second time, then the
inspection should be considered deficient.
Use of this procedure is estimated to produce a spurious result approximately 1% of the time. That is, results
of this procedure will call for further examination when no examination is warranted in approximately 1 out of
100 dwelling units tested.
BIAS AND PRECISION:
Do not use these bias and precision data to correct for substrate bias. These bias and precision data
were computed without substrate correction from samples with reported laboratory results less than 4.0
mg/cm2 lead. There were 143 testing locations with a laboratory reported result equal to or greater than 4.0
mg/cm2 lead. Of these, 1 had an XRF reading less than 1.0 mg/cm2. These data are for illustrative purposes
only. Actual bias must be determined on the site. Inconclusive ranges provided above already account for
4 of 5
Figure D-7 continued. XRF Performance Characteristic Sheet for the Princeton
Gamma-Tech XK-3.
D-99
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
Princeton Gamma-Tech, Inc.; XK-3
bias and precision. Bias and precision ranges are provided whenever significant variability was found between
machines of the same model. Units are in mg/cm2.
MEASURED
AT
0.0 mg/cm2
0.5 mg/cm2
1 .0 mg/cm2
2.0 mg/cm2
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
BIAS
(mg/cm2)
0.9
1.3
-0.1
0.9
0.8
0.2
0.9
1.3
0.0
1.1
0.8
0.4
0.9
1.3
0.0
1.3
0.8
0.6
0.9
1.3
0.1
1.7
0.7
1.0
BIAS RANGE
(mg/cm2)
(0.6, 1.9)
(-0.3, 0.2)
(0.5, 1.4)
(0.4, 1.7)
(-0.1, 1.0)
(0.7, 1.9)
(-0.2, 0.2)
(0.7, 1.6)
0.2, 1.6)
0.1, 1.1)
0.7, 2.0)
-0.1, 0.2)
0.9, 1.7)
(0.0, 1.6)
(0.3, 1.3)
( 0.7, 2.0)
(0.1, 0.2)
( 1.4,2.1)
(-0.3, 1.6)
(0.8, 1.5)
PRECISION"
(mg/cm2)
0.6
0.6
0.3
0.5
0.5
0.4
0.6
0.6
0.4
0.8
0.6
0.6
0.6
0.7
0.4
1.0
0.6
0.7
0.6
0.8
0.6
1.4
0.7
0.9
PRECISION
RANGE (mg/cm2)
( 0.2, 0.6)
( 0.2, 0.3)
( 0.4, 0.5)
( 0.4, 0.5)
( 0.3, 0.5)
( 0.5, 0.7)
( 0.3, 0.4)
( 0.4, 0.9)
( 0.4, 0.6)
( 0.3, 0.9)
( 0.6, 0.8)
( 0.4, 0.5)
(0.5, 1.1)
( 0.4, 0.7)
(0.3, 1.2)
( 0.6, 0.9)
( 0.5, 0.6)
(0.6, 1.6)
( 0.4, 0.8)
(0.3, 1.7)
'Precision at 1 standard deviation
A document titled Methodology for XRF Performance Characteristic Sheets provides an explanation of
he statistical methodology used to construct the data in the sheets and provides empirical results from using
he recommended inconclusive ranges or thresholds for specific XRF instruments. For a copy of this
document call the National Lead Information Center Clearinghouse at 1-800-424-LEAD.
This XRF Performance Characteristic Sheet is a joint product of the U.S. Environmental Protection Agency (EPA) and the U.S
Department of Housing and Urban Development (HUD). The issuance of this sheet does not constitute rulemaking. The
information provided here is intended solely as guidance to be used in conjunction with Chapter 7 of the Guidelines for the
Evaluation and Control of Lead-Based Paint Hazards in Housing. EPA and HUD reserve the right to revise this guidance.
Please address questions and comments on this sheet to: Director, Office of Lead-Based Paint Abatement and Poisoning
Prevention, U.S. Department of Housing and Urban Development, Room B-133, 451 Seventh St, S.W., Washington, DC 20410.
5 of 5
Figure D-7 continued.
XRF Performance Characteristic Sheet for the Princeton
Gamma-Tech XK-3.
D-100
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
Table D-51. Inconclusive Ranges For Princeton Gamma-Tech XK-3 Readings Where Substrate
Correction Is Not Performed, But Substrate Correction Is Recommended.
DESCRIPTION
Results not corrected for substrate bias on any substrate
SUBSTRATE
Brick
Concrete
Drywatl
Metal
Plaster
Wood
INCONCLUSIVE RANGE
(mg/cm2)
0.9 to 2.4
0.9 to 2. 9
0.9 to 1.1
0.9 to 2.9
0.9 to 2.2
0.9 to 1 .8
D-101
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
Table D-52. Bias Estimates, and Their Standard Errors, of Princeton Gamma-Tech XK-3 Readings.
