*¦ _
WJ
PRO"^
TECHNICAL ASSISTANCE DOCUMENT
FOR SAMPLING AND ANALYSIS OF
OZONE PRECURSORS FOR THE
PHOTOCHEMICAL ASSESSMENT
MONITORING STATIONS PROGRAM -
Revision 2 - April 2019
-------
EPA-454/B-19-004
April 2019
TECHNICAL ASSISTANCE DOCUMENT FOR SAMPLING AND ANALYSIS OF OZONE
PRECURSORS FOR THE PHOTOCHEMICAL ASSESSMENT MONITORING STATIONS
PROGRAM - Revision 2 - April 2019
U.S. Environmental Protection Agency
Office of Air Quality Planning and Standards
Air Quality Assessment Division
Research Triangle Park, NC
li
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
DISCLAIMER
The statements in this document, with the exception of referenced requirements, are intended
solely as guidance. This document is not intended, nor can it be relied upon, to create any rights
enforceable by any party in litigation with the United States. The Environmental Protection
Agency (EPA) may decide to follow the guidance provided in this document, or to act at
variance with the guidance based on its analysis of the specific facts presented.
Mention of commercial products or trade names should not be interpreted as endorsement.
Some types of instruments currently in use may be described in text or in example figures or
tables. Sometimes these products are given as a typical and perhaps well-known example of the
general class of instruments. Other instruments in the class are available and may be fully
acceptable.
iii
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
CONTENTS
1.0 INTRODUCTION 1
1.1 Scope and Purpose 1
1.2 Overview of TAD Sections 2
1.2.1 Notable Changes from the 1998 TAD 2
1.3 Background 3
1.4 References 4
2.0 UPDATED REGULATIONS 5
2.1 PAMS Required Sites - Collocation with NCore 5
2.2 PAMS Parameters 6
2.3 References 9
3.0 DATA QUALITY PLANNING AND QUALITY AS SURANCE 10
3.1 Data Quality Obj ectives 10
3.2 Data Quality Indicators 10
3.3 Measurement Quality Objectives 11
3.3.1 Representativeness 12
3.3.1.1 Temporal Representativeness 12
3.3.1.2 Spatial Representativeness - Chemical Measurement Probe
Siting Criteria 13
3.3.1.2.1 Inlet Probe Height 13
3.3.1.2.2 Spacing from Obstructions 14
3.3.1.2.3 Spacing from Trees 14
3.3.1.2.4 Spacing from Roadways 14
3.3.1.3 Spatial Representativeness - Meteorological Parameters 14
3.3.2 Completeness 16
3.3.2.1 Make-up Sample Policy - Carbonyls Only 17
3.3.3 Precision 18
3.3.4 Bias 20
3.3.4.1 Assessing Laboratory Bias 21
3.3.4.2 Assessing Field Measurement Bias 21
3.3.4.2.1 Field Site Proficiency Testing for Speciated VOCs 21
3.3.4.2.2 Assessing Field Bias for Carbonyls 22
3.3.4.2.3 Ongoing Bias Assessment for Speciated VOCs and
Continuous Gas Monitors 22
3.3.4.2.4 Through-the-Probe Auditing 23
3.3.5 Sensitivity 23
3.3.5.1 Method Detection Limits 24
3.3.5.1.1 Frequency of Method Detection Limit
Determination 28
3.3.5.1.2 MDL Measurement Quality Obj ectives 29
3.3.5.1.3 Determining MDLs via 40 CFRPart 136 Appendix
B - Method Update Rule 29
3.4 Quality Assurance Project Plan 34
iv
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
3.4.1 Development of the National PAMS Required Site Program QAPP 35
3.4.2 PAMS Required Site QAPP Program Deviations 36
3.5 Standard Operating Procedures 36
3.6 Good Scientific Practices 38
3.6.1 Data Consistency and Traceability 38
3.7 References 38
4.0 VOLATILE ORGANIC COMPOUNDS BY AUTO-GC 40
4.1 Priority and Optional Volatile Organic Compounds 40
4.2 Instrumentation - Measuring VOCs with an Auto Gas Chromatograph with
Flame Ionization Detection 42
4.2.1 Summary of Method 42
4.2.2 Sample Introduction and Collection 43
4.2.2.1 Probe Inlet 43
4.2.2.2 Sample Collection Requirements 45
4.2.3 Automatic Gas Chromatograph (Auto-GC) 45
4.2.3.1 Instrument Sensitivity 46
4.2.3.2 Moisture Management 46
4.2.3.3 Thermal Desorption 48
4.2.3.4 Separation of Compounds 50
4.2.3.5 Flame Ionization Detection 50
4.2.4 Compound Identification 51
4.2.4.1 Compound Retention Time 51
4.2.4.2 Signal-to-Noise Ratio 53
4.2.5 Auto-GC Data File Naming 54
4.3 Method Detection Limits for Auto-GC 55
4.3.1 MDL Blank Component, MDLb 56
4.3.2 MDL Standard Spike Component, MDLsp 56
4.3.3 Redetermination ofMDLs 59
4.4 Auto-GC Interferences 59
4.4.1 Ozone Interference 59
4.4.2 Moisture 60
4.4.3 T emperature 60
4.4.4 Source-Oriented Interferences 61
4.4.5 Problematic Compounds for Auto-GC 62
4.5 Calibration of Auto-GCs 63
4.5.1 Standard Materials 63
4.5.1.1 Primary Calibration Standard 63
4.5.1.2 Secondary Source Calibration Verification Standard 63
4.5.1.3 Retention Time Standard 64
4.5.1.4 Zero Air 64
4.5.2 Retention Time Establishment and Calibration Convention and
Procedure 65
4.5.2.1 Static Dilution 66
4.5.2.2 Dynamic Dilution 68
4.5.2.3 Pulsed Standard Delivery 70
v
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
4.5.2.4 Humidification 71
4.5.3 Auto-GC Calibration Curves 72
4.5.4 Second Source Calibration Verification 73
4.6 Auto-GC Operation and Quality Control 73
4.6.1 System Blanks 76
4.6.2 Continuing Calibration Verification (CCV) 77
4.6.3 Precision Check 77
4.6.4 Retention Time Standard 77
4.7 References 80
5.0 CARBONYL COMPOUNDS VIA EPA COMPENDIUM METHOD TO-11A 83
5.1 General Description of Sampling Method and Analytical Method 83
5.2 Minimizing Bias 84
5.3 Carbonyls Precision 84
5.3.1 Sampling Precision 85
5.3.1.1 Collocated Sample Collection 85
5.3.1.2 Duplicate Sample Collection 86
5.3.2 Laboratory Precision 87
5.4 Managing Ozone 87
5.4.1 Copper Tubing Denuder/Scrubber 88
5.4.2 Sorbent Cartridge Scrubbers 89
5.4.3 Other Ozone Scrubbers 89
5.5 Collection Media 89
5.5.1 Lot Evaluation and Acceptance Criteria 89
5.5.2 Cartridge Handling and Storage 90
5.5.3 Damaged Cartridges 91
5.5.4 Cartridge Shelf Life 91
5.6 Carbonyls Method Detection Limits 91
5.6.1 C arb ony 1 s MDL Procedure 92
5.6.1.1 Selecting a Spiking Level 92
5.6.1.2 Preparing MDL Spikes and Blanks 93
5.6.1.3 Extraction and Analysis of MDL Spikes and Blanks 93
5.6.1.4 MDL Calculation 93
5.6.1.5 Ongoing Determination of MDLs 95
5.6.2 Example Carbonyls MDL Scenario and Calculation 96
5.7 Carbonyls Sample Collection Equipment, Certification, and Maintenance 97
5.7.1 Sampling Equipment 98
5.7.1.1 Sampling Unit Zero Check (Positive Bias Check) 98
5.7.1.2 Carbonyls Sampling Unit Flow Calibration 99
5.7.1.3 Moisture Management 100
5.7.2 Sampling Train Configuration and Presample Purge 100
5.7.3 Carbonyl Sampling Inlet Maintenance 101
5.8 Sample Collection Procedures and Field Quality Control Samples 101
5.8.1 Sample Collection Procedures 101
5.8.1.1 Sample Setup 101
5.8.1.2 Sample Retrieval 102
vi
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
5.8.1.3 Sampling Schedule and Duration 103
5.8.1.4 Carbonyls Sample Chain of Custody 103
5.8.2 Field Quality Control Samples 104
5.8.2.1 Field Blanks and Exposure Blanks 104
5.8.2.2 Trip Blanks 105
5.8.2.3 Collocated Samples 106
5.8.2.4 Duplicate Samples 106
5.8.2.5 Field Matrix Spikes 106
5.8.2.6 Breakthrough Samples 107
5.9 Carbonyls Extraction and Analysis 107
5.9.1 Analytical Interferences and Contamination 107
5.9.1.1 Analytical Interferences 107
5.9.1.2 Labware Cleaning 107
5.9.1.3 Minimizing Sources of Contamination 108
5.9.2 Reagents and Standard Materials 108
5.9.2.1 Solvents 108
5.9.2.2 Calibration Stock Materials 108
5.9.2.3 Secondary Source Calibration Verification Stock Materials 109
5.9.2.4 Holding Time and Storage Requirements 109
5.9.3 Cartridge Holding Time and Storage Requirements 109
5.9.4 Cartridge Extraction 109
5.9.4.1 Laboratory Extraction Batch Quality Control Samples 109
5.9.4.2 Cartridge Extraction Procedures 110
5.9.5 Analysis by HPLC Ill
5.9.5.1 Instrumentation Specifications Ill
5.9.5.2 Initial Calibration Ill
5.9.5.3 Secondary Source Calibration Verification Standard 112
5.9.5.4 Continuing Calibration Verification 112
5.9.5.5 Replicate Analysis 113
5.9.5.6 Compound Identification 113
5.9.5.7 Data Review and Concentration Calculations 114
5.10 Summary of Quality Control Parameters 116
5.11 References 119
6.0 OXIDES OF NITROGEN 120
6.1 NO/NOy 121
6.2 True NO; 121
6.2.1 Photolytic Conversion Chemiluminescent Detection NO2 Instruments 121
6.2.2 Cavity Attenuated Phase Shift (CAPS) Instruments 123
6.2.3 Cavity Ring-down Spectroscopy (CRDS) Instruments 124
6.2.4 True NO2 FEM Instrument Response 124
6.2.5 Minimizing Bias in NO2 Measurements 125
6.2.6 Generation of NO2 Standards 126
6.2.6.1 Gas Phase Titration 126
6.2.6.2 Dilution of Standard NO2 Gas 127
6.2.7 Calibration of True NO2 Instruments 128
vii
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
6.2.8 TrueN02 Sampling 128
6.2.9 Method Detection Limits for Continuous Gaseous Criteria Pollutant
Methods 128
6.2.9.1 Determining the MDLb 129
6.2.9.2 Determining the MDLsp 130
6.2.9.3 Calculating and Verifying the Instrument MDL 131
6.2.9.4 Ongoing Determination of the Instrument MDL 132
6.2.9.5 Example MDL Calculation for Continuous Gaseous Criteria
Pollutant Monitors 132
6.2.10 True NO2 Quality Control 134
6.3 NOy 135
6.4 References 136
7.0 OZONE 138
7.1 References 139
8.0 METEOROLOGY 140
8.1 Wind Speed and Wind Direction 140
8.2 Temperature 141
8.3 Relative Humidity 142
8.4 Solar Radiation 143
8.5 Ultraviolet Radiation 143
8.6 Barometric Pressure 144
8.7 Precipitation 144
8.8 Mixing Layer Height 144
8.8.1 Definition and Measurement of Mixing Layer Height 144
8.8.2 Ceilometer Theory of Operation 146
8.8.3 Ceilometer Siting and Installation 148
8.8.4 Ceilometer Operations 149
8.8.5 Ceilometer Mixing Height Calculations 149
8.8.6 Mixing Height Data Files and Data Validation 151
8.9 Quality Assurance/Quality Control for Meteorological Measurements 152
8.10 References 155
9.0 DATA HANDLING 156
9.1 Data Collection 156
9.1.1 Validation of Data Reduction and Transformation Systems and
Software 156
9.2 Data Backup 156
9.3 Recording of Data 157
9.3.1 Paper Records 157
9.3.2 Electronic Data Capture 157
9.3.3 Error Correction 157
9.3.3.1 Manual Integration of Chromatographic Peaks 158
9.4 Numerical Calculations 158
9.4.1 Rounding 159
viii
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
9.4.2 Calculations Using Significant Digits 159
9.4.2.1 Addition and Subtraction 159
9.4.2.2 Multiplication and Division 159
9.4.2.3 Standard Deviation 160
9.4.2.4 Logarithms 160
9.5 In-house Control Limits 160
9.5.1 Warning Limits 160
9.5.2 Control Limits 161
9.6 Negative Values 161
9.6.1 Negative Concentrations 161
9.6.2 Negative Physical Measurements 161
10.0 PAMS DATA VERIFICATION AND VALIDATION 162
10.1 Data Verification 164
10.1.1 Routine (Self) Review 165
10.1.2 Technical Review 167
10.2 Data Validation 168
10.2.1 Level 1 Data Validation 169
10.2.1.1 Identification of Outliers 170
10.2.2 Level 2 Data Validation 171
10.2.3 Level 3 Data Validation 171
10.3 Reporting of Validated Data to AQS 171
10.3.1 Reporting Values below Method Detection Limits 171
10.4 Data Validation Tools and Methods 172
10.4.1 Data Validation Visualization Methods 172
10.4.2 Data Validation Tools 178
10.5 Data Verification and Validation Records 179
10.6 Data Flagging 179
10.7 Data Verification and Validation of Speciated VOCs 179
10.7.1 Speciated VOCs Data Sources 180
10.7.1.1 Calibration Data 180
10.7.1.2 Auto-GC Reports and Datafiles 180
10.7.1.3 Chromatographic Data File Identification 181
10.7.1.4 Auto-GC Chromatograms 181
10.7.1.5 Instrument Maintenance and Site Logbooks 183
10.7.2 Speciated VOCs Data Verification Procedures 183
10.7.2.1 Correcting Chromatography Data 184
10.7.2.2 Routine Auto-GC Operator Checks 184
10.7.2.3 Technical Review of Speciated VOCs Data 185
10.7.3 Speciated VOCs Data Validation Procedures 191
10.7.3.1 Level 1 Data Validation 192
10.7.3.2 Level 2 Data Validation - Historical Data Comparisons 193
10.7.3.3 Level 3 Data Validation - Parallel Consistency Checks 193
10.7.4 Speciated VOCs Visualization Methods 193
10.7.4.1 Time Series Graphs 193
10.7.4.2 Scatter Plots 194
ix
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
10.7.4.3 Fingerprint Plots 194
10.7.4.4 Comparison with Other Parameters 194
10.8 Carbonyl Data Verification and Validation 195
10.8.1 Site Operator Verification Activities 196
10.8.2 ASL Verification and Validation Activities 196
10.8.2.1 ASL Sample Receipt 196
10.8.2.2 ASL Sample Extraction 197
10.8.2.3 ASL Sample Analysis 197
10.8.2.4 ASL Overall Technical Review 197
10.8.3 Carbonyls SLT Monitoring Agency Data Verification and Validation 200
10.8.3.1 Manual Inspection of Carbonyls Collection Data 200
10.8.3.2 Review of ASL Data 201
10.8.3.3 Review of Supporting QC Data 201
10.8.3.4 SLT Monitoring Agency Carbonyls Data Validation 202
10.8.3.4.1 Level 1 Carbonyls Data Validation 202
10.8.3.4.2 Level 2 Carbonyls Data Validation 204
10.8.3.4.3 Level 3 Carbonyls Data Validation 204
10.9 Data Verification and Validation of Ozone and Nitrogen Oxides 210
10.9.1 Ozone 211
10.9.2 Nitrogen Oxides, including True NO2 212
10.10 Verification and Validation of Routine Meteorological Measurements 212
10.10.1 Routine Meteorology Data Verification 212
10.10.1.1 Site Operator Routine Checks 213
10.10.1.2 Data Verification Performed by DAS 215
10.10.1.3 Technical Review of Meteorology Data 215
10.10.2 Meteorology Data Validation 216
10.10.2.1 Level 1 Validation of Meteorology Data 216
10.10.2.2 Level 2 Validation of Meteorology Data 217
10.10.2.3 Level 3 Validation of Meteorology Data 217
10.10.2.4 Reporting Validated Data to AQS 218
10.11 Using Surface Meteorology Measurements for Data Validation 218
10.12 References and Further Reading 218
11.0 REPORTING DATA TO AQS 220
11.1 Coding Ambient and Quality Assurance Data for AQS 220
11.2 Reporting PAMS Parameters to AQS 221
11.3 AQS Reporting Units 221
11.4 Corrections to Data Uploaded to AQS 222
11.5 AQS Qualifiers 222
11.5.1 AQS Qualification for Low Concentration Data 225
FIGURES
Figure 3-1. Graphical Representation of the MDL and Relationship to a Series of Blank
Measurements in the Absence of Background Contamination 25
X
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
Figure 3-2. Graphical Representation of the MDL and Relationship to a Series of Measurements
at the MDL Value 26
Figure 4-1. Determination of Chromatographic Peak Signal-to-Noise Ratio 54
Figure 4-2. Example Auto-GC Sampling Sequence for Ambient and QC Samples 76
Figure 5-1. Collocated and Duplicate Carbonyls Sample Collection 86
Figure 5-2. Qualitative Identification of Target Analytes 114
Figure 6-1. Schematic Diagram of Photolytic Chemiluminescence NO2/NO/NOX FEM 122
Figure 6-2. Simulated Squarewave LED Light before the Cavity and Attenuated Phase Shifted
Waveform after Passing through the Cavity 123
Figure 6-3. Schematic Diagram of Aerodyne CAPS NO2 Monitor 124
Figure 6-4. NO2 FRM and FEM Response Time 125
Figure 6-5. Calibration of CAPS NO2 Analyzer using NO2 Dilution Method 127
Figure 6-6. Summary of Commercially-Available NOy Analyzers 136
Figure 7-1. Simplified Representation of Tropospheric Ozone Chemistry Reactions and
Processes 139
Figure 8-1. Diurnal Variation of the Planetary Boundary Layer Structure 146
Figure 8-2. Vaisala CL51 Ceilometer 147
Figure 8-3. Example Vertical Backscatter Profile 148
Figure 8-4. Ceilometer Configuration 149
Figure 8-5. Example Graphical Display of Mixing Height using BL-View 151
Figure 10-1. Schematic of PAMS Data Generation, Verification, Validation, and Reporting... 164
Figure 10-2. Time Series Plot of Ethane 173
Figure 10-3. Scatter Plot of Propane and TNMOC 174
Figure 10-4. Fingerprint Plots of PAMS Target VOC Analytes 175
Figure 10-5. Stacked Bar Chart of Ethane, Propane, n-Butane, and n-Pentane 176
Figure 10-6. Example Box Plots of Formaldehyde Concentrations at Seven Sites 177
Figure 10-7. Example Meteorological Sensor Visual Checklist 214
TABLES
Table 2-1. NCore Station Parameters 5
Table 2-2. Priority and Optional PAMS Required Site Chemical Parameters 8
Table 2-3. PAMS Required Site Meteorological Parameters 9
Table 3-1. Data Quality Indicators and Associated Measurement Quality Objectives 11
Table 3-2. Minimum Distance for Inlet Probes to Roadways 15
Table 3-3. One-sided 99th Percentile Student's t Values 32
Table 3-4. PAMS Required Site National QAPP Elements 35
Table 4-1. PAMS Priority and Optional VOCs Measured by Auto-GC 40
Table 4-2. Example Auto-GC File Naming Convention 55
Table 4-3. Auto-GC Quality Control Standard Conventions 75
Table 4-4. Speciated VOCs Quality Control Parameters Summary 78
Table 5-1. Carbonyl Target Compounds Measured by Method TO-11A 84
Table 5-2. Maximum Background per Lot of DNPH Cartridge 90
Table 5-3. Example Carbonyls MDL Determination 96
Table 5-4. Carbonyls Field Blank Acceptance Criteria 105
xi
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
Table 5-5. Summary of Quality Control Parameters for Carbonyls Analysis 117
Table 6-1. Ambient Air Intercomparison Results for NO2 FEM (photolytic conversion) versus
FRM (molybdenum bed conversion) Method Reported by Beaver et al., 2013 122
Table 6-2. Example True NO2 MDL Determination 133
Table 6-3. Quality Control Parameters and Acceptance Criteria for True NO2 135
Table 8-1. Quality Control Parameters for Meteorology Measurements 154
Table 10-1. Speciated VOCs Data Screening Checks 186
Table 10-2. Major Sources of Carbonyls in the Atmosphere 195
Table 10-3. Carbonyls Data Validation Table 205
Table 10-4. Example Screening Criteria for Ozone 211
Table 10-5. Example Screening Criteria for N02/NO/NOx/NOy 212
Table 11-1. AQS Parameters and Recommended Reporting Units 222
Table 11-2. AQS Qualifiers for PAMS 223
Table 11-3. AQS Quality Assurance Qualifier Flags for Various Concentrations Compared to a
Laboratory's MDL and SQL 226
APPENDIX A: EPA ROUNDING GUIDANCE
APPENDIX B: AQS CODING GUIDANCE FOR PAMS QUALITY ASSURANCE DATA
xii
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
ACRONYMS
ACN acetonitrile
ADQ audit of data quality
AGL above ground level
AMTIC Ambient Monitoring Technology Information Center
ANP annual network plan
AQS Air Quality System
ASL analytical support laboratory
ASOS Automated Surface Observing System
auto-GC automatic gas chromatograph
BL-View Vaisala Boundary Layer View software
BTEX benzene, toluene, ethylbenzene, and total xylenes
C carbon
C2 compounds containing two carbon atoms
C6 compounds containing six carbon atoms
C12 compounds containing twelve carbon atoms
CAA Clean Air Act
CAPS cavity attenuated phase shift
CASAC Clean Air Scientific Advisory Committee
CBS A core-based statistical area
CCV continuing calibration verification standard
CDS chromatography data system
CFR Code of Federal Regulations
cm centimeter
CO carbon monoxide
COA certificate of analysis
COC chain of custody
CRDS cavity ringdown spectrometer
CSN chemical speciation network
CV coefficient of variation
DART Data Analysis and Reporting Tool
DAS data acquisition system
DDC dynamic dilution calibrator
DF dilution factor
DNPH 2,4-dinitrophenylhydrazine
DQI data quality indicator
DQO data quality objective
ECN effective carbon number
EMP enhanced monitoring plan
EPA United States Environmental Protection Agency
ESMB extraction solvent method blank
xiii
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
FB field blank
FID flame ionization detector
FEM federal equivalent method
FEP fluorinated ethylene propylene
FRM federal reference method
g gram(s)
GC gas chromatograph
GC-FID gas chromatograph with flame ionization detection
GPT gas phase titration
HAP hazardous air pollutant
HC hydrocarbon
HCF hydrocarbon-free
HPLC high performance liquid chromatograph
HVAC heating ventilation and air conditioning
ICAL initial calibration
IDL instrument detection limit
IMPROVE Interagency Monitoring of Protected Visual Environments
IPA instrument performance audit
KI potassium iodide
L liter(s)
LCS laboratory control sample
LCSD laboratory control sample duplicate
LED light-emitting diode
LIMS laboratory information management system
LPM liters per minute
M molar
m meter(s)
m3 cubic meter(s)
MB method blank
MDL method detection limit
MFC mass flow controller
mg milligram(s)
min minute(s)
mL milliliter(s)
ML minimum level
MLH mixing layer height
mm millimeter(s)
mM millimolar
MPV multi-point verification
xiv
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
MQO
measurement quality objective
MS
mass spectrometer
MUR
method update rule
Hg
microgram(s)
|iL
microliter(s)
|im
micrometer(s)
n
number
NAAQS
National Ambient Air Quality Standards
NATTS
National Ambient Air Toxics Trends Stations
NCore
National Core
ND
non-detect
netCDF
Network Common Data Form
ng
nanogram(s)
NIST
National Institute of Standards and Technology
nm
nanometer(s)
NO
nitrogen oxide
NO2
nitrogen dioxide
NOx
sum of nitrogen oxide and nitrogen dioxide
NOy
oxides of nitrogen with nitrogen atom charge > +2: sum of NO + NOx + NO;
NOz
oxides of nitrogen excluding NOx: NOy - NOx
NO A A
National Oceanic and Atmospheric Administration
NPAP
National Performance Audit Program
NPN
n-propyl nitrate
O2
oxygen molecule
O3
ozone molecule
OTR
Ozone Transport Region
PAMS
photochemical assessment monitoring station
PAMSHC
PAMS hydrocarbons
PAN
peroxyacetyl nitrate
PBL
planetary boundary layer
PBM
propane benzene mix
PDA
photodiode array
PFA
perfluoroalkoxy
PLOT
porous layer open tubular
PM
particulate matter
PM2.5
particulate matter with aerodynamic diameter <2.5 microns
PM10
particulate matter with aerodynamic diameter <10 microns
POC
parameter occurrence code
ppb
part(s) per billion
ppbC
parts per billion carbon
ppbV
part(s) per billion by volume
ppm
part(s) per million
ppt
part(s) per trillion
xv
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
PQAO
primary quality assurance organization
psi
pound(s) per square inch
psia
pound(s) per square inch absolute
PT
proficiency test
PTFE
polytetrafluoroethylene
QA
quality assurance
QAPP
quality assurance project plan
QC
quality control
QS
quality system
RAID
redundant array of independent disks
RAOB
The Universal RAwinsonde OBservation program
RH
relative humidity
RPD
relative percent difference
RSD
relative standard deviation
RF
response factor
RT
retention time
RTP
Research Triangle Park
RTS
retention time standard
SB
solvent blank
S:N
signal to noise
SLT
State, Local, and Tribal
SO2
sulfur dioxide
SOP
standard operating procedure
SQL
sample quantitation limit
sscv
second source calibration verification
STP
standard conditions of temperature and pressure
SYSB
system blank
TAD
technical assistance document
TD
thermal desorption
TNMOC
total non-methane organic carbon
TSA
technical systems audit
HP
through-the-prob e
UHP
ultra-high purity
UV
ultraviolet
VOC
volatile organic compound
ZAG
zero air generator
xvi
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
1.0 INTRODUCTION
1.1 Scope and Purpose
The Technical Assistance Document (TAD) for Sampling and Analysis of Ozone Precursors was
initially published in October 1991 to aid air monitoring agencies responsible for implementing
photochemical assessment monitoring stations (PAMS). Several revisions were made to the
initial TAD in 1994 and 1995 and were incorporated into Appendix N of the PAMS
Implementation Manual; a major revision to the TAD was published in September 1998, which
included modifications necessary following advances in the methodology for measuring
pollutants and meteorological parameters of interest for PAMS.
The purpose of this document is to provide guidance in support of the required monitoring and
associated measurements resulting from the revisions to 40 Code of Federal Regulations (CFR)
Part 58 Appendix D Section 5.0 related to ozone precursor monitoring. This scope includes
guidance and technical information to State, Local, and Tribal (SLT) air monitoring agencies
(henceforth referred to as "monitoring agencies") as well as to Environmental Protection Agency
(EPA) Regions responsible for measuring meteorological parameters and ozone precursors in
ambient air. Described herein are specific methods for the collection and analysis of speciated
volatile organic compounds (VOCs), speciated carbonyl compounds, "true" nitrogen dioxide
(NO2), and the local meteorology including temperature, relative humidity (RH), and wind
speed, among other parameters. This TAD describes the quality system for the monitoring
program, but not in detail. A separate quality assurance project plan (QAPP) is being developed
for monitoring agencies to utilize as a template to develop and prescribe aspects of quality
assurance/quality control (QA/QC) associated with collecting PAMS data.
Technical guidance presented in this TAD is a combination of lessons learned from experienced
PAMS operators and from experts responsible for assessing PAMS data over the past 15 years as
well as best practices gained from instrument manufacturers in development of new and updated
instruments. The technologies described are mature and have evolved since the inception of the
PAMS program such that their advantages and limitations are better understood. These
technologies will be under continual evaluation and improvement as data are gathered and
analyzed following the implementation of the PAMS Required Monitoring program in June
2019.
The updated regulations in 40 CFR Part 58 Appendix D Section 5h require that monitoring
agencies in states with moderate and above ozone non-attainment and states in the Ozone
Transport Region (OTR, which includes Connecticut, Delaware, the District of Columbia,
Maine, Maryland, Massachusetts, New Hampshire, New Jersey, New York, Pennsylvania,
Rhode Island, Vermont, and Virginia) develop enhanced monitoring plans (EMPs). While EMPs
may be developed utilizing much of the guidance in this TAD, development of EMPs is outside
the scope of this guidance document. EPA has provided additional guidance for monitoring
agencies to prepare an EMP at the following link:
https://www3.epa.eov/ttnam.til/files/am.bient/pams/PAMS%20EMP%20Guidance.pdf
1
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
1.2 Overview of TAD Sections
This document is organized to present guidance and information in the likely order in which they
are needed when establishing a PAMS Required Site and operating an analytical support
laboratory. The organization of this document follows the EPA's plan-do-check-act feedback
loop to facilitate continuous improvement to meet data quality objectives (DQOs) of the PAMS
Required Site program. Aspects of the program pertaining to planning are addressed first
followed by implementation and data collection, data verification, and data and program
assessment.
1. Background - Brief overview of the history of the PAMS Program, PAMS measurement
parameters, and noteworthy revisions from the 1998 TAD.
2. Planning - Discussion of aspects related to data quality planning and QA.
3. Chemical Parameters - Measurements of chemical constituents in ambient air -
determining method detection limits (MDLs) and measuring VOCs, carbonyls, oxides of
nitrogen, and ozone.
4. Meteorological Parameters - Description of surface and upper air measurements of
interest to ozone formation.
5. Data Handling - Procedures and policies for collection, manipulation, backup, archival,
calculations, verification, validation, and reporting.
1.2.1 Notable Changes from the 1998 TAD
This revision of the PAMS TAD incorporates many of the changes to the PAMS network since
the most recent revision of the TAD in 1998. Several of the updates to the PAMS network in that
time no longer apply, and are not addressed. The most important change incorporated involves
the elimination of the requirement for an array of upwind and downwind sites to be operated in
PAMS areas; this has been updated to instead prescribe a single urban PAMS monitoring site
collocated at the national core (NCore) site within the core-based statistical area (CBSA) per the
regulations promulgated in October 2015. Additionally, as better time resolution of ozone
precursors is desired, this TAD eliminates the volumes of guidance on collection of precursor
VOCs in canisters and the subsequent laboratory analysis. These details have been replaced by
guidance targeted to the setup and operation of auto-gas chromatographs (GCs) for hourly VOC
measurements.