MEASURED AT
0.0 mg/cm2
0.5 mg/cm2
1 .0 mg/cm2
2.0 mg/cm2
CORRECTED?
No
Yes
No
Yes
No
Yes
No
Yes
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
BIAS (mg/cm2)
0.9
1.3
-0.1
0.9
0.8
0.2
0.1
0.2
-0.2
-0.2
0.1
-0.2
0.9
1.3
0.0
1.1
0.8
0.4
0.0
0.2
-0.1
0.0
0.0
0.0
0.9
1.3
0.0
1.3
0.8
0.6
0.0
0.2
0.0
0.2
0.0
0.2
0.9
1.3
0.1
1.7
0.7
1.0
-0.1
0.1
0.2
0.7
-0.2
0.6
STANDARD ERROR FOR BIAS
0.06
0.05
0.03
0.06
0.05
0.03
0.05
0.05
0.03
0.06
0.05
0.03
0.10
0.05
0.10
0.06
0.05
0.03
0.09
0.05
0.09
0.06
0.05
0.04
0.22
0.10
0.21
0.10
0.09
0.06
0.19
0.09
0.19
0.10
0.09
0.06
0.47
0.21
0.43
0.20
0.21
0.13
0.40
0.20
0.39
0.20
0.21
0.13
D-102
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
Table D-53. Precision Estimates, and Their Standard Errors, of Princeton Gamma-Tech XK-3 Readings.
MEASURED AT
0.0 mg/cm2
0.5 mg/cm2
1.0 mg/cm2
2
£.u mcj/cm
CORRECTED?
Kin
mu
No
Yes
I CO
No
I1U
Yes
No
I1U
Yes
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick
Concrete
Drywall
Metal
Plaster
Wood
PRECISION' (mg/cm2)
0.6
0.6
0.3
0.5
0.5
0.4
0.5
0.6
0.3
0.6
0.6
0.5
0.6
0.6
0.4
0.8
0.6
0.6
0.5
0.7
0.4
0.8
0.6
0.6
0.6
0.7
0.4
1.0
0.6
0.7
0.5
0.7
0.4
1.0
0.7
0.8
0.6
0.8
0.6
1.4
0.7
0.9
0.5
0.8
0.4
1.3
0.7
1.0
STANDARD ERROR
FOR PRECISION
0.04
0.04
0.03
0.05
0.04
0.03
0.03
0.04
0.02
0.05
0.03
0.03
0.04
0.04
0.09
0.06
0.04
0.04
0.03
0.04
0.04
0.05
0.04
0.04
0.04
0.07
0.16
0.09
007
0.06
0.03
007
006
0.08
0.06
0.06
0.04
0.12
0.26
0.13
0.13
0.09
0.03
0.12
0.11
0.13
0.11
0.09
'Precision at 1 standard deviation
D-103
-------
APPENDIX D: XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
Table D-54. Classification Results For the Princeton Gamma-Tech XK-3 K-shell 15-Second Readings
(Corrected and Uncorrected), Classified Using the Inconclusive Ranges and Threshold Value
Reported in the XRF Performance Characteristic Sheet and Compared to Laboratory Results
in mg/cm2 Lead Classified Using the 1.0 mg/cm2 Lead Federal Standard For Data Taken
During the EPA/HUD Field Study.
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
Brick (Uncorrected)
Concrete (Uncorrected)
Drywall (Uncorrected)
Metal (Uncorrected)
Plaster (Uncorrected)
Wood (Uncorrected)
TOTAL'
INCONCLUSIVE RANGE
OR THRESHOLD
0.9 to 1.3
0.8 to 1.7
1.0
0.4 to 1.8
0.7 to 1.4
0.9 to 1.4
0.9 to 2.4
0.9 to 2.9
0.9 to 1.1
0.9 to 2.9
0.9 to 2.2
0.9 to 1.8
FALSE POSITIVE
RATE
6.3% (9/144)
2.6% (10/388)
2.1% (5/237)
2.2% (7/314)
4.0% (16/403)
3.5% (18/51 9)
6.3% (9/144)
5.4% (21/388)
3.0% (7/237)
4.5% (14/314)
8.2% (33/403)
4.2% (22/51 9)
3.2% (65/2005)
FALSE NEGATIVE
RATE
2.4% (1/42)
3.6% (2/56)
(0/0)
12.0% (11/92)
6.8% (4/59)
7.1% (16/224)
2.4% (1/42)
1.8% (1/56)
-
5.4% (5/92)
1.7% (1/59)
4.0% (9/224)
7.2% (34/473)
INCONCLUSIVE
RATE
4.8% (9/1 86)
20.5% (91/444)
0.0% (0/237)
25.1% (102/406)
19.3% (89/462)
6.1% (45/743)
36.0% (67/186)
56.5% (251/444)
2.1% (5/237)
46.1% (187/406)
35.3% (163/462)
12.9% (96/743)
13.6% (336/2478)
'Total results are for values reported in PCS.