For monitoring agencies conducting canister sampling and analysis for VOCs, the previous 1998
PAMS TAD can be consulted. Characterization of instrument sensitivity is also updated, and
includes determination of MDLs by the method update rule (MUR) promulgated in September
2017, which takes into account the method background and updates the MDL definition to
represent the lowest concentration detectable above background. Another important update has
been facilitated by new technologies that permit the specific measurement of "true" NO2 as
distinguished from NOx. Molybdenum conversion instruments reduce numerous nitrogen species
in addition to NO2 which results in overestimated concentrations of NO2. The new generation of
N02-specific instruments offers additional advantages in that they also operate with much faster
response time, which is important in measuring atmospheric NO2 when concentrations are
changing rapidly in ambient air. EPA has funded optimization of Compendium Method TO-11A
2
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
for measuring carbonyl concentrations, and this TAD includes updated guidance based on the
outcomes of the optimization work. Lastly, with the development and improvement in ceilometer
technology, characterization of the mixing layer height (MLH) is discussed within the
meteorology guidance in Section 8.8.
1.3 Background
The Clean Air Act (CAA) Amendments of 1990 required the EPA Administrator to promulgate
rules for monitoring to obtain comprehensive and representative data on air pollution. One result
of this was that the EPA promulgated a final rule in 40 CFR, Part 58 on February 12, 1993,
which required enhanced monitoring of ozone, oxides of nitrogen, VOCs, and selected carbonyl
compounds in ambient air and monitoring of meteorological parameters. The rule required states
to establish PAMS as part of their existing State Implementation Plan monitoring networks in
ozone non-attainment areas classified as serious, severe, or extreme.
The first PAMS sites began monitoring in 1994. Since that time, the ozone national ambient air
quality standards (NAAQS) shifted from a 1-hour standard to an 8-hour standard in 1997, with a
form based on the 3-year average of the annual fourth-highest daily maximum 8-hour average
ozone concentration. At the same time the standard shifted from a 1-hour to an 8-hour standard,
(July 1997), the NAAQS concentration was reduced from 0.12 parts per million (ppm) to 0.080
ppm. This was further reduced to 0.075 ppm in March 2008 and to 0.070 ppm in December
2015. The chief objective of the PAMS program is to provide a database of information on ozone
and its precursors to assist state and local air pollution control agencies in evaluating, tracking
the progress of, and if necessary, refining control strategies for attaining the ozone NAAQS. A
secondary objective is to utilize the PAMS data to prepare air quality trends, evaluate and refine
photochemical model performance, and assist state and local agencies in implementing
regulatory controls. Concurrent with the decrease of the ozone NAAQS, the national average
ozone concentration has decreased by approximately 30% between 1980 and 20091. While the
number of serious and above non-attainment areas has decreased, the number of non-attainment
areas remained nearly the same.
In April 2011, EPA published a white paper titled "White Paper on EPA's PAMS Network Re-
engineering project", in which EPA reviewed the PAMS program, which is responsible for
monitoring ambient air for chemical constituents responsible for contributing to ground-level
ozone (O3).2 The Clean Air Scientific Advisory Committee (CASAC) Air Monitoring and
Methods Subcommittee held public teleconferences through May and July 2011 to review the
PAMS Network Re-Engineering project. The PAMS program had generated a large quantity of
ozone precursor and meteorology data which the air monitoring community felt were
underutilized. EPA received requests from various PAMS stakeholders, including the National
Association of Clean Air Agencies and state and local monitoring agencies to review the PAMS
program to identify areas of improvement to make collected data more useful to the intended
users. Much of the equipment in use at PAMS sites was nearing or past the end of use cycle, and
it was sensible to re-evaluate the methods and equipment prior to appropriating funds to replace
equipment.
3
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
In the re-evaluation of the PAMS network begun in 2011, EPA identified the value of flexibility
for state and local agencies to determine the best way to address monitoring to:
1. Minimize redundancy while providing robust information for defining ozone gradients
2. Better target monitoring resources tailored to each area's ozone problem which is unique
based on the mix of sources, topography, and meteorology. The one-size fits all approach
is overly rigid and requires SLT agencies to expend resources that may offer little benefit
to their specific problem. Allowing the local and regional air quality agencies flexibility
to determine the appropriate monitoring plans for their network increases the likelihood
of developing effective control strategies.3
3. Focus measurements on PAMS VOCs which impact ozone formation. Following a
Northeast States for Coordinated Air Use Management review of approximately 15
years' worth of PAMS VOCs, it became clear that target compounds were rarely
measured at concentrations that would significantly impact ozone formation. As a result,
EPA performed a review of the existing PAMS target compound list to potentially revise
the list. This review evaluated whether compounds in the existing PAMS target
compound list could be eliminated from the list or made optional, due to the overall
reactivity adjusted average concentration, reactivity adjusted average concentration
during 9 a.m. morning rush hour on high ozone days, reactivity adjusted average
concentration based on geography, and whether the compound was a hazardous air
pollutant and/or a high priority secondary organic aerosol precursor. The resulting list of
target compounds was separated into priority compounds (mandatory monitoring
required) and optional compounds (recommended, but monitoring not required). For
optional compounds deemed to be important to the ozone precursor chemistry in a
specific region, the responsible agencies should monitor for those compounds. This
reduced list of priority compounds should allow agencies to reduce the costs associated
with collecting, evaluating, and reporting data for compounds which may not be relevant
to their specific region.
PAMS regulations in 40 CFR Part 58 Appendix D Section 5.0 were amended concurrently with
the revision to the 2015 ozone NAAQS reduction to 0.070 ppm to reflect the outcomes of the
2011 re-evaluation. PAMS monitoring for the Required Site network is to begin implementation
on June 1, 2019. This TAD describes the equipment, policies, and procedures to collect PAMS
measurements for the PAMS Required Site network.
1.4 References
1. EPA Air Trends website, www.epa.eov/airtrends/ozone.html
2. EPA-CASAC-11-010
https://vosemite.epa. eov/sab/sabproduct.nsf/8412C8765AE2BC80852579190072D70A/$File
/EPA-CASAC-11 -010-unsiei
3. US EPA. Ambient Air Monitoring Strategy for State, Local, and Tribal Air Agencies. Office
of Air Quality Planning and Standards. December 2008
and fluxes.
4
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
2.0 UPDATED REGULATIONS
The updated regulations in 40 CFR Part 58 Appendix D Section 5.01 promulgated in October
2015 prescribe the updates to the required PAMS monitoring associated with the revision to the
8-hour ozone NAAQS. These revised regulations standardize the operation of the PAMS
network at approximately 43 geographically separated PAMS Required Sites and require the
measurement of a common list of pollutants and meteorological parameters. These are described
in Section 2.2.
2.1 PAMS Required Sites - Collocation with NCore
The updated regulations require PAMS monitoring (a PAMS Required Site) at each NCore site
within a CBSA having a population of 1,000,000 persons or more. To meet the requirements in
the regulations promulgated in October 2015, all PAMS Required Sites are to be operational and
reporting quality assured and validated data for the required parameters to EPA's Air Quality
System (AQS) by June 1, 2019. As of the publication of this TAD, there are several "early
adopter" sites which have accelerated their monitoring timeline to allow them additional time for
developing their PAMS Required monitoring program. Guidance in this TAD has been revised
from that in the earlier draft versions and may be further revised and updated based on lessons
learned and best practices identified by these early adopter programs.
PAMS Required Sites are to be located at NCore Network stations within the CBSA unless a
waiver is granted by the EPA Regional Administrator. The NCore Network comprises 63 urban
and 17 rural sites which integrate advanced measurements for particles, pollutant gases, and
meteorology. Many NCore sites are formerly National Ambient Air Monitoring Stations. Most
NCore sites have been operating since the formal network start on January 1, 2011. Parameters
collected at NCore stations include those listed in Table 2-1.
Table 2-1. NCore Station Parameters 2
i sni' i it
Cumnu'iil^
PM2.5 speciation
organic and elemental carbon, major ions and trace metals
(24-hour average; every third day); part of the Interagency Monitoring of
Protected Visual Environments (IMPROVE) or Chemical Speciation
Network (CSN)
PM2.5 Federal Reference
Method (FRM) mass
24-hr average at least every third day
Continuous PM2.5 mass
1-hour reporting interval; federal equivalent method (FEM) or pre-FEM
monitors
PM10-2.5 mass
filter-based or continuous
Ozone (O3)
Carbon monoxide (CO)
Sulfur dioxide (SO2)
Nitrogen oxide (NO)
Total reactive nitrogen (NOy)
all gases through continuous FRM or FRM monitors - capable of trace levels
(low ppm and below) where needed
Surface meteorology
wind speed and direction (reported as "Resultant"), temperature, relative
humidity
5
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
Collocation of PAMS Required Sites at existing NCore stations allows monitoring organizations
and EPA to leverage existing infrastructure and monitoring agency policies and procedures, and
provides the ability to evaluate numerous collocated chemical and meteorological parameters. Of
particular importance for interpreting PAMS speciated precursor data are ozone and
meteorological data collection. Addition of PAMS monitoring at NCore sites increases the value
of the sites and establishes a wider national network to better characterize the ozone problem as
well as provide a more complete picture of air quality in the associated urban environments.
2.2 PAMS Parameters
The new regulations promulgated in October 2015 specify that the following chemical and
meteorological parameters will be measured at PAMS Required Sites minimally commencing
June 1 through August 31 of each year (sites are encouraged to monitor outside this period,
particularly at sites where ozone season extends before or after this three-month period).
Chemical measurement parameters and meteorological parameters are detailed in Table 2-2 and
Table 2-3, respectively.
o Ozone - Ozone measurements are already required at NCore monitoring stations.
Sites that elect to exercise the waiver option to locate the Required PAMS station at a
location other than the NCore station (i.e., alternate PAMS site) will be required to
monitor for ozone as prescribed for NCore monitoring stations. Each Required PAMS
Site is to continuously monitor for ozone and report the hourly averaged ozone
concentration.
o VOCs - All Required PAMS Sites are to measure the priority speciated VOCs
(classified as olefin, aromatic, paraffin, halogenated, monoterpene olefin, alkyne, or
alcohol) listed in Table 2-2 as well as the total non-methane organic carbon
(TNMOC). It is strongly suggested that all Required PAMS Sites take hourly
speciated VOC measurements with auto-GCs. Each Required PAMS Site is to report
the hourly averaged concentration of each priority compound VOC listed in Table 2-2
and is encouraged to report the hourly averaged concentration of the optional
compounds (note that carbonyls are denoted in the table and are to be measured by
EPA Method TO-11 A)3. There is a waiver option to allow collection of three 8-hour
canister samples every third day (as an alternative to hourly speciated VOC
measurements) at locations where auto-GCs may not be appropriate (e.g., where VOC
concentrations are too low or where the predominant VOCs may not be measurable
by the auto-GC technique) or for logistical or other programmatic constraints.
o Carbonyls - All Required PAMS Sites are to conduct carbonyl sampling with a
frequency of three sequential 8-hour samples on a one-in-three-day basis (-90
samples per PAMS sampling season). The regulations permit an alternative of
reporting hourly averaged formaldehyde concentrations; however, at publication of
this TAD, EPA has not formally evaluated such instrumentation to provide hourly
formaldehyde concentrations. Should such instruments be evaluated following
finalization of this TAD, EPA plans to communicate the appropriate guidance and
requirements for their operation. A complete list of the target carbonyl compounds
can be found in Table 2-2. The TO-11A method, as used in the National Ambient Air
Toxics Trends Stations (NATTS) program,4 will be used for PAMS Required Sites.
6
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
Sites may additionally elect to conduct episodic carbonyl measurements to provide
additional insight into ozone formation at that specific site as well as inputs for ozone
and air quality modelling.
o Oxides of Nitrogen - All Required PAMS Sites will monitor for NO and NOy (total
oxides of nitrogen) in addition to true NO2, where the latter will be measured with a
cavity attenuated phase shift (CAPS) spectroscopy direct-reading NO2 instrument, a
cavity ringdown spectrometer (CRDS) instrument, or a photolytic-converter NOx
analyzer. EPA has indicated that sites will employ true NO2 instruments with FRM or
FEM status. As of the time of release of this document, there are not currently CRDS
instruments approved as FRM or FEM.
o Meteorology Measurements - All Required PAMS Sites will measure the
meteorological parameters listed in Table 2-3. Although EPA is suggesting the use of
ceilometers for mixing layer height, other types of meteorological equipment that
provide for an indication of mixing layer height can be proposed in the monitoring
agency PAMS Implementation Plan appended to the monitoring agency's annual
network plan (ANP). Sites may apply for a waiver to allow meteorological
measurements to be obtained from other nearby sites (e.g., National Oceanic and
Atmospheric Administration [NOAA] Automated Surface Observing System [ASOS]
sites). Discussions with NOAA regarding ASOS site data indicate that the ceilometers
in use are not sufficiently sensitive and will not readily provide the MLH, therefore
monitoring agencies are encouraged to operate their own ceilometer.
7
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
Table 2-2. Priority and Optional PAMS Required Site Chemical Parameters
l'i'iiii il\ ('lienm al I'lii'iiiiH-U'i'o
(l\ii|uin(l)
AQS
1'arameler
( ode
( (impound
( la»>.
()pliuiial ( 'hemiral
I'araiiK'U'i'x
\ys
I'arameler
( ode
( iiiiipiiiind
( law
1,2,3-trimelhylbenzene
45225
aromatic
1,3,5-lrimclhylbcnzene
45207
aromatic
1,2,4-trimethylbenzene
45208
aromatic
1-pentene
43224
olefin
1-butene
43280
olefin
2,2-dimethylbutane
43244
paraffin
2,2,4-trimethylpentane
43250
paraffin
2,3,4-trimethylpentane
43252
paraffin
acetaldehyde
43503
carbonyl
2,3-dimethylbutane
43284
paraffin
benzene
45201
aromatic
2,3-dimethylpentane
43291
paraffin
eis-2-butene
43217
olefin
2,4-dimethylpentane
43247
paraffin
ethane
43202
paraffin
2-methylheptane
43960
paraffin
ethylbenzene
45203
aromatic
2-methylhexane
43263
paraffin
ethylene
43203
olefin
2-methylpentane
43285
paraffin
formaldehyde
43502
carbonyl
3-methylheptane
43253
paraffin
isobutane
43214
paraffin
3-methylhexane
43249
paraffin
isopentane
43221
paraffin
3-methylpentane
43230
paraffin
isoprene
43243
olefin
acetone
43551
carbonyl
m&p-xylenes
45109
aromatic
acetylene
43206
alkyne
m-ethyltoluene
45212
aromatic
cis-2-pentene
43227
olefin
n-butane
43212
paraffin
cyclohexane
43248
paraffin
n-hexane
43231
paraffin
cyclopentane
43242
paraffin
n-pentane
43220
paraffin
isopropylbenzene
45210
aromatic
o-ethyltoluene
45211
aromatic
m-diethlybenzene
45218
aromatic
o-xylene
45204
aromatic
methylcyclohexane
43261
paraffin
p-ethyltoluene
45213
aromatic
methylcyclopentane
43262
paraffin
propane
43204
paraffin
n-decane
43238
paraffin
propylene
43205
olefin
n-heptane
43232
paraffin
styrene
45220
aromatic
n-nonane
43235
paraffin
toluene
45202
aromatic
n-octane
43233
paraffin
trans-2-butene
43216
olefin
n-propylbenzene
45209
aromatic
ozone
44201
criteria pollutant
n-undecane
43954
paraffin
true NO2
42602
criteria pollutant
p-diethylbenzene
45219
aromatic
total non-methane organic carbon
43102
total VOCs, non-
methane
trans-2-pentene
43226
olefin
a-pinene
43256
monoterpene
olefin
fi-pinene
43257
monoterpene
olefin
1,3 butadiene
43218
olefin
benz aldehyde
45501
carbonyl
carbon tetrachloride
43804
halogenated
ethanol
43302
alcohol
tetrachloroethylene
43817
halogenated
8
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
Table 2-3. PAMS Required Site Meteorological Parameters
I'simmclcr
AQS I'iimineler Code
hourly averaged ambient temperature
62101
hourly vector-averaged wind direction
61104
hourly vector-averaged wind speed
61103
hourly averaged atmospheric pressure
64101
hourly averaged relative humidity
62201
hourly precipitation
65102
hourly averaged mixing layer height
61301
hourly averaged solar radiation
63301
hourly averaged ultraviolet radiation
63302
2.3 References
1. 40 CFR Part 58 Appendix D, available at (accessed March 2018): https://www.ecfr.gov/cgi-
bin/retrieveECFR?n=40y6.0.1.1.6#ap
2. NCore Multipollutant Monitoring Network website on EPA AMTIC (accessed March 2018):
https://www3.epa.gov/ttn/amtic/ncore.html
3. US EPA. Additional Revisions to the Photochemical Assessment Monitoring Stations
Compound Target List. October 2, 2017
https://www3.epa.gov/ttnamtil/files/ambient/pams/targetlist.pdf
4. NATTS Technical Assistance Document, October 2016, Available at (accessed March 2018):
https://www3.epa.gov/ttnamtil/files/am.bient/airtox/ \ \ II S%20TAD%2.0Revision%20?« R
NAL%20Qctober%20; If
9
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
3.0 DATA QUALITY PLANNING AND QUALITY ASSURANCE
The purpose of the PAMS Required stations network is to measure the concentrations of ozone
and its precursors (NOy and VOCs) and characterize the meteorological conditions under which
ozone precursors contribute to ozone formation in CBS As having populations greater than
1,000,000. The re-engineered PAMS network monitoring slated to begin June 1, 2019 builds on
the data collected and experience gained from the PAMS network initiated in 1994 which was
established to provide data to assist air monitoring agencies in evaluating, tracking the progress
of, and refining control strategies for attaining the ozone NAAQS. Ambient concentrations of
ozone precursors are used to track VOCs and NOx emission inventory reductions, better
characterize the nature and extent of the ozone problem and prepare air quality trends. The
database of PAMS data allows air quality modelers to evaluate photochemical model
performance, which is integral to adjusting current control strategies and developing future
effective and efficient control strategies.
3.1 Data Quality Objectives
DQOs are qualitative and quantitative statements derived from the DQO Planning Process that
clarify the purpose of the study, define the most appropriate type of information to collect,
determine the most appropriate conditions under which to collect that information, and specify
tolerable levels of potential decision errors.1 DQOs define the quality of and the acceptable
levels of uncertainty in the measurements. Stated another way, DQOs are statements describing
"how good" the measurements need to be to control decision risk(s) within known levels of
confidence and to ensure that collected data are of sufficient quantity and quality to be fit for the
stated purpose.
A formal DQO process was not undertaken to determine a PAMS Required Site DQO. Rather,
the measurement quality objectives (MQOs) for the various data quality indicators (DQIs) were
established based on the expertise of EPA modelers and data analysts and their data quality
needs for ozone and ozone precursor model evaluation and model inputs. Monitoring agencies
measuring PAMS parameters and other experts in PAMS measurements reviewed the proposed
MQOs to ensure they were reasonable and attainable. The MQOs prescribed herein will be
reevaluated and potentially revised once EPA modelers and data analysts work with the PAMS
data from the first year of the program (anticipated to be June through August 2019).
Additionally, if more sensitive or accurate measurement methods become available and are
deemed to be necessary to meet modelers' needs, the MQOs may be modified and refined to
accommodate the updated methods.
3.2 Data Quality Indicators
In order to achieve the data quality needs identified by EPA modelers and data analysis staff, the
MQOs in the next section were assigned to the following DQIs: representativeness,
completeness, precision, bias, and sensitivity.
10
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
3.3 Measurement Quality Objectives
DQIs of representativeness, completeness, precision, bias, and sensitivity are to meet specific
MQOs, or acceptance criteria. The MQOs for each of the DQIs are shown in Table 3-1. Note that
the MQOs for ozone and oxides of nitrogen (true NO2, NO, and NOy) will follow QA/QC
requirements prescribed in the CFR (40 CFR Part 50 and 40 CFR Part 58) and in the QA
Handbook (QA Handbook Volume II, Appendix D, January 2017).
Table 3-1. Data Quality Indicators and Associated Measurement Quality Objectives
Mi'llmri or Piiniim-k-r
l)QI
Mi'iisu mill-ills
Ri'l>ri'si-iil;ili\iiU'ss
(S;iiii|>lin75
True NO2 and
NO/NOy
Continuous, hourly
average
<±15.1% or
± 1.5 ppbd
whichever is
greater
< 15.1% or
1.5 ppbc
whichever is
greater
<0.001 ppm
>75
Ozone
< ± 7.1% or±
1.5 ppbd
whichever is
greater
<7.1% or 1.5
ppbc
whichever is
greater
< 0.002 ppm
> 90% (avg) daily
max available in
O3
season with min of
75% in any 1 year1
TO-11A (carbonyls)
Three sequential 8-
hour samples every 3rd
daye-f
<25g
< 15h
< 0.25 (ig/m3
>85
Ki'|)ivsiiil;ili\ I'lii'ss
(Siiinplin^
l"lV(|lll-|llA )¦'
lii;is
Piviisiiui
Si'iisiii\ ii\
(ki'siiluiidii)
ClIllipll-U-IK-SS (",'ii)
Ambient Temperature
<±0.5 "C
<0.1 "C
Relative Humidity
< ± 5% RH
< 0.5% RH
Barometric Pressure
< ± 3 hPa
<0.1 hPa
< ± 0.2 m/s or
Wind Speed
± 5%,
whichever is
greater
<0.1 m/s
Wind Direction
Continuous, hourly
average
< ± 5 degrees
not specified
< 1 degree
>75
Solar Radiation
<±5%
< 1 Watt/m2
UV Radiation
<±5%
< 0.01 Watt/m2
Precipitation
<± 10%
< 0.25 mm/hr
< ± 5 m or
Mixing Layer Height
± 1%,
whichever is
greater
< 10 m
a Spatial representativeness is addressed in monitor siting as specified in Sections 3.3.1.2 and 3.3.1.3.
b Assessed with twice-annual PT samples and across the entire PAMS season as the upper bound of the mean absolute value of
the percent differences across all single-point QC checks. For functional form of the calculation, see 40 CFR 58 Appendix A
Section 4.1.3, Equations 3, 4 and 5.
11
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
Table 3-1 (continued). Data Quality Indicators and Associated Measurement Quality
Objectives
c Measured as the upper bound of the coefficient of variation (CV) across all single-point QC checks in the PAMS season. For
functional form of the calculation, see 40 CFR 58 Appendix A Section 4.1.2, Equation 2. Acceptance criteria listed here for
criteria pollutants duplicate those in the EPA QA Handbook validation templates. Changes to the QA Handbook requirements
will supersede those criteria listed here.
d Measured as the upper bound of the mean absolute value of the percent differences across all single-point QC checks in the
PAMS season. For functional form of the calculation, see 40 CFR 58 Appendix A Section 4.1.3, Equations 3, 4, and 5.
Acceptance criteria listed here for criteria pollutants duplicate those in the EPA QA Handbook validation templates. Changes to
the QA Handbook requirements will supersede those listed here.
e Carbonyls sampling by TO-11A may be substituted with continuous formaldehyde monitoring and reporting of the hourly
average. MQOs for continuous formaldehyde monitors have not been established at the time this document was written.
f Carbonyls sampling will follow the l-in-3 day sampling schedule as prescribed in Table Bl-2 and the national sampling
calendar available at the following link on AMTIC: https://www3.epa.gov/ttn/amtic/calendar.html
B Assessed with twice-annual PT samples.
h Measured as the coefficient of variation of the RPDs across, as applicable, all (i) duplicate/collocated field-collected cartridges;
(ii) duplicate LCSs; and (iii) replicate laboratory analyses in the entire PAMS season. See Sections 2.5.1 and 2.5.2 of the NATTS
2011-2012 Quality Assurance Annual Report available here:
https://www3.epa.gov/ttnamtil/files/ambient/airtox/NATTS20112012OAARfinal.pdf
1 Refer to 40 CFR Part 50 Appendix U, Section 4
3.3.1 Representativeness
One of the most important DQIs of any ambient air monitoring network is representativeness.
This term refers to the degree to which data accurately represent the frequency distribution of a
specific variable in the population (e.g., concentration of pollutants in air for the spatial scale of
interest). Population uncertainty, i.e., the spatial and temporal components of error, affects
representativeness, as does measurement uncertainty. The latter is controlled through the
selection of appropriate instrumentation and measurement techniques and by specifying
applicable MQOs for important DQIs. Population uncertainty is controlled through the selection
of appropriate boundary conditions such as the monitoring area and the sampling time
period/frequency of sampling. The PAMS Required Sites were selected to capture the
spatiotemporal variability inherent to urban scale measurements in a specific location and, when
taken together over the entire PAMS network, to regional and national scale measurements in the
tropospheric boundary layer over the time period (June through August) during which ozone
concentrations are expected to be greatest in most of the continental United States.
3.3.1.1 Temporal Representativeness
To adequately characterize the diurnal concentrations and weekday/weekend day pattern of
ozone and ozone precursors, the sampling frequency for each of the required parameters is
prescribed as follows to capture the ozone precursor concentrations from June 1 to August 31
(referred to as ozone season). The sampling frequency for ozone, true NO2, NOy, speciated
VOCs, and meteorological parameters is for sampling to occur continuously daily and the
collected data averaged over each hour. Due to the labor-intensive aspects of manual sample
collection onto cartridge media and the need to collect an air volume sufficient to enable
sensitive measurements, carbonyls sampling is not required hourly. Rather, such is required on a
one-in-three day schedule and consists of three sequential 8-hour samples on a given sampling
day. Carbonyls samples are to be collected per the national sampling calendar:
https://www3.epa.gov/ttn/amtic/calendar.html
12
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
Sites may choose to deploy instalments capable of continuous measurements of carbonyls. In
such cases, PAMS Required monitoring agencies are to follow the QA/QC requirements and
guidance developed by EPA {note: such requirements and guidance have not been proposed at
the time of this document's publication). Carbonyls data collected by continuous methods are to
be reported as the hourly average as for the other parameters. The three sequential 8-hour sample
collections every third day provide a sufficient number of data points at a sufficient time
resolution to ensure that the measurements characterize the diurnal concentration pattern over the
course of the PAMS season at a given PAMS Required Site.
3.3.1.2 Spatial Representativeness — Chemical Measurement Probe Siting Criteria
To obtain information on ozone precursors on an urban spatial scale at locations where
significant population exposure to tropospheric ozone may occur, PAMS Required Sites will be
preferentially collocated with NCore stations in CBSAs having populations greater than
1,000,000.
Sampling inlet probes and equipment are to be properly sited to ensure the conditions are
representative of the ambient air in the tropospheric boundary layer of the geographic area
intended to be represented by the site. As such, sites and inlet probes are to comply with the
siting criteria in 40 CFR Part 58 Appendix E. PAMS Required Sites located at NCore sites
typically already meet the siting criteria and therefore are representative of the near-surface
atmospheric conditions.
Some general guidelines for probe and manifold inlet placement are:1
probes should not be placed next to air outlets such as exhaust fan openings or
chimney flues
horizontal probes are to extend beyond building overhangs
probes should not be placed near physical obstructions such as chimneys which can
affect the air flow near the probe
probes need to be accessible for performance evaluation auditors
height of the probe above the ground depends on the pollutant being measured
design of the probe system should be such that both analyzer and gas calibrator
exhaust are vented outside for safety reasons, and operators should periodically check
the outside vent line to ensure it is not clogged or blocked. The outside vent line
should be of minimal length to prevent blocking of the exhaust with debris.
The inlet probe is to be minimally 1 meter vertically or horizontally away from any supporting
structure, wall, parapet, penthouse, etc., and away from dusty or dirty areas. If the inlet probe is
located near the side of a building or a wall, then it should be located on the windward side of the
building or wall relative to the prevailing wind direction during PAMS season.
3.3.1.2.1 Inlet Probe Height
Inlet probes and equipment are to be placed at the following heights:
PAMS VOCs 2 to 15 m above ground level (AGL)
13
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
PAMS carbonyls 2 to 15 m AGL
NOy > 10 m AGL (sites are to be compliant for NCore measurements)
Ozone 2 to 15 m AGL (sites are to be compliant for NCore measurements)
N02 2 to 15 m AGL
3.3.1.2.2 Spacing from Obstructions
Inlet probes are to have unrestricted airflow in a 270-degree arc and the predominant wind
direction is to be included in this arc. As much as possible, inlet probes should not be located on
the side of a building, but if such is the case, must have unrestricted airflow in an arc of at least
180 degrees. This arc must include the predominant wind direction for the PAMS season. As
most of the PAMS Required Sites will be located at existing NCore sites, EPA does not expect
that probes will be mounted on the side of a building.
The inlet probe is to be minimally twice the distance from the potential obstruction as the
potential obstruction extends above the inlet probe. For example, if a wall extends 2 meters
above the inlet probe, the inlet probe is to be 4 meters or more horizontally from the wall.
3.3.1.2.3 Spacing from Trees
Trees can provide surfaces for O3 or NO2 adsorption or reactions and may act as obstructions to
air flow when of a sufficient height and leaf canopy density. To avoid such interferences, inlet
probes are to be minimally 10 meters and optimally > 20 meters from the dripline of the nearest
tree.
3.3.1.2.4 Spacing from Roadways
Mobile sources represent a significant source of ozone precursors; therefore, it is important to
ensure that monitoring sites are sufficiently displaced from roadways since the goal of PAMS
monitoring is to provide urban scale measurements. Minimum separation distances from
roadways assume PAMS Required Sites represent urban scale, and as such are to comply with
Table E-l of 40 CFR Part 58 Appendix E, reproduced below in Table 3-2. Note that these
minimum separation distances are to also be maintained from other motor vehicle traffic areas
such as parking garages and parking lots.
3.3.1.3 Spatial Representativeness — Meteorological Parameters
Siting of meteorology equipment for the required measurements is specific to each instrument
type. General siting criteria for the meteorology instruments follow.