D-104
-------
APPENDIX D: XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
D.9 XRF Performance Characteristic Sheet for the Niton XL-309 Spectrum
Analyzer and Related Results
D-105
-------
APPENDIX D: XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
Niton Corporation; XL-309 Spectrum Analyzer
EFFECTIVE DATE: August 24, 1995 EDITION NO.: 1
MANUFACTURER AND MODEL:
Make: Niton Corporation
Model: XL-309 Spectrum Analyzer
Source: Cd109
Note: This sheet supersedes all previous sheets for the XRF instrument of the make, model, source,
and software versions shown above.
EVALUATION DATA SOURCE AND DATE:
This sheet is supplemental information to be used in conjunction with Chapter 7 of the HUD Guidelines
for the Evaluation and Control of Lead-Based Paint Hazards in Housing ("HUD Guidelines"). Performance
parameters shown on this sheet are calculated from the EPA/HUD evaluation using archived building
components. Testing was conducted March 1995 on approximately 150 test locations using a single
instrument with an October 1994 source at 10 mCi initial strength while running software version 1.2C. These
results supersede the 1993 testing of XL prototypes reported in the document titled: A Field Test of Lead-
Based Paint Testing Technologies.
FIELD OPERATION GUIDANCE
OPERATING PARAMETERS:
Performance parameters shown in this sheet are applicable only when operating the instrument under
the same conditions as the evaluation testing and using the procedures described in Chapter 7 of the HUD
Guidelines. Operating parameters include:
• Manufacturer-recommended warm-up and quality control procedures
• Use the Multifamily Decision Flowchart for determining the presence of lead on a component type in
multifamily housing
• Nominal 20-second readings for L-shell results or 120-second readings for K-shell results on three
locations per component for single-family housing and one location per component for multifamily
housing
• Calibration checks are taken using the red (1.02 mg/cm2) NIST Standard Reference Material (SRM
No. 2579) paint film
• Lead-based paint is defined as paint with lead equal to or in excess of 1.0 mg/cm2.
XRF CALIBRATION CHECK:
Chapter 7 of the HUD Guidelines recommends using a calibration check procedure to determine the
operating condition of the XRF instrument. If the observed calibration check average minus 1.02 mg/cm2 is
greater than the positive (plus) calibration check tolerance value, or less than the negative (minus) calibration
check tolerance value, then the instructions provided by the manufacturer should be followed in order to bring
1 of 7
Figure D-8. XRF Performance Characteristic Sheet for the Niton XL-309 Spectrum
Analyzer.
D-106
-------
APPENDIX D: XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
Niton Corporation; XL-309 Spectrum Analyzer
the instrument back into control before any more XRF testing is done. This calibration check is estimated to
produce an incorrect result (that is, a finding that the instrument is out of calibration) very infrequently - once
out of every 200 times this procedure is followed
minus value = -0.1 mg/cm2
plus value = +0.1 mg/cm2
FOR XRF RESULTS BELOW 4.0 mg/cm2, SUBSTRATE CORRECTION RECOMMENDED FOR:
None
SUBSTRATE CORRECTION NOT RECOMMENDED FOR:
Brick, Concrete, Drywall, Metal, Plaster, and Wood
HOW TO CLASSIFY READINGS:
This section describes how to apply the readings and other information displayed by this instrument to
determine the presence or absence of lead in paint using the procedures recommended in Chapter 7 of the
HUD Guidelines. These guidelines recommend classifying XRF results as positive, negative, or inconclusive
compared to the 1.0 mg/cm2 standard. But because this instrument displays readings and ancillary
information useful for classification purposes, an algorithmic procedure is described that makes use of not only
the XRF reading but some of the other displayed information as well.
As detailed below, the algorithm for classifying results is first applied to 20-second nominal L-shell
readings followed by 120-second nominal K-shell readings to resolve inconclusive results and laboratory
analysis of paint-chip samples, if necessary. For a listing of laboratories recommended by the EPA National
Lead Laboratory Accreditation Program (NLLAP) for the analysis of samples to perform additional
confirmational analysis, call the National Lead Information Center Clearinghouse at 1-800-424-LEAD.
XRF results are classified using threshold values. For the XL-309, threshold values are the only values
provided for classifying results. Results are classified as positive if they are greater than or equal to the
threshold, and as negative if they are less than the threshold. There is no inconclusive classification when
using threshold values. However, inconclusive results still may be obtained regardless of whether decisions
are based on L-shell readings, K-shell readings, or both, as described below. Use all digits that are reported
by the instrument. Threshold values, which were determined for comparing results to the 1.0 mg/cm2
standard, are provided in the following table.
2 of 7
Figure D-8 continued. XRF Performance Characteristic Sheet for the Niton XL-309
Spectrum Analyzer.
D-107
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
Niton Corporation; XL-309 Spectrum Analyzer
DESCRIPTION
Results not corrected for substrate bias
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
THRESHOLD' (mg/cm2)
1.0
1.0
1.0
1.0
1.0
1.0
Application of the decision making methodology recommended in this PCS can result in inconclusive
results regardless of whether decisions are based on L-shell readings, K-shell readings, or both.
This instrument displays its lead-based paint measurements as both L-shell and K-shell readings based on
the corresponding L-shell and K-shell X-ray fluorescence (refer to Chapter 7 of the HUD Guidelines for
more details). The L-shell readings (or L-readings) are displayed as a numerical result alone, or as a
numerical result preceded by either one greater-than symbol (">") or preceded by two greater-than symbols
("»"). The two greater-than symbols will only be displayed when the detected lead level is greater than 5.0
mg/cm2. Since the maximum lead level reported by this instrument is 5.0 mg/cm2, lead levels greater than
5.0 mg/cm2 are displayed as "»5.0". Other examples of how L-readings can be displayed (in mg/cm2 units)
are "0.6" and ">0.9". The numerical display alone implies that the instrument measured the lead in the paint
at the displayed level using L-shell X-ray fluorescence; 0.6 mg/cm2 in the example. A number preceded by
a single greater-than symbol indicates that the measurable lead is deeply buried in the paint and the detected
lead level is greater than the displayed value. In the example, >0.9 indicates that the instrument detected lead
deeply buried in paint at a level greater than 0.9 mg/cm2. K-shell readings (or K-readings) are displayed in
one of two ways: 1) as a single K-reading plus and minus a "precision" value or 2) as an upper K-reading and
lower K-reading.
The algorithm used for testing in multifamily housing differs slightly from that used in single-family housing.
This is because the recommended number of readings per testing combination varies between the two types
of housing. (A testing combination is a location on a painted surface as defined in Chapter 7 of the HUD
Guidelines). In multifamily housing, the HUD Guidelines recommend taking a single XRF reading on a testing
combination. In single-family housing, three XRF readings are recommended on each testing combination.
MULTIFAMILY HOUSING XRF RESULT CLASSIFICATIONS
A. Take a single 20-second nominal reading on each testing combination.
B.
Classify the L-reading based on the type of information displayed.
If two oreater-than symbols are displayed then:
• Classify the »5.0 L-reading as POSITIVE
3 of 7
Figure D-8 continued.
XRF Performance Characteristic Sheet for the Niton XL-309
Spectrum Analyzer.
D-108
-------
APPENDIX D: XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
Niton Corporation; XL-309 Spectrum Analyzer
If one areater-than symbol is displayed then:
• Classify the L-reading as POSITIVE if the numerical result that follows the greater than symbol
is equal to or greater than 1.0.
• Classify the L-reading as INCONCLUSIVE if the numerical result that follows the greater than
symbol is less than 1.0.
If the numerical L-readina is displayed alone (that is. without any preceding areater-than symbols)
then:
• Classify the L-reading as POSITIVE if the numerical result is equal to or greater than 1.0.
• Classify the L-reading as NEGATIVE if the numerical result is less than 1.0.
C. Resolution of results classified as inconclusive.
All results classified as inconclusive above require further investigation. Take a 120-second nominal
XRF reading and use the K-shell reading. In multifamily housing, resolve the inconclusive
classification with a single K-shell reading or laboratory analysis as described below.
• Classify the result as POSITIVE if either the K-reading minus the displayed precision value or
the lower K-reading is equal to or greater than 1.0.
• Classify the result as NEGATIVE if either the K-reading plus the displayed precision value or the
upper K-reading is less than 1.0.
Classify the result as INCONCLUSIVE if neither of the above decision rules using the K-reading
provided a classification which can occur when the upper K-reading is equal to or greater than
1.0 or the lower K-reading is less than 1.0.
To resolve a remaining INCONCLUSIVE classification, remove a paint-chip sample and have it
analyzed in a laboratory as described in Chapter 7 of the HUD Guidelines.
SINGLE-FAMILY HOUSING XRF RESULT CLASSIFICATIONS:
D. Take three 20-second nominal readings on each testing combination.
E. Classify each L-reading using the methodology described in item A under Multifamily Housing XRF
Result Classifications.
F. Classification of the XRF result for a given testing combination is obtained by combining the individual
results of the three L-shell readings as follows:
• A POSITIVE classification is obtained if at least two of the three individual L-readings are
classified as positive.
4 of 7
Figure D-8 continued. XRF Performance Characteristic Sheet for the Niton XL-309
Spectrum Analyzer.