Wind Speed and Wind Direction: The standard height for surface layer wind measurements is
10 meters (m) AGL.2'4'5 The location of the site for the wind measurements should ensure that
the horizontal distance to obstructions (e.g., buildings, trees) is at least 10 times the height of the
obstruction.2'5 An obstruction may be man-made (e.g., a building) or natural (a tree). A wind
instrument should be securely mounted on a mast that will not twist, rotate, or sway. If a wind
instrument must be mounted on the roof of a building, it should be mounted high enough to be
out of the wake of an obstruction. Roof mounting is not a good practice and should only be
resorted to when absolutely necessary. Sensor height and its height above/below any
obstructions, as well as the character of nearby obstructions, should be documented.
14
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
Table 3-2. Minimum Distance for Inlet Probes to Roadways
Uoadwav
average dailv traffic,
vehicles per dav
.Minimum
distance 11
(meters)
.Minimum
distance " h
(meters)
<1,000
10
10
10,000
10
20
15,000
20
30
20,000
30
40
40,000
50
60
70,000
100
100
>110,000
250
250
a Distance from the edge of the nearest traffic lane. The distance for intermediate traffic counts should be interpolated from the
table values based on the actual traffic count.
b Applicable for ozone monitors whose placement has not already been approved as of December 18, 2006.
An open lattice tower is the recommended structure for monitoring of meteorological
measurements at the 10-m level. In the case of wind measurements, certain precautions are
necessary to ensure that the measurements are not significantly affected by turbulence in the
immediate wake of the meteorological tower. To avoid such tower effects, the wind sensor
should be mounted on a mast a distance at least one tower width above the top of the tower, or if
the tower is higher than 10 m, on a boom projecting horizontally from the tower. In the latter
case, the boom should extend a distance at least twice the diameter/diagonal of the tower from
the nearest point on the tower. The boom should project into the direction which provides the
least distortion for the most important wind direction (i.e., into the prevailing wind).
Ambient Temperature and Relative Humidity: The standard height for surface layer ambient
temperature and RH measurements is 2 m AGL 4 If a tower is used, the temperature sensor
should be mounted on a boom which extends at least one tower width/diameter from the tower.
The measurement should be made over a uniform plot of open, level ground at least 9 m in
diameter centered on the sensor. The surface should be covered with non-irrigated or un-watered
short grass or, in areas which lack a vegetation cover, natural earth. Concrete, asphalt, and oil-
soaked surfaces and other similar surfaces should be avoided to the extent possible. The sensor
should be at least 30 m horizontally from any such paved area. If these siting criteria (open
ground and distance from paved surfaces) cannot be achieved, it should be identified in site
characterization documentation. Other areas to avoid include extraneous energy sources
(subway entrances, rooftops, electrical transmission equipment), large industrial heat sources,
roof tops, steep slopes, hollows, high vegetation, swamps, snow drifts, standing water, tunnels,
drainage culverts, and air exhausts. The distance to obstructions for accurate temperature
measurements should be at least four times the obstruction height.
Solar Radiation and Ultraviolet Radiation: Solar radiation and ultraviolet radiation
measurements should be taken in a location with an unrestricted view of the sky in all directions.
15
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
In general, locations should be avoided where there are obstructions that could cast a shadow or
reflect light on the sensor; light colored walls or artificial sources of radiation should also be
avoided. The horizon as viewed from the pyranometer should not exceed 5 degrees. Sensor
height is not critical for pyranometers; consequently, tall platforms or roof tops are typical
locations. Regardless of where the pyranometer is sited, it is important to ensure that the level of
instrument is maintained and that the glass dome is cleaned as necessary. To facilitate leveling,
pyranometers should be equipped with an attached circular spirit level.
Barometric Pressure: Barometric pressure instruments should be located in a ventilated shelter
about 2 m AGL. The height of the station above mean sea level and the height of the pressure
sensor AGL should be documented.
Precipitation: Precipitation gauges should be located on level ground in an open area.
Obstructions should not be closer than two to four times their height from the instrument. The
area around the precipitation gauge should be covered with natural vegetation and should not be
located on a paved or hard surface (e.g., the roof of a monitoring shelter) from which splashing
may occur. The mouth of the gauge should be level and should be as low as possible while still
precluding in-splashing from the ground; also, to avoid becoming snow-covered, 30 centimeters
[cm] AGL is the recommended minimum height). A wind shield/wind screen (such as an Alter-
type wind shield consisting of a ring with approximately 32 free-swinging separate metal leaves)
should be employed to minimize the effects of high wind speeds.
Mixing Layer Height: The ceilometer for determining mixing layer height measurements is
intended for more macro-scale application than are the surface meteorological measurements.
Consequently, the ceilometer need not be located at the PAMS site, but may be placed nearby.
Factors that should be considered in selecting a site for the MLH monitor include whether the
upper-air measurements for the proposed location are likely to provide the necessary data to
characterize the meteorological conditions associated with high ozone concentrations, and the
extent to which data for the proposed location may augment an existing upper-air network. The
ceilometer should be securely installed on a stable level surface such as a concrete pad or
wooden platform suitably located to provide an unobstructed view of the sky. A wide-open
location is recommended where there are no tall trees, overhead lines, or antennas nearby.
Proximity to powerful radars should also be avoided. Any object in the cone projecting upward
created by an angle of 25° from vertical will impede the ability of the ceilometer to properly
measure atmospheric backscatter. Common interfering objects would include powerlines, tree
branches, tower support guidewires, flagpoles, or similar features which may be permanently or
transiently present above the ceilometer.
3.3.2 Completeness
Generation of a dataset sufficient to characterize the diurnal concentration pattern of ozone,
ozone precursors, and meteorological parameters of interest to PAMS requires that a minimum
number of the intended measurements be valid. The MQOs for completeness are specified for
each parameter as detailed in Table 3-1.
For hourly measurements (formaldehyde [when continuous measurement instruments are
deployed and approved for use], ozone, NOy, and true NO2, and for meteorological
16
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
measurements), 45 minutes will be considered a valid hour and 18 hours a valid day (Section
6.4.1 of the Quality Assurance Handbook for Air Pollution Measurement Systems Volume II
[EPA-454/B-17-001 January 2017]). Note that due to limitations with the instrument cycling for
sample collection and measurement, a valid sampling hour for speciated VOCs is 40 minutes of
sampling for the hour, for which 30 minutes of this 40-minute period will occur during the
reported hour (air collection may commence no sooner than 10 minutes before or 30 minutes
past the beginning of the hour). For an 8-hour carbonyls sample to be valid, the sample will be
collected for 8 hours ± 20 minutes, and air collection will commence within 15 minutes of the
scheduled collection start time, adjusted for clock discrepancy. The overall completeness listed
in Table 3-1 will be based on acquiring data for the entire PAMS season determined as the total
valid samples out of the samples possible. For continuous measurements, this will be based on
approximately 2208 hours (24 hours for 92 days). For carbonyls samples, the total possible
sampling days is 30 or 31 days, depending on the sampling calendar for the year. For carbonyls
sample collection, if a sample day is missed (if one or more of the three sequential samples from
the sample day is invalidated), a null code "AF" (scheduled but not collected) will be reported to
AQS for the sample run date; if the sample is invalidated, an appropriate null code will also be
reported to AQS for the sample run date.
A valid sample is one that was collected, analyzed, and reported to AQS without null flags. If a
collected data point is voided or invalidated for any reason (reported with a null flag), this data
point does not count toward completeness. For continuous measurement methods whose
measurements are reported as the hourly average, uncollected or invalidated sample results are
lost, and cannot be made up. For carbonyls sample collection, a make-up sample collection
should be attempted as soon as practical according to the make-up sampling policy below.
3.3.2.1 Make-up Sample Policy - Carbonyls Only
Samples and sample results may be invalidated for a number of reasons. In all cases, the data are
entered in AQS flagged with a null code indicating the data are invalid. In order to increase the
likelihood of attaining the completeness MQO of > 85%, make-up sampling events should be
collected when a carbonyls sample or sample result is invalidated.
A replacement carbonyls sample set (three 8-hour samples) should be collected as close to the
original sampling date as possible, preferably before the next sampling date. Scheduling make-up
sampling in this way helps to minimize potential bias introduced to the PAMS season
concentration average due to differences in ambient diurnal concentration pattern from the
originally scheduled sample date. The diurnal pattern is most strongly impacted by the particular
day of the week (weekday versus weekend day).
In order to be temporally representative of the PAMS season concentration at a given site, the
carbonyls sample dates are to be as evenly distributed as possible to capture concentrations that
fluctuate seasonally or according to weather patterns. It is not acceptable to delay make-up
sampling until the end of the PAMS season, as this may bias the data to be more representative
of the conditions during the month of August than that of the entire PAMS season. Therefore, it
is important to analyze carbonyls samples as soon as possible to determine if make-up samples
are needed (if samples are lost during extraction and analysis).
17
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
To summarize, carbonyls make-up samples should be collected within PAMS season:
1. Before the next scheduled sampling date
2. Within two weeks of the missed collection date, with preference given that the
rescheduled date occurs on a weekday or weekend day to match that of the original
schedule.
3.3.3 Precision
Reproducibility is a key component of ensuring concentration results at one site are comparable
to those at other sites and are comparable over time. The PAMS program only specifies precision
MQOs for the chemical (non-meteorological) parameters; see Table 3-1. Collocated and/or
duplicate measurements for precision assessments are only possible for carbonyls; in order to
evaluate precision for the other methods, replicate analyses of a calibration verification standard,
or one-point QC check, are performed. In addition, for speciated VOC measurements, a
continuing calibration check (CCV) is prescribed to be analyzed twice sequentially on a weekly
basis.
The precision of the entire carbonyls method (collection, extraction, and analysis) is evaluated by
collection of collocated and/or duplicate field samples (samples representing the same air parcel
collected at the same time). The combination of laboratory handling and analytical precision may
be estimated by preparing replicate laboratory control samples (LCS) taken through all
laboratory procedures (extraction and analysis). Laboratory analytical precision alone is assessed
by the replicate analysis of a sample extract. For sites that are not collecting field precision
samples (collocated or duplicate), laboratory precision will be assessed by extraction and
analysis of LCS/LCS duplicate (LCSD) pairs and replicate analysis. Precision measurement
requirements for each PAMS Required Site should be detailed in each monitoring agency's ANP
or PAMS QAPP.
The precision MQOs for the continuous methods (ozone, true NO2, and NOy) are based on an
evaluation of each site's PAMS season precision data and assessed as described in 40 CFRPart
58 Appendix A Sections 2.3.1, 3.1.1, 4(b), and 4.1.2 and in the validation tables in Appendix D
of Quality Assurance Handbook for Air Pollution Measurement Systems Volume II (EPA-454/B-
17-001 January 2017). Additional guidance on precision for NOy measurements is provided in
Section 4.3.1.1 of the Technical Assistance Document (TAD) for Precursor Gas Measurements
in the NCore Multi-pollutant Monitoring Network (EPA-454/R-05-003, Version 4, September
200J).
Speciated VOCs precision is assessed for each individual chemical parameter by determining the
absolute relative percent difference (RPD) of the pairwise (N = 2) replicate back-to-back
precision checks (CCV analyses), which should not exceed 25% RPD. These individual
precision pairs are evaluated as QC checks to ensure ongoing instrument performance in the
same way the ongoing CCV analyses are evaluated to demonstrate instrument performance
which may result in qualification of individual sampling hours until the instrument performance
returns to conformance. The absolute RPD is calculated as the difference between the paired
measurements divided by the mean of the pair, expressed as a percentage:
18
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
IS — S I
%Absolute RPD = — ¦ 100%
"^Sci'Sb
where: Sa = first measurement in pair
Sb = second measurement in pair
Xsa,sb= population mean of measurements Sa and Sb
Precision for the PAMS season (or measurements for the calendar year) is estimated and
assessed by calculating the upper bound of the coefficient of variation (CV) across all single-
point CCV (daily and weekend precision CCV analyses) analyses in the PAMS season. The
individual percent difference (di) for each individual check is calculated for each target
compound as follows:
d = Cm—£n 10Q
Cn
where: Cm = measured concentration
Cn = nominal concentration
For the functional form of the calculation, refer to 40 CFR Part 58 Appendix A Section 4.1.2,
Equation 2, reproduced here:
where n is the number of single point checks for aggregation and Xo.i,n-i is the 10th percentile of
a chi-squared distribution with n-1 degrees of freedom.
For carbonyls, the precision of the entirety of the method is evaluated by calculation of the %CV
using the duplicate and/or collocated sample pairs aggregated across the PAMS season, as
follows:
%CV = 100-
(Pi - r,)
,05-(A +rt)_
2 n
where:
pt = the result of the analysis performed on the primary sample within the z'th pair,
rt = the result of the analysis performed on either the collocated or duplicate sample
within the z'th pair, and
n = the number of primary-collocated and primary-duplicate sample pairs.
For evaluation of laboratory extraction and analysis precision only, /;, and n are the LCS/LCSD
pairs. Precision of the analytical method is evaluated by calculating the %CV of replicate
measurement pairs.
19
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
Attainment of the precision MQOs, as calculated using the aggregated single-point QC checks
(continuous gas analyzers), ongoing CCVs (speciated VOCs), and various
collocated/duplicate/replicate pairs (carbonyls) should be evaluated on a monthly basis so that
preventative or corrective action(s) may be taken to avoid or recover from nonconformances,
respectively. One method for accomplishing such for the continuous gas analyses is by
generating the AMP600 report within AQS.
3.3.4 Bias
Bias is the difference of a measurement from a true or accepted value and can be negative or
positive. As much as possible, bias should be minimized as biased data input to trends analysis or
to models may result in incorrect trends evaluation or inaccurate model outcomes. Bias may
originate in several places within the measurement system such as sample introduction and
collection, instrument calibration, and analysis steps. Sources of sample introduction and
collection bias include, but are not limited to, poorly maintained (dirty) sampling inlets and flow
paths, flow paths with incompatible materials, leaks within the flow path, incorrectly calibrated
flows or out-of-calibration sampling instruments, elevated and unaccounted-for background
within the instrument or on collection media, and poor sample handling techniques resulting in
contamination or loss of analyte. Sources of instrument calibration bias include, but are not
limited to, poor hygiene or technique in standards preparation, incorrectly calibrated or out of
tolerance equipment used for standards preparation and use of contaminated or incompatible
materials in standards preparation activities. Incorrect input of standards materials theoretical
(true or certified) values when establishing calibration can likewise impart a bias to
measurements. Analysis bias may result from instrument response drift or infrequent or
inappropriate instrument maintenance leading to enhanced or degraded analyte responses.
Attainment of acceptably low bias is verified by performing periodic calibration checks which
may also include analysis of a second source standard in the case of speciated VOCs analysis and
carbonyls analysis. Instruments that demonstrate bias not meeting the specified acceptance
criteria are to be re-calibrated following maintenance or corrective action, as needed. Bias
MQOs are detailed in Table 3-1.
Independent assessment of the bias for speciated VOCs and carbonyls will be performed with
proficiency test (PT) samples. EPA (or a support contractor) will provide the PTs spiked with
target analytes at concentrations unknown to the site or laboratory. More information on these
PTs is given in the following subsections. Independent assessment of the bias for true NO2,
ozone, and NO will be conducted as part of the National Performance Audit Program (NPAP).
Bias for speciated VOCs, true NO2, ozone, and NO will also be evaluated across the entire
PAMS season as the upper bound of the mean absolute value of the percent differences across all
CCVs and/or single-point QC checks. For functional form of the calculation, see 40 CFR 58
Appendix A Section 4.1.3, Equations 3, 4 and 5. Furthermore, bias for carbonyls is assessed by
ongoing flow rate checks.