D-109
-------
APPENDIX D: XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
Niton Corporation; XL-309 Spectrum Analyzer
• A NEGATIVE classification is obtained if at least two of the three individual L-readings are
classified as negative.
• An INCONCLUSIVE classification is obtained if at least two of the three individual L-readings are
classified as inconclusive or if one L-reading is classified as positive, another is classified as
negative, and the third is classified as inconclusive.
G. Resolution of results classified as inconclusive.
Any results classified as inconclusive require further investigation in the same manner as described above
for multifamily housing with one exception. Take three 120-second nominal K-readings instead of a single
one. Obtain a classification by combining the individual results of the three K-readings. Resolve the
inconclusive classification by classifying the combined K-shell readings or with laboratory analysis as
described below.
• A POSITIVE classification is obtained if at least two of the three individual K-readings where
classified as positive.
A NEGATIVE classification is obtained if at least two of the three individual K-readings where
classified as negative.
• An INCONCLUSIVE classification is obtained if at least two of the three individual K-readings
where classified as inconclusive.
To resolve a remaining INCONCLUSIVE classification, remove a paint-chip sample and have it
analyzed in a laboratory as described in Chapter 7 of the HUD Guidelines.
INSTRUCTIONS FOR EVALUATING XRF TESTING:
Chapter 7 of the HUD Guidelines recommends several options for evaluating XRF testing. Among those
options is the following procedure which may be used after XRF testing has been completed. In single-family
housing, an XRF result is the average of three readings taken on a testing combination. (A testing
combination is a location on a painted surface as defined in Chapter 7 of the HUD Guidelines). In multifamily
housing, an XRF result is a single reading taken on a testing combination. If a multifamily housing
development is being retested, randomly select two units from within the development from which the ten
testing combinations should be randomly selected.
Randomly select ten testing combinations for retesting from each house or from the two selected units.
Conduct XRF retesting at the ten testing combinations selected for retesting.
Determine if the XRF testing in the units or house passed or failed the test by applying the steps below.
Compute the Retest Tolerance Limit by the following steps:
Determine XRF results for the original and retest XRF readings. Do not correct the original or
5 of 7
Figure D-8 continued. XRF Performance Characteristic Sheet for the Niton XL-309
Spectrum Analyzer.
D-110
-------
APPENDIX D: XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
Niton Corporation; XL-309 Spectrum Analyzer
retest results for substrate bias. In single-family housing a result is defined as the average of
three readings. In multifamily housing, a result is a single reading. Therefore, there will be ten
original and ten retest XRF results for each house or for the two selected units.
Compute the average of the original and re-test result for each of the ten testing combinations.
Square the average for each testing combination.
Add the ten squared averages together. Call this quantity C.
Multiply the number C by 0.0072. Call this quantity D.
Add the number 0.032 to D. Call this quantity E.
Take the square root of E. Call this quantity F.
Multiply F by 1.645. The result is the Retest Tolerance Limit.
Compute the overall average of all ten retest XRF results over all ten testing combinations selected
for retesting.
Take the difference of the overall average of the ten original XRF results and the overall average of
the ten retest XRF results. If the difference is negative, drop the negative sign.
If the difference of the overall averages is less than the Retest Tolerance Limit, the inspection has
passed the retest. If the difference of the overall averages equals or exceeds the Retest Tolerance
Limit, this procedure should be repeated with ten new testing combinations. If the difference of the
overall averages is equal to or greater than the Retest Tolerance Limit a second time, then the
inspection should be considered deficient.
Use of this procedure is estimated to produce a spurious result approximately 1 % of the time. That is, results
of this procedure will call for further examination when no examination is warranted in approximately 1 out of
100 dwelling units tested.
BIAS AND PRECISION:
These bias and precision data were computed without substrate correction using the using 20-second
L-shell readings from samples with reported laboratory results less than 4.0 mg/cm2 lead. Readings reported
by the instrument in the ">x" or "»x" format were not used in the computation. There were 15 test locations
with a laboratory reported result equal to or greater than 4.0 mg/cm2 lead. Of these, 12 readings were
reported in the ">x" or "»x" format, but of the 3 remaining, 1 had an XRF reading less than 1.0 mg/cm2.
These data are for illustrative purposes only. Substrate correction is not recommended for this XRF
instrument. Units are in mg/cm2.
6 of 7
Figure D-8 continued. XRF Performance Characteristic Sheet for the Niton XL-309
Spectrum Analyzer.
D-111
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
XRF PERFORMANCE CHARACTERISTIC SHEET
Niton Corporation; XL-309 Spectrum Analyzer
MEASURED AT
0.0 mg/cm2
0.5 mg/cm2
1 .0 mg/cm2
2.0 mg/cm2
SUBSTRATE
All
All
All
All
BIAS (mg/cm2)
0.0
0.0
0.0
-0.1
PRECISION* (mg/cm2)
<0.1
0.2
0.3
0.5
'Precision at 1 standard deviation
A document titled Methodology for XRF Performance Characteristic Sheets provides an explanation of
the statistical methodology used to construct the data in the sheets and provides empirical results from using
the recommended inconclusive ranges or thresholds for specific XRF instruments. For a copy of this
document call the National Lead Information Center Clearinghouse at 1-800-424-LEAD.