Laboratories not meeting bias MQOs will take corrective action, as appropriate. Corrective
action will depend on which analytes did not meet criteria, the number of analytes not meeting
criteria, the magnitude of the bias, and may involve qualification or invalidation of reported
20
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
ambient results depending on the severity and scope of bias nonconformance(s). Attainment of
the bias MQOs, as calculated using the aggregated single-point QC checks (continuous gas
analyzers) and ongoing CCVs (speciated VOCs), should be evaluated on a monthly basis so that
preventative or corrective action(s) may be taken to minimize the likelihood that data may
require invalidation as a result of end-of-season data review.
3.3.4.1 Assessing Laboratory Bias
Each laboratory performing analysis of carbonyls samples collected at PAMS Required Sites is
to participate in the PT program for carbonyls. Minimally annually, EPA (or its support
contractor) will prepare one or more 2,4-dinitrophenylhydrazine (DNPH) cartridges spiked with
known amounts of target carbonyl compounds. Cartridges will be dispatched to each support
laboratory to be extracted and analyzed per the laboratory's standard procedures. The
concentrations of the target compounds are blind to the laboratory. The monitoring agency then
reports the measured results to the PT provider who will compile the reported concentrations for
evaluation against the nominal spiked value and against the overall PAMS Required Site
Network average (with outliers removed). Support laboratories that do not meet the bias criterion
of ± 25% of the target value (likely the PAMS Required Site network average measured
concentration) are required to take corrective action.
The PT program for carbonyls is well-established for many of the laboratories that analyze
carbonyls for air toxics monitoring programs. Successful participation in air toxics carbonyls
PTs may satisfy the requirement for participating in the PAMS carbonyls PT at EPA's discretion.
3.3.4.2 Assessing Field Measurement Bias
3.3.4.2.1 Field Site Proficiency Testing for Speciated VOCs
Each PAMS Required Site is to participate in the PAMS PT program for speciated VOCs. EPA
or and EPA support contractor (PT provider) will prepare a sample mixture with a known
concentration of speciated VOCs in a stainless steel canister or aluminum cylinder and dispatch
it to each PAMS Required Site to be analyzed with the site's auto-GC. The PT will be conducted
minimally annually, and likely biannually, once prior to the beginning of PAMS season and just
before the end of PAMS season. The concentrations of the target compounds will be blind to the
monitoring site. The PT will focus on the priority compounds and may also contain a suite of
optional compounds of interest. The monitoring agency will report the measured results to the
PT provider who will compile the reported concentrations for evaluation against the nominal
spiked value and against the overall PAMS Required Site Network average (with outliers
removed). PAMS Required Sites which do not meet the bias criterion of ± 25% of the target
value (likely the PAMS Required Site network average measured concentration) for the priority
compounds will take corrective action. Further information on the PAMS PT program will be
available at the following link on EPA Ambient Monitoring Technology Information Center
(AMTIC):
https://www3.epa.gov/ttnamtil/pamseuidance.html
For speciated VOCs, there are a number of variables that may impact the overall network
average results; therefore, at EPA's discretion, PT results may be further broken down by
21
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
instrument manufacturer to investigate bias or variability associated with the different instrument
models.
3.3.4.2.2 Assessing Field Bias for Carbonyls
The direction of the flow rate bias in carbonyls samplers is opposite that of the bias introduced in
the reported concentrations. That is, flow rates that are biased low result in overestimation of in-
air concentrations whereas flow rates that are biased high result in underestimation of in-air
concentrations.
Indicated flow rates for carbonyls are to be within ± 10% of the certified flow transfer standard.
Corrective action should be taken when this criterion is not met including, but not limited to,
recalibration of the sampling unit flow control device which may involve adjusting the flow
linear regression response (slope and intercept). Sampling units that cannot meet these flow
accuracy specifications are not to be utilized for sample collection. Additionally, following a
failing calibration or calibration check, monitoring agencies are to evaluate sample data collected
since the last acceptable calibration or calibration check, and such data may be subject to
qualification or invalidation. Corrective action is recommended for flow calibration checks that
indicate flows approaching, but not exceeding the appropriate flow acceptance criterion. EPA
recommends monitoring agencies perform multipoint calibration of the flow control device(s)
prior to the commencement of PAMS season, check monthly during PAMS season, and perform
a final check at the conclusion of PAMS season. Since PAMS Required Sites will preferentially
be collocated atNCore sites where flow rate checks of PM2.5 monitors are performed monthly,
monitoring agencies can schedule carbonyls flow checks to coincide with those for PM2.5.
Sampling bias for carbonyls is also characterized by challenging field collection instruments with
analyte-free humidified zero air or nitrogen (zero checking) as discussed further in Section
5.7.1.1.
3.3.4.2.3 Ongoing Bias Assessment for Speciated VOCs and Continuous Gas Monitors
Bias for speciated VOCs, true NO2, ozone, and NO is evaluated across the entire PAMS season
as the upper bound of the mean absolute value of the percent differences across all single-point
QC checks. For functional form of the calculation, see 40 CFR 58 Appendix A Section 4.1.3,
Equations 3, 4 and 5, reproduced below. Acceptance criteria are given in Table 3-1.
The bias estimator is an upper bound on the mean absolute value of the percent differences (di -
as detailed above in Section 3.3.3):
\bias\ = AB + t0.95,n-i- —p
Vn
where n is the number of single-point QC checks being aggregated, /o.95.0-1 is the 95th quantile of
a t-distribution with n-1 degrees of freedom; the quantity AB is the mean of the absolute values
of the percent differences and is calculated per the following equation:
22
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
n
= --Yldil
n Z—i
AB
n
i=1
and where the quantity AS is the standard deviation of the absolute value of the percent
differences and is calculated using the following equation:
AS = jn ' Yi=i\di\2 — (Ei=i\di\)2
n(n — 1)
3.3.4.2.4 Through-the-Probe Auditing
Each PAMS Required Site will be audited for ozone and true NO2, and for NO/NOy (as
practical), according to the frequency and procedures defined by the EPA NPAP by delivery of
challenge gases provided through-the-probe (TTP). Independent NPAP auditors provide
challenge gases of known concentration, blind to the site operator, to the monitoring station
inlets. Site operators report the measured concentration of each audit concentration level and the
site is evaluated on the bias of the reported measurement compared to the known concentration.
Acceptance criteria are described within a February 2011 EPA memorandum:
https://www.epa.gov/sites/prodiiction/files/2i documents/201102171owlevelstatmemo.pdf
As of the publication of this TAD, the NPAP program prescribes that each site within a primary
quality assurance organization (PQAO) is to be audited every five years, however, this frequency
may be subject to change for PAMS Required Sites per EPA directive. The TTP audit is required
to be performed as described in 40 CFR Part 58 Appendix A and further details and information
can be found in the NPAP program documents on the EPA AMTIC site:
https://www3.epa.eov/ttnamtil/npaplist.html
In addition to independent audits conducted as part of NPAP, monitoring agencies are required
to conduct TTP audits of the criteria gas analyzers annually as prescribed in 40 CFR Part 58
Appendix A Section 3.1.3.
Monitoring agencies should evaluate the conversion efficiency of the NOy channel of the NOy
instrument to ensure the conversion of n-propyl nitrate (NPN) to NO exceeds 95%.5
3.3.5 Sensitivity
EPA defines sensitivity as "the capability of a method or instrument to discriminate between
measurement responses representing different levels of the variable of interest."6 An important
aspect of sensitivity is the ability of the method to differentiate a signal arising from the presence
of the target analyte from any combination of signal and noise that is measured in the absence of
the target analyte. This lower limit (at which a measured signal is sufficiently large so as to be
due to the presence of the target analyte) is established experimentally for O3, true NO2, NOy,
carbonyls, and VOCs by conducting an MDL study. In order to ensure that the chemical
23
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
measurement methods are sufficiently sensitive for reported concentration values to be useful in
trends evaluations and for model input, MDLs should not exceed those detailed in Table 3-1.
The MDL and sample quantitation limit ([SQL], defined as 3.18 times the MDL concentration)
provide information on the concentration at which both positive identification and accurate
quantification are expected, respectively. That the SQL is greater than the MDL reflects the fact
that as concentrations of target analytes are measured closer to the MDL or below the MDL, the
resulting measurements become less accurate (decrease in precision and increase in bias). While
all measured concentrations (even those less than the MDL) of positively identified target
analytes are to be reported to AQS, the accuracy associated with each reported concentration is
related to the corresponding MDL and SQL.
The SQL is equivalent to ten-fold the standard deviation of seven MDL measurements, which
was defined in draft EPA guidance in 19948 as the minimum level (ML). The factor of 3.18 was
derived by dividing 10 standard deviations by 3.14 (the one-sided 99th percentile Student's t
value for seven replicates). The MDL process in 40 CFR Part 136 Appendix B is protective
against reporting false positives such that 99% of the measurements made at the determined
MDL value are positively detected (determined to be different from the detector's response in the
absence of the analyte) but does not attempt to characterize precision or address accuracy at the
determined MDL concentration. Method accuracy and precision are expected to be attained at
concentrations at and above the SQL (ML).
Meteorology instruments should meet the resolution specifications listed in Table 3-1. The
resolution specifications provided by the manufacturer will indicate that instruments are suitable
for PAMS Required Site meteorological use. Sensitivity for meteorology instrument
measurements is fundamentally different than for chemical measurement instruments for which
the lowest concentration differentiable from background is useful. For ambient meteorology
measurements, resolution, or the ability to differentiate between two similar measurements, is of
interest, since the conditions to be quantified are not challenging to detect in the same way that
low concentrations of chemical species are. For example, the ability to discern between
temperatures of 24.2°C and 24.6°C is important; however, it is not important to be able to
measure the lowest temperature possible since such is not a concern for ambient monitoring.
3.3.5.1 Method Detection Limits
The MDL as prescribed in 40 CFR Part 136 Appendix B was initially developed and applied to
wastewater analyses.9 Since then, this procedure has been applied to a variety of other matrices
and analysis methods to approximate the lowest concentration (or amount) of analyte that can be
reported with 99% confidence that the actual concentration (or amount) is greater than zero. As
can be seen in Figure 3-1, the Gaussian curve represents analysis of contamination-free method
(matrix) blanks and the distribution of their concentration values around zero. The small area of
the blank values to the right of the MDL value (indicated by the vertical dashed line) represent
the 1% of values that would be considered false positives (an analyte reported to be present when
it in fact absent).
24
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
C
o
•4—1
o
0)
¦*—>
(D
"O
SX f
O
o
c
0
ZJ
cr
Q>
v_
LL
1 % chance of
false positive
Not drawn to scale
0 MDL
Concentration
Figure 3-1. Graphical Representation of the MDL and Relationship to a Series of Blank
Measurements in the Absence of Background Contamination
(Credit: Reference 10 as adapted from Reference 11)
In practical terms, this MDL procedure provides a conservative detectability estimate and aims to
ensure that there is a 1% false positive rate (incorrectly reporting the presence of an analyte when
it is in fact absent) at the determined MDL concentration. In many cases, the analyte will be
qualitatively identified (positively detected) at concentrations below the MDL with a signal
distinguishable from instrumental noise. That is to say, the MDL procedure is not protective of
false negatives, which is incorrectly concluding that the analyte is absent when it is in fact
present; in fact, 50% of the time the analyte present at the MDL concentration will be measured
at less than the MDL (the compound will not be reported as 'detected' when it is in fact found to
be present).12 This can be seen in Figure 3-2; the solid Gaussian curve represents a series of
measurements at the MDL concentration. The measurements in the shaded portion of the curve
to the left of the MDL value are false negatives or values measured at less than the MDL despite
the fact that such measurements meet qualitative identification criteria. Therefore, if an analyte is
measured at the MDL concentration, the analyte is present 99% of the time; however, for
analytes measured at or less than the MDL concentration, 50% of the time the analyte may also
be present.
25
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
50 % chance of
false negative -
sx t
0 MDL
Concentration
Figure 3-2. Graphical Representation of the MDL and Relationship to a Series of
Measurements at the MDL Value
(Credit: Reference 10 as adapted from Reference 11)
In summary:
• 99% of the results measured > MDL are in fact greater than zero (there is a 1% false
positive rate, or chance that such measurements are not actually greater than zero)
• 50% of actual concentrations at the MDL will be reported as > MDL
• 50% of actual concentrations at the MDL will be reported as < MDL (they will be
false negatives) even though they may still be qualitatively identified and may still in
fact be valid identifications
The MDL as described in 40 CFR Part 136 Appendix B and in Reference 9 is a statistical
estimate of the lowest concentration at which there is a 99% chance that the concentration is
greater than zero. The MDL procedure is not simply a characterization of the noise of the
instrument nor is it meant to estimate measurement accuracy at the MDL concentration. The
MDL is also not an estimate of the precision or variability of the method, although method
precision is critical to a method's sensitivity. Note that such is in accord with the definition in
EPA QA/G-5 of sensitivity as the capability of a method to differentiate varying levels of
analyte; higher precision methods will have greater sensitivity than lower precision methods.
Emphasis is given to the overall MDL, rather than the instrument detection limit (IDL), in that
the former is a characterization of the sensitivity of the entire measurement method, inclusive of
the potential effects of the sampling pathway, the sample extraction and preparation process (for
carbonyls), and of the sample matrix. Characterization of the analysis instrument sensitivity
26
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
exclusive of other portions of the method (e.g., extraction process, sampling matrix), as
determined by the IDL, should not be substituted for the MDL. The IDL establishes the lowest
concentration that may be differentiated from signal background and noise at a prescribed level
of statistical confidence; knowledge of the IDL is particularly helpful when attempting to
determine, for example, why MDLs may be elevated.
There are known limitations to the 40 CFR Part 136 Appendix B MDL procedure, not the least
of which is that it is a "compromise between statistical respectability and requirements of cost
and time."10'11 More specifically for the PAMS Required Site program, the MDL procedure
prescribed in this TAD does not explicitly take into account the impact of all portions of the
method from collection through analysis, and excludes the effects that the sampling pathway
may impart to the MDL. To capture such effects, the MDL study would be conducted through
the sampling probe and such is impractical for gases, particularly because the spiked samples are
to be analyzed on three separate dates. For carbonyls and VOCs, sourcing gaseous standards is
difficult and delivering low concentration gases suitable for determining MDLs requires
expensive equipment. The impact of the sampling process on detectability is minimized by
requiring that ozone and true NO2 be challenged TTP, by strongly recommending that continuing
calibration verifications for VOCs are introduced as close to the sample inlet probe as feasible,
and by strongly recommending that bias checks are performed for carbonyls field samplers.
The MDL concentration, as defined in 40 CFR Part 136 Appendix B, is determined statistically
by preparing and analyzing minimally seven separate aliquots of a standard spike prepared in the
method matrix. For continuous gaseous analyzers, this would involve measuring standard
aliquots over the course of seven separate measurement periods (each approximately 20 minutes,
or a sufficient period of stable measurements). All portions of the method and matrix are to be
included in the preparation and analysis, as applicable and feasible, such that as much of the
variability of the method and as many of the possible matrix effects are taken into account. The
MDL procedure is an iterative process and, to be meaningful, the MDL procedure is to be
performed as prescribed herein.