This XRF Performance Characteristic Sheet is a joint product of the U.S. Environmental Protection Agency (EPA) and the U.S.
Department of Housing and Urban Development (HUD). The issuance of this sheet does not constitute rulemaking. The
information provided here is intended solely as guidance to be used in conjunction with Chapter 7 of the Guidelines for the
Evaluation and Control of Lead-Based Paint Hazards in Housing EPA and HUD reserve the right to revise this guidance.
Please address questions and comments on this sheet to: Director, Office of Lead-Based Paint Abatement and Poisoning
Prevention, U.S. Department of Housing and Urban Development, Room B-133,451 Seventh St, S.W., Washington, DC 20410.
7 of 7
Figure D-8 continued.
XRF Performance Characteristic Sheet for the Niton XL-309
Spectrum Analyzer.
D-112
-------
APPENDIX D: XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
Table D-55. Bias Estimates, and Their Standard Errors, of Niton XL-309 Spectrum Analyzer Readings.
MEASURED AT
0.0 mg/cm2
0.5 mg/cm2
1 .0 mg/cm2
2.0 mg/cm2
CORRECTED?
No
No
No
No
SUBSTRATE
All
All
All
All
BIAS (mg/cm2)
0.0
0.0
0.0
-0.1
STANDARD ERROR FOR BIAS
0.06
0.03
0.05
0.12
Table D-56. Precision Estimates, and Their Standard Errors, of Niton XL-309 Spectrum Analyzer.
MEASURED AT
0.0 mg/cm2
0.5 mg/cm2
1 .0 mg/cm2
2.0 mg/cm2
CORRECTED?
No
No
No
No
SUBSTRATE
All
All
All
All
PRECISION' (mg/cm2)
<0.1
0.2
0.3
0.5
STANDARD ERROR
FOR PRECISION
0.00
0.02
0.03
0.04
'Precision at 1 standard deviation
D-113
-------
APPENDIX D:
XRF PERFORMANCE CHARACTERISTIC SHEETS AND RELATED
RESULTS
Table D-57. Classification Results For the Niton XL-309 20-Second L-Shell Readings (Uncorrected) and 120-
Second K-Shell Readings (Uncorrected), Classified Using the Threshold Values and
Methodology Reported in the XRF Performance Characteristic Sheet and Compared to
Laboratory Results in mg/cm2 Lead Classified Using the 1.0 mg/cm2 Lead Federal Standard For
Data Taken From Archive Testing.
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster*
Wood
TOTAL
THRESHOLD
1.0
1.0
1.0
1.0
1.0
1.0
FALSE POSITIVE
RATE
0.0% (0/1)
0.0% (0/2)
0.0% (0/14)
0.0% (0/27)
0.0% (0/29)
5.0% (2/40)
1.8% (2/1 13)
FALSE NEGATIVE
RATE
0.0% (0/2)
(0/0)
(0/0)
9.1% (1/11)
40.0% (2/5)
0.0% (0/23)
7.3% (3/41)
INCONCLUSIVE
RATE'
0.0% (0/3)
0.0% (0/2)
2 1.4% (3/1 4)
5.3% (2/38)
1 1 .8% (4/34)
14.3% (9/63)
11. 7% (18/1 54)
'For this table, readings were classified following the procedures stated in the PCS for this instrument. The inconclusive rate
was defined as the percentage of samples tested which were classified as inconclusive after both a 20-second L-shell reading
and a 1 20-second K-shell reading. However, in the testing of this instrument, 18 samples were classifed as inconclusive when
tested by the 20-second L-shell reading, and none of these 18 samples were resolved (classified) by the 1 20-second K-shell
reading.
'Testing of four additional plaster samples was completed at the National Institute of Standards and Technology with the same
instrument that was used for testing at the archive site. The inclusion of these four additional plaster samples has the following
affect in the table. The false negative rate for plaster decreased to 33.3% (3 out of 9) and the overall false negative rate
increased to 8.9% (4 out of 45). The false positive and inconclusive rates are unchanged. Specifically, one L-shell reading
classified a sample as negative that had a laboratory ICP result greater than 1 .0 mg/cm2 lead and another L-shell reading
correctly classified a second sample as positive. The two remaining samples were classified by the 20-second L-shell readings
as inconclusive, which, in turn, were correctly classified as positive by 120-second K-shell readings.
D-114
-------
APPENDIX E: XRF PERFORMANCE CHARACTERISTIC SHEET: AN EXAMPLE
E.1 Introduction
The XRF Performance Characteristic Sheets (PCSs) currently available from the
National Lead Information Center (1-800-424-LEAD) are duplicated in appendix D. In
this appendix, an example XRF Performance Characteristic Sheet illustrates the
minimal amount of information needed to test for lead-based paint according to Chapter
7 of the HUD Guidelines for the Evaluation and Control of Lead-Based Paint Hazards in
Housing. The information provided in this example PCS is strictly for illustrative
purposes only and is not intended to be used as a PCS for field testing. The model in
this appendix is proposed as a possible new format for the XRF Performance
Characteristic Sheets after publication of this report.
E.2 Example
The example PCS is shown on the following pages.
E-1
-------
APPENDIX E: XRF PERFORMANCE CHARACTERISTIC SHEET: AN EXAMPLE
XRF PERFORMANCE CHARACTERISTIC SHEET (PCS)
ABC Company, Inc.; XRF I
EFFECTIVE DATE: May 1, 1997
(MANUFACTURER AND MODEL:
Make: ABC Company, Inc.
Model: XRF I
Source: Co57
Note: This sheet supersedes all previous sheets for the XRF instrument of tftBBtalW, model,
and source shown above. '._,
USE AND EVALUATION DATA SOURCE:
This sheet is supplemental information to be used in <
apter 7 of the HUD Guidelines
Guidelines"). It is
for the Evaluation and Control of Lead-Based Paint Hazards
important that the user of this sheet be familiar with Chapter 7 of th^ tftfc<|«jfj*8nes. Performance
parameters shown in this sheet are based on EPA/HUD evaluation using aiefUved building components.
OPERATING PARAMETERS AND INSTRUCTIONS:
Performance parameters shown in this PCS at* appBeatJle only if the instrument is operated under
the same conditions as in the evaluation W^y^^i^ft^i^iedures described in Chapter 7 of the
HUD Guidelines: - -' v J^'h''?
XRF results, as defined in Chapter 7of tt» HyQ Guidelines, should be calculated using all digits
that are reported by the instrument
XRF results obtained in
or 30-second reading times, are
mode with either nominal 20-second
sheet
Calibration checks are performed Using rtSttirtal 20-second standard mode readings and
the red (1.02 mg/cm2)NIST Standard Reference Material (SRM No. 2579) paint film
Substrate
covered
ion values
(1.02mg/a
fated from XRF readings obtained on bare substrate
paint film
imeters can be applied to the Multifamily Decision Flowchart
id on a component type in murtifamily housing
:| joint product of the U.S. Environmental Protection Agency (EPA) and the U.S.
(HUD). The issuance of this sheet does not constitute rutomaking. The
as guidance to be used in conjunction with Chapter 7 of the Guidelines for On
Hazards in Housing. EPA and HUD reserve the right to revise this guidance.
on this sheet to: Director, Office of Lead-Based Paint Abatement and Poisoning
and Urban Development, Room B-133,451 Seventh St, S.W., Washington, DC
1of3
Figure E-1. Example PCS.
E-2
-------
APPENDIX E: XRF PERFORMANCE CHARACTERISTIC SHEET: AN EXAMPLE
XRF PERFORMANCE CHARACTERISTIC SHEET (PCS)
ABC Company, Inc.; XRF I
INSTRUMENT CALIBRATION CHECK:
Calibration check readings are taken in standard mode with nominal 20-second reading times. If the
observed calibration check average minus 1.02 mg/cm2 is greater than the PLUS VALUE, or less than the
MINUS VALUE, the manufacturer's instructions should be followed to bring the instrument into control
before additional testing is done. This calibration check is designed to minimize the chance of rejecting
a properly calibrated instrument: only about one out of every 200 times this procedure is followed.
MINUS VALUE = -0.4 mg/cm2
PLUS VALUE = +0.2 mg/cm2
SUBSTRATE CORRECTION.
For XRF results below 4.0 ma/cm2, substrate correction is recommended for
Metal and Wood , ;
Substrate correction is not recommended for:
Brick, Concrete, Drywall, and Plaster ^C
Substrate correction value computation:
XRF results are corrected for substrate bias by subtracting from each XRF result a correction value
determined separately for each house (single-family housing) or for each development (multifamily
housing) for each substrate type. The correction values are 'computed as follows:
• Using the same XRF instrument, take three readings on a bare substrate area covered with the red
NIST SRM (1 .02 mg/cm2) paint film. Repeat this procedure by taking three more readings on a
second bare substrate area, of the same substrate, covered with the red NIST SRM (1 .02 mg/cm2)
paint film. - ,, s , ;
• Compute the correction value for each substrate type that requires substrate correction by computing
the average of all six readings as shown below. ,
For each substrate type:
5'* + 6"- Reading _ 1 Q2mglcm*
2 of 3
Figure E-1 continued. Example PCS.