The MDL procedure adopted for the PAMS Required Site program builds upon the 40 CFR Part
136 Appendix B by adding some aspects of the promulgated MUR.13 The MUR recognizes that
the CFR procedure assumes that blank values are centered around a concentration of zero and
does not take into account the potential for background contamination to be present in the sample
collection process or on media and does not make allowances for instrument zero drift. If there is
a consistent background level of contamination on the sample collection media (as is typical for
carbonyls on DNPH cartridge media) or a consistent signal in the absence of the target analyte,
measured matrix blank values will not be centered around zero; rather, they will be centered on
the mean blank value. In such cases the MDL is to be defined as the value that is statistically
significantly greater than the blank value and the 40 CFR Part 136 Appendix B procedure will
underestimate the MDL. This occurs since the resulting standard deviation of the MDL replicates
(and thus the calculated MDL concentration) prepared in the presence of background
contamination will not be different than if there were no discernable background (standard
deviation simply evaluates the difference in the spread of the values, not the magnitude of the
individual values). The MUR considers the process background and adjusts for blank levels that
are not centered around zero.
27
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
Specific guidance for determining MDLs for each of the chemical analysis methods is described
in each respective section of this document (Sections 4.3, 5.6, and 6.2.9, for speciated VOCs,
carbonyls, and continuous gaseous monitoring for true NO2 and other criteria pollutants,
respectively). The MDL MUR procedure adds few additional steps than those required in the 40
CFR Part 136 Appendix B procedure. The net effect is that if there is little or no contribution of
background contamination on the sampling media or negligible zero drift, the MDL will be
similar to that determined by 40 CFR Part 136 Appendix B. If the sampling media or other
aspects of the preparation procedures or matrix contributes blank contamination or additional
signal to the blank, the determined MDL will incorporate this average background concentration.
In such cases, the MDL as determined by the modified MUR procedure will be the concentration
at which there is a 99% chance that the concentrations reported at this level are in fact greater
than the mean blank level.
The MUR-modified 40 CFR Part 136 Appendix B protocol maintains the 50% false negative rate
of the original procedure, which is generally recognized as unacceptable for the purposes of
environmental monitoring.10'11 As a result, concentrations measured at less than the MDL, as
long as the qualitative identification criteria have been met (analyte is positively detected), are
valid and necessary for trends analysis and substituting or censoring concentrations measured at
less than the MDL is not permitted. EPA recognizes that many monitoring organizations are not
comfortable reporting concentrations measured less than the MDL as these concentrations are
outside of the calibrated range of the instrument and are associated with an unknown and
potentially large uncertainty. However, actual values reported, even when less than the MDL, are
more valuable from a data analyst's standpoint and far superior than censored or substituted
values. Addition of qualifiers as prescribed in Section 11.5.1 in Table 11-3 indicates when values
are near, at, and below detection limits; these qualifiers indicate when larger uncertainty should
be expected with such concentrations.
3.3.5.1.1 Frequency of Method Detection Limit Determination
MDLs are to be determined minimally annually prior to PAMS season or when changes to the
instrument or preparation procedure result in significant changes to the sensitivity of the
instrument and/or procedure. Examples of situations where MDLs should be re-determined
include, but are not limited to:
• Detector or lamp replacement
• Replacement of the plumbing in the auto-GC leading to changes in background levels
of speciated VOCs, such as changing the moisture management system component
for auto-GCs (i.e., the Nafion™ dryer or cryogenic water trap)
• Replacement of the auto-GC preconcentrator trap(s) which directly impacts the
instrument sensitivity
• Changing the cleaning procedure for sample collection media or labware which
results in a change in background contamination levels
It is recommended that MDLs be determined following annual maintenance of the instruments,
as annual maintenance will likely include some of the items listed above.
28
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
3.3.5.1.2 MDL Measurement Quality Objectives
To ensure that measurements of ozone precursors in ambient air are sufficiently sensitive to
assess trends in concentrations, a minimum required method sensitivity, or MDL MQO, has been
established for each of the chemical parameters. Monitoring agencies and supporting analytical
support laboratories (ASLs) should meet (have MDLs equal to or less than) the applicable MDL
MQOs listed in Table 3-1.
The MDL MQOs are based on concentrations deemed to be reasonably achievable by the
associated methods and instruments while meeting the needs of modelers for model evaluation.
While analytical methods prescribed in this TAD are capable of meeting the MDL MQOs,
MDLs may be elevated above the MDL MQOs due to background contamination for certain
specific analytes. The convention listed in 40 CFR Part 136 Appendix B accounted for
instrumental limitations during the determination of MDLs but did not consider background or
interferences, which, in certain instances, may be several-fold higher than the MDL MQO. Note
that this is typically an issue related to analysis of carbonyls by Method TO-11 A, yet typical
measured concentrations of carbonyls such as formaldehyde in ambient air (~1 ug/m3) exceed the
MDL MQO (-0.25 ug/m3) where the latter was established by accounting for typical media
background levels of carbonyls such as formaldehyde.
3.3.5.1.3 Determining MDLs via 40 CFR Part 136 Appendix B - Method Update Rule
MDLs are determined according to the updated MDL procedure described in 40 CFR Part 136
Appendix B, the MUR.13 MDLs should be determined for each instrument employed to measure
PAMS Required Site parameters. Monitoring agencies are to report the determined MDL for
each parameter to AQS for each sample result reported (as part of the AQS reporting string).
Specific to ASLs, those utilizing multiple instruments for carbonyls analysis should perform
MDL studies for each instrument (the same samples or extracts may be used for all analysis
instruments) from which PAMS carbonyls data are reported. In instances where multiple
instruments are employed for reporting carbonyls for PAMS Required Sites (e.g., two or more
high performance liquid chromatograph [HPLC]-ultraviolet [UV] instruments), there are two
conventions for how to report the MDLs. The preferred convention is to maintain an MDL for
each instrument and report the respective MDL from the instrument on which a given sample
was analyzed. Alternatively, the most conservative (highest) MDL from the multiple instruments
can be reported - though this may not reflect the MDL associated with the sample analysis. It is
not appropriate to average the MDL values for reporting.
The MDL procedure described in this section is adopted from the procedure as given in 40 CFR
Part 136 Appendix B with several changes, based on those proposed in the CFR on February 19,
2015, to explicitly include in the MDL the background (i.e., contamination) contribution of the
sample collection media or instrument background and to incorporate temporal variability in
laboratory preparation (where applicable) and instrument performance.13
For each chemical analysis method (true NO2, ozone, NOy [as possible], speciated VOCs, and
carbonyls), the MDL by MUR involves measuring a minimum of seven "spikes," prepared at a
specifically chosen low concentration, and a minimum of seven blanks. The spikes and blanks
29
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
are prepared and/or measured over the course of three different, preferably non-consecutive,
calendar days to incorporate temporal variability in instrument performance. Calculations,
acceptance criteria, and reporting to AQS are discussed within the individual chemical
measurement methods MDL sections.
After all spikes and blanks are analyzed, two MDL values are calculated: one MDL for the
spiked samples according to the convention in 40 CFR Part 136 Appendix B (MDLsp) and one
MDL for the method blanks which includes the media and/or matrix background contribution
(MDLb).
The first step is to select a spiking level for preparing the MDL spiked samples. If too low of a
spiking level is chosen, the analyte may not be reliably detected by the instrument. If too high of
a spiking level is chosen, the variability of the method near the actual limits of detection may not
be properly characterized. An appropriate spiking level may be selected by considering the
following (in order of importance):
1. The concentration at which the instrument signal-to-noise (S:N) ratio is three- to five-
fold for the analyte.
2. The concentration at which qualitative identification criteria for the analyte are lost
(note that this will be approximately the concentration determined from the MDL
process absent of blank contamination).
3. Analysis of a suite of blank samples - calculate the standard deviation of the
measured concentration and multiply by 3.
4. Previously acceptable MDL studies and related experience.
Note that the MDL spiking level should not be within the calibration curve; rather, the MDL
spiking level should be less than the lowest non-zero calibration standard to best approximate the
instrument response at the MDL. Concentrations within the calibration curve are required to
meet method precision and bias acceptance criteria and are of a high enough concentration that
qualitative identification is certain.
The second step is to prepare the seven or more separate spiked samples (at the level determined
in the first step) and seven or more method (matrix) blank samples. In order to best mimic
procedures conducted in the field, each spike and blank sample should be, to the extent feasible,
subjected to the same procedures performed to process field samples, and include all portions of
the sample matrix and steps in preparation for analysis. Method matrix blanks and spiked
samples should be prepared and measured over the course of three different preparation batches
preferably on non-consecutive days.
An efficient method to determine the MDL following this convention is to prepare and analyze
an MDL sample periodically over the course of several weeks or months. In this scenario, one
MDL sample (or up to three) would be prepared and measured (every week, for example) and
after seven or more data points have been collected for the MDL samples and for the associated
method matrix blanks (which are analyzed routinely as ongoing QC), the MDL could be
calculated. This would alleviate the need to dedicate a significant contiguous block of time to
preparing and analyzing MDL samples and method blanks.
30
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
The following should be taken into consideration during preparation of the MDL samples for
carbonyls:
1. Spiked samples are prepared in matrix (DNPH cartridge).
2. Blanks or blank media which do not meet cleanliness criteria for a given analyte
should trigger root cause analysis to determine the source of the contamination and
should not be used to determine the method blank portion of the MDL.
The third step is to analyze the samples against a valid calibration curve. QC criteria for the
analysis should be met (blanks or zeroes, continuing calibration or span/precision checks,
secondary source QC standards, LCS, calibration checks, etc.). The samples should be analyzed
over the course of minimally three different dates. Note that it is expected and acceptable that
one or more of the applicable qualitative identification criteria (e.g., signal-to-noise ratio or
qualifier ion abundance where applicable - note that retention times for chromatographic method
should remain within typical acceptable limits) will not be met for MDL spikes, but they may
still be included in the MDLsp calculation.
1. Perform all MDL calculations in the final units applicable to the method (e.g., part
per billion carbon [ppbC], |ig/m3).
2. Calculate the MDL of the spiked samples, MDLsp:
a. Following acquisition of the concentration data for each of the seven or more
spiked samples, calculate the standard deviation of the calculated concentrations
for the spiked samples (%>). Include all replicates unless a technically justified
reason can be cited (faulty injection, power glitch, etc.), or if a result can be
statistically excluded as an outlier.
b. Calculate the MDL for the spiked samples (MDLsp) by multiplying .vsp by the one-
sided 99th percentile Student's t-statistic corresponding to the number of spikes
analyzed according to Table 3-3. Other values of t for additional samples (n > 34)
may be found in standard statistical tables.
MDLsp = .VsP ¦ t
c. Compare the resulting calculated MDLsp value to the nominal spiked amount. The
nominal spiked level will be greater than MDLsp and less than 10-fold MDLsp,
otherwise the determination of MDLsp should be repeated with an adjusted
spiking concentration. For MDLsp values greater than the nominal spike level, the
MDL spiking level should be adjusted higher by approximately two- or three-fold.
For nominal spike levels that are greater than 10-fold the MDLsp, the MDL
spiking level should be adjusted lower by approximately two- or three-fold.
3. Calculate the MDL of the method matrix blanks, MDLb:
a. If none of the method matrix blanks provide a numerical result for the analyte, the
MDLb does not apply. A numerical result includes both positive and negative
values for positively identified analytes. Non-numeric values such as "ND" (non-
detect) would result when the analyte is not positively identified. Only method
31
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
matrix blanks that meet the specified qualitative criteria for identification (S:N,
etc.) are to be given a numerical result.
Table 3-3. One-sided 99th Percentile Student's t Values
Nil in her of Ml)l
Siimplcs (id
Doiiivos ol'
l-rmlom
V (II- 1 )
Siiuk'iil's I
Value
7
6
3.143
8
7
2.998
9
8
2.896
10
9
2.821
11
10
2.764
12
11
2.718
13
12
2.681
14
13
2.650
15
14
2.624
16
15
2.602
17
16
2.583
18
17
2.567
19
18
2.552
20
19
2.539
21
20
2.528
22
21
2.518
23
22
2.508
24
23
2.500
25
24
2.485
26
25
2.479
27
26
2.473
28
27
2.467
29
28
2.462
30
29
2.457
31
30
2.453
32
31
2.449
33
32
2.445
34
33
2.441
b. If the method matrix blank pool includes a combination of non-numeric (ND) and
numeric values for the target analyte, set the MDLb to equal the highest of the
method blank results. If more than 100 method matrix blank results are available
for the analyte, set the MDLb to the level that is no less than the 99th percentile of
the method matrix blanks. In other words, for n method blanks where n> 100,
rank order the concentrations. The value of the 99th percentile concentration
(n 0.99) is the MDLb. For example, to determine MDLb from a set of 129 method
blanks where the highest ranked method blank concentrations are 1.10, 1.15, 1.62,
1.63, and 2.16, the 99th percentile concentration is the 128th value (129 0.99 =
127.7, which rounds to 128), or 1.63. Alternatively, spreadsheet programs can be
employed to interpolate the MDLb more precisely.
c. If all concentration values for the method matrix blank pool are numeric values
(negative, zero, or positive), calculate the MDLb as follows:
32
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
i. Calculate the average concentration of the method blanks (xb). If Xb < 0,
let Xb = 0.
ii. Calculate the standard deviation of the method blank concentrations, 5b.
iii. Multiply Sb by the one-sided 99th percentile Student's t-statistic
corresponding to the number of blanks analyzed according to Table 3.3.
Other values of t for additional samples (n > 34) may be found in standard
statistical tables.
iv. Calculate MDLb as the sum of Xb and the product of sb and the associated
student's T value:
MDLb = Xb+ 5b-t
4. Compare MDLsp and MDLb. The higher of the two values is reported as the MDL for
the given target analyte.
5. If the MDL is determined as the MDLsp, the determined MDL should be verified by:
a. Preparing one or more spiked samples at one- to five-fold the determined MDL
and analyzing the sample per the method to ensure the determined MDL is
reasonable. Recall that at the MDLsp concentration there is a 50% chance that the
analyte will not be detected; however, the analyte should be detected at two- to
five-fold the determined MDL.
b. Comparing the measured values to reasonable bias acceptance criteria for the
measured concentration of the MDL verification samples. For example, an MDL
verification that recovers 2% of the nominal amount is not realistic, nor is one that
recovers 300%. Appropriate potential acceptance limits are to double the
acceptance window prescribed by the method for the given analyte. For example,
TO-11A normally permits formaldehyde LCS recoveries to be 80 to 120% (±
20%) error), therefore the MDL verification acceptance limits would be
established at 60 to 140%> recovery. Note that agencies may develop alternate
acceptance criteria through control charts or other similar tools. For methods with
a significant background or matrix contamination, blank subtraction may be
necessary to evaluate the recovery of the MDL verification sample (note this is
unlikely if the MDLb is not higher than the MDLsp).
c. Examining the MDL procedure for reasonableness if the verification sample is
outside of the laboratory-defined acceptance criteria. Such an examination might
include investigating the signal-to-noise ratio of the analyte response in the spiked
samples, comparing the MDL to existing instrument detection limits (if known),
and relying on analyst experience and expertise to evaluate the MDL procedure
and select a different spiking level. The MDL study should then be repeated with
a different spiking level.