E-3
-------
XRF PERFORMANCE CHARACTERISTIC SHEET (PCS)
ABC Company, Inc.; XRF I
LBP CLASSIFICATION USING INCONCLUSIVE RANGES AND THRESHOLD VALUES.
XRF results are classified using either a threshold or an inconclusive range. With an inconclusive
range, XRF results are classified as positive if they are greater than or equal to the upper bound of the
inconclusive range, as negative if they are less than or equal to the lower bound, and as inconclusive
otherwise. With a threshold, XRF results are classified as positive if they are greater than or equal to the
threshold, and as negative if they are less than the threshold. There is no inconclusive classification with
a threshold. If substrate correction is recommended, always correct XRF results before using the
inconclusive ranges and thresholds shown below.
QUICK MODE
RPAOINft DF^f^RIPTION
Results corrected for substrate bias
on metal and wood substrates only
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
INCONCLUSIVE RANGE (mg/cm2)
LOWER BOUND
0.8
0.8
0.7
0.8
0.8
0.8
UPPER BOUND
09
0.9
0.9
0.9
0.9
0.9
20-SECOND OR 30-SECOND STANDARD MODE
READING DESCRIPTION
Results corrected for substrate bias
on metal and wood substrates only
SUBSTRATE
Brick
Concrete
Drywall
Metal
Plaster
Wood
THRESHOLD
(mg/cm2)
0.8
0.8
0.8
0.9
0.9
0.9
ADDITIONAL INFORMATION;
For a listing of laboratories recommended by the EPA National Lead Laboratory Accreditation Program
(NLLAP) or for the document tilted Methodology for XRF Performance Characteristic Sheets, which
describes the statistical methodology used to construct the data in the sheets and provides empirical
results from using the recommended inconclusive ranges or thresholds for specific XRF instruments, call
the National Lead Information Center Clearinghouse at 1-800-424-LEAD.
3 of 3
Figure E-1 continued. Example PCS.
E-4
-------
50272-101
REPORT DOCUMENTATION
PAGE
1. REPORT NO
EPA 747-R-95-008
3. Recipient's Accession No
4 Title and Subtitle
METHODOLOGY FOR XRF PERFORMANCE CHARACTERISTIC SHEETS
5. Report Date
September 1997
7. Author(s)
Cox, D.C.; Haugen, M.M.; Koyak, R.A.; Schmehl, R.L.
8 Performing Organization Rept. No
9. Performing Organization Name and Address
QuanTech, Inc.
1911 North Fort Myer Drive, Suite 1000
Rosslyn, Virginia 22209
10. Project/Task/Work Unit No
11. Contract © or Grant (G) No
68-D3-0004
12. Sponsoring Organization Name and Address
U.S. Environmental Protection Agency
Office of Pollution Prevention and Toxics
Washington, DC 20460
13. Type of Report & Period Covered
Technical Report
14.
15. Supplementary Notes
In addition to the authors listed above, the following key staff members were major contributors to the study:
Thomas Kelly of Battelle; Paul Constant, Jack Balsinger, Bruce Die), Dennis Hooton, and Gary Wester of Midwest
Research Institute; and Gary Dewalt of QuanTech.
16. Abstract (Limit: 200 words)
Information collected from an EPA/HUD field study conducted in 1993 provided background for Chapter 7
of the HUD Guidelines for the Evaluation and Control of Lead-Based Paint Hazards in Housing. A primary
conclusion of the field study is that testing by K-shell XRF instruments, with laboratory confirmation of
inconclusive XRF results, and with substrate correction in cases where this is effective in reducing bias, is a
viable way to test for lead-based paint. This approach can produce satisfactory results for classifying the
paint on architectural components using the federal threshold of 1.0 mg/cm2. This information is included
in Chapter 7. Since it was anticipated that there would be ongoing improvements XRF instrumentation as
the demand for testing for lead-based paint increased, Chapter 7 was written to defer to easily updated
documents for providing testing guidance for specific instruments called XRF Performance Characteristic
Sheets (PCSs). This report describes the methodology used to develop the PCSs.
17. Document Analysis a. Descriptors
Lead-based paint, lead-based paint testing, field evaluation, archive testing, recommendations for testing for lead in
paint, XRF Performance Characteristic Sheets, PCS, PCSs.
b. Identifiers/Open-Ended Terms
X-ray fluorescence instrument, XRF instrument, portable XRF, inductively coupled plasma-atomic emission
spectrometry, ICP-AES, ICP
c. COSATI Field/Group
18. Availability Statement
19. Security Class (This Report)
Unclassified
20 Security Class (This Page)
Unclassified
21. No. of Pages
227
22. Price
(SeeANSI-Z39.18)
OPTIONAL FORM 272 (4-77)
(Formerly NTIS-35)
Department of Commerce
------- |