33
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
3.4 Quality Assurance Project Plan
The monitoring agency quality system is the framework that ensures that defensible data of
appropriate quality (those that meet the network MQOs for the various DQIs) are generated and
reported to EPA so that the program-specific DQOs are attained. The QAPP is the roadmap for
the design of each agency's quality system (QS) for the specific monitoring program. It describes
the framework of the resources, responsible individuals, and actions to be taken to attain the
PAMS Required Site quality requirements.
Given the importance of the QAPP, each monitoring organization operating a PAMS Required
Site and/or ASL performing analysis of PAMS samples is to have an up-to-date and fully
approved QAPP. All major stakeholders involved in the monitoring organization's and/or
laboratory's PAMS Required Site Program work should provide input to and review the QAPP
to ensure that aspects of the QAPP for which the stakeholders are responsible are accurately and
adequately described. The QAPP should be minimally be approved and signed by the monitoring
organization's PAMS Program Manager (however named) and by the cognizant EPA Regional
office staff person (or EPA Regional office delegate as defined in the grant language) in the EPA
Region in which the monitoring site and/or laboratory exists. The original approved QAPP will
then be kept on file with the monitoring agency. Additional approvers would include a
monitoring agency QA representative and other appropriate managers, as applicable.
The PAMS program QAPP is to provide an overview of the work to be conducted, describe the
need for and objectives of the measurements, and define the QA/QC activities to be applied to
the project such that the monitoring objectives are attained. The QAPP should detail the sites to
be operated, the measurements to be made, and should include information for staff responsible
for project management, instrument operation, sample collection, laboratory analysis, QA,
training, safety, data review, and data reporting.
Review of the QAPP on an annual basis (or as required by the Region), conduct of audits and
assessments, and implementation of effective corrective action ensure that PAMS Required Sites
and supporting ASLs are in fact achieving program objectives, and, if not, are implementing
corrective actions, as needed.
Two mechanisms will be available to the monitoring agencies for development of their QAPPs:
1. Each monitoring agency may develop its own QAPP and have it reviewed and approved.
Monitoring agencies are familiar with the procedures for developing QAPPs and securing
their approval by EPA Regional staff. Minimally, EPA Regions would review the QAPP
to ensure the performance specifications described in the national QAPP are to be
implemented, their achievement documented, and any deviations from prescribed MQOs
are identified along with information supporting how such deviations will not adversely
impact data quality.
2. Monitoring agencies can utilize the EPA-developed national QAPP and add specific
information and details to the QAPP as described below.
34
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
The PAMS Required Site QAPP for each monitoring agency should include the DQIs and
associated MQOs listed above in Section 3.3, and should include QAPP elements listed in
Section 3.4.1 to ensure that data of sufficient and comparable quality and quantity are generated
across the entire PAMS Required Site network and that intra- and inter-monitoring site
concentration trends may be successfully detected. The PAMS Required Site Program DQO(s)
[if enacted], DQIs, and MQOs take precedent over regional and SLT monitoring objectives for
the associated PAMS sampling and analysis that is performed unless the SLT requirements are
more stringent than those indicated for PAMS. For example, monitoring agencies are free to
prescribe more conservative acceptance criteria (e.g., lower blank acceptance concentrations,
more stringent recovery ranges, etc.).
3.4.1 Development of the National PAMS Required Site Program QAPP
In order to ensure data quality comparability across the PAMS Required Sites, EPA detailed the
main aspects of QA for the PAMS Required Site program within a National PAMS Required
Site QAPP. This National QAPP followed the form described in EPA OA/R-5, EPA
Requirements for Quality Assurance Project Plans14 and the related document, EPA OA/G-5,
Guidance for Quality Assurance Project Plans15 As described in the September 2016 PAMS
QAIP, EPA developed the sections in black font in Table 3-4 below.
Table 3-4. PAMS Required Site National QAPP Elements
QAPP Element
A1
Title and Approval Sheet
A2
Table of Contents
A3
Distribution List
A4
Project/Task Organization
A5
Problem Definition/Background
A6
Project/Task Description
A7
Quality Objectives and Criteria for Measurement Data
A8
Special Training Requirements/Certification
A9
Documentation and Records
B1
Sample Process (Network) Design
B2
Sampling Methods Requirements
B3
Sample Handling and Custody Requirements
B4
Analytical Methods Requirements
B5
Quality Control Requirements
B6
Instrument/Equipment Testing, Inspection & Maintenance
B7
Instrument Calibration and Frequency
B8
Inspection/Acceptance Requirements for Supplies and Consumables
B9
Data Acquisition Requirements for Non-direct Measurements
BIO
Data Management
CI
Assessments and Response Actions
C2
Reports to Management
D1
Data Review, Validation, and Verification Requirements
D2
Validation and Verification Methods
D3
Reconciliation and User Requirements
Monitoring agencies are expected to complete the remaining sections (A3, A8, A9, B3, B6, B8,
and BIO) to describe aspects of their quality system specific to their monitoring organization and
35
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
PAMS program. Monitoring agencies operating PAMS Required Sites may adopt the National
QAPP after adding these details or may modify the National QAPP to address specific portions
of their PAMS monitoring QA program which differ from those in the default document.
3.4.2 PAMS Required Site QAPP Program Deviations
As the PAMS Required Site program is new to many of the monitoring agencies in the network,
monitoring agencies are encouraged to develop policies and procedures as closely as possible to
those described in this TAD, the National QAPP, and National standard operating procedures
(SOPs). Acceptance criteria specified in the QAPP and SOPs are prescribed to meet program
quality objectives; however, method deviations are permitted provided the acceptance criteria for
representativeness, completeness, precision, bias, and sensitivity are met and can be
demonstrated to be scientifically sound and defensible.
Planned method deviations are to be described in the monitoring organization's QAPP and are to
be approved by the cognizant EPA Regional office (or delegate as detailed in the grant
language). Adjustments to storage conditions and holding times or deviations that permit
exceedances to the prescribed method acceptance criteria or to PAMS Required Site MQOs will
require technical justification for Regional approval, as such would allow data of a quality lower
than, and not comparable to, that required to be generated in the PAMS Required Site network.
3.5 Standard Operating Procedures
Each monitoring agency's PAMS Required Site QAPP should list the pertinent SOPs, however
named, to be followed to conduct the PAMS Required Site work. Each monitoring organization
and support organization conducting PAMS Required Site work will develop and maintain
SOPs, however named, which describe in detail the procedures for performing various activities
needed to report ambient air concentrations to AQS, including sample collection, sample
analysis, data reduction, and data review, among others. It is not acceptable to simply cite a
method document (e.g., EPA Compendium Method TO-11 A) or instrument manual as the SOP,
although these documents may serve as the basis for an SOP and may be referenced in the SOP.
EPA recommends following the G-6 SOP format described in Guidance for Preparing Standard
Operating Procedures (SOPs), EPA/600/B-07/001 April 2007:
https://www.epa.eov/sites/production/files/2015-06/dociiments/g6-final.pdf
Instrument manuals and the compendium methods do not include sufficient detail on the specific
procedures and/or equipment necessary to perform the procedures and generally offer several
different procedures or conventions for performing activities or operating equipment. EPA is
developing SOPs for several of the auto-GC instruments, true NO2 with the CAPS instrument,
ceilometer, and carbonyls sample collection and analysis. Monitoring agencies may adopt these
SOPs or may edit the SOPs to more accurately describe procedures specific to their monitoring
site and agency as long as the MQOs for the DQIs are met. The purpose of providing national
SOPs for these instruments and procedures is to reduce the work required by monitoring
agencies for developing thorough SOPs and to encourage consistency across the PAMS network
36
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
such that data are collected in a similar manner irrespective of the site or monitoring
organization.
SOPs should reflect current practice and the work monitoring agencies or support organizations
perform should be in accordance with SOPs. SOPs are to be written with sufficient detail to
enable an individual with limited experience or knowledge of the procedure to successfully
perform the procedure when unsupervised. Production, review, revision, distribution, and
retirement of SOPs should conform to the requirements prescribed by the monitoring agency
document control system such that only the current approved procedures and policies are
followed.
SOPs should prescribe the details of the activities applicable to operation and calibration of field
instruments, field sample collection, preparation and analysis of the samples in the laboratory,
and data review, reduction, and reporting. SOPs should minimally cover the following aspects of
the PAMS Required Site program (note that these aspects may be arranged as desired and
convenient within multiple SOPs, provided each aspect is addressed):
• Calibration, operation, determination of MDLs (where applicable), and maintenance
of instruments for measuring: true NO2, ozone, NOy, speciated VOCs, and
meteorological parameters
• Calibration, operation, and maintenance of carbonyls sampling units as well as
sample collection, preservation, extraction, and analysis of carbonyls samples;
• Calibration of critical support equipment; and
• Data handling (calculations, transformations, etc.), verification, validation, and
reporting.
Additional SOPs should be prepared as necessary to cover routine procedures and repetitive
tasks which, if performed incorrectly, could affect data quality. Such routine activities include,
but are not limited to, sample chain of custody (COC) and performing numerical calculations
(describing rounding, significant figures, etc.).
For portions of measurements, sample collection, or analysis that are contracted or otherwise
performed elsewhere (not by the responsible PAMS Required Site monitoring agency), the
monitoring organization will reference the SOP of the third party in its PAMS Required Site
QAPP. If the support organization (ASL or other entity) is other than the national contract ASL
(the contract for which is administered by EPA), the monitoring agency will maintain a current,
approved copy of the third-party's SOP(s) on file. Monitoring agencies will ensure that such
third-party organization QAPPs and SOPs are available and that third-party laboratories' quality
systems and MQOs are consistent with the requirements of the National QAPP and National
SOPs.
The author of each SOP should be an individual knowledgeable with the activity who has the
responsibility for the veracity and defensibility of the document's technical content. A team
approach may be followed to develop the SOP, especially for multi-tasked processes where
experience of several individuals is critical to the procedure. SOPs should be approved by the
37
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
cognizant manager and QA representative and should be revised when they no longer reflect
current practices. At a minimum, SOPs are to be reviewed by the author and a member of QA to
determine if revisions are needed; these reviews and revisions are to be documented. Review is
recommended to occur annually prior to PAMS season, but should not exceed three years, and
the review and revision period should be prescribed in the monitoring agency's PAMS Required
Site QAPP, Quality Management Plan, or similar controlled policy document. Once a new
version of the SOP is effective, the previous version is retired and made inaccessible so that it
may not be referenced for conducting procedures.
3.6 Good Scientific Practices
Good scientific practices, including instrument calibration and proper recording of observations,
measurements, and instrument conditions, are equally important in both the field and in the
laboratory. Such practices are necessary to generate data which are consistent, comparable,
standardized, traceable, and defensible. Appropriate aspects of good laboratory and field
practices are to be detailed in each agency's PAMS QS. The need for such practices is given
below.
3.6.1 Data Consistency and Traceability
To be able to verify that the PAMS Required Site network generates data of quantity and quality
sufficient to evaluate the PAMS quality objectives, data collection and generation activities are
to be traceable to calibrated instruments, certified standards, and to activities conducted by
individuals with the appropriate and documented training. Traceability in this case refers to
ensuring the existence of a documentation trail which allows reconstruction of the activities
performed to collect and analyze a sample and to the certified standards and calibrated
instrumentation employed to determine target analyte concentrations. To specifically ensure
attainment of overall network bias requirements, each reported concentration is to be traceable to
a measurement of known accuracy, be it from an analytical balance, volumetric flask, calibrated
auto-GC, mass flow controller, etc. Maintaining this traceability from sample collection to final
results reporting assures that PAMS data are credible and defensible, and that the root cause of
nonconformances may be found and corrected which thereby enables continuous improvement in
PAMS program activities. Instrument calibration specifications and frequencies are provided
within the individual methods sections in Sections 4, 5, 6, and 8.
3.7 References
1. Quality Assurance Handbook for Air Pollution Measurement Systems. Volume II - Ambient
Air Quality Monitoring Program, EPA-454/B-17-001, U. S. Environmental Protection
Agency, January 2017.
2. Quality Assurance Handbook for Air Pollution Measurement Systems. Volume IV -
Meteorological Measurement, EPA-454/B-08-002, U. S. Environmental Protection Agency,
January 2008.
3. On-site Meteorological Program Guidance for Regulatory Modeling Applications, EPA-
454/R-99-005. Research Triangle Park, NC: U. S. Environmental Protection Agency,
February 2000.
38
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
4. Guide to Meteorological Instruments and Methods of Observation, WMO No. 8 Geneva,
Switzerland: World Meteorological Organization, 2014.
5. Technical Assistance Document (TAD) For Precursor Gas Measurements in the NCore
Multi-Pollutant Monitoring Network, Version 4, EPA-454/R-05-003, September 2005.
Available at (accessible March 2018):
https://www3.epa.gov/ttnamtil/ncore/eiiidance/tadversion4.pdf
6. Environmental Protection Agency. (December 2002). Guidance for Quality Assurance
Project Plans. EPA QA/G-5 (EPA Publication No. EPA/240/R-02-009). Office of
Environmental Information, Washington, DC. Available at (accessed March 2018):
https://www.epa.gov/sites/prodiiction/files/2015-06/dociim.en.ts/g5-final.pdf
7. Standard Practice for Characterizing Surface Wind Using a Wind Vane and Rotating Cup
Anemometer, ASTM Designation: D 5741-96, Annual Book of ASTM Standards, Section
11, 1996.
8. National Guidance for the Permitting, Monitoring, and Enforcement of Water Quality-based
Effluent Limitations Set Below Analytical Detection/Quantitation Levels, Draft Report.
United States EPA, 1994
9. Glaser, J. A., Foerst, D. L., McKee, G. D., Quave, S. A., & Budde, W. L. (1981). Trace
analyses for wastewaters. Environmental Science and Technology, 75(12), 1426-1435.
10. Boyd, R. K., Basic, C., & Bethem, R. A. (2008). Trace Quantitative Analysis by Mass
Spectrometry. West Sussex, England: John Wiley and Sons.
11. Childress, C. J. O., Foreman, W. T., Connor, B. F., & Maloney, T. J. (1999). New Reporting
Procedures Based on Long-Term Method Detection Levels and Some Considerations for
Interpretations of Water-Quality Data Provided by the U.S. Geological Survey National
Water Quality Laboratory - Open-File Report 99-193. US Geological Survey. Available at
(accessed March 2018): http://water.iisgs.gov/owq/OFR 99-193/
12. Keith, L.H. (1992). Environmental Sampling and Analysis: A Practical Guide. Chelsea, MI:
Lewis Publishers, pp. 93-119.
13. Definition and Procedure for the Determination of the Method Detection Limit, Revision.
EPA Office of Water, EPA 821-R-16-006, December 2016. Available at (accessed March
2018): https://www.epa.gov/sites/production/files/2016-12/documents/m dl -
procedut >df
14. Environmental Protection Agency. (March 2001). EPA Requirements for Quality Assurance
Project Plans. EPA QA/R-5 (EPA Publication No. EPA/240/B-01-003). Office of
Environmental Information, Washington, DC. Available at (accessed March 2018):
https://www.epa.gov/sites/prodiiction/files/2016-06/dociiments/r5-final 0 ixlf
15. Environmental Protection Agency. (December 2002). Guidance for Quality Assurance
Project Plans. EPA QA/G-5 (EPA Publication No. EPA/240/R-02-009). Office of
Environmental Information, Washington, DC. Available at (accessed March 2018):
https://www.epa.gOv/sites/prodiiction/files/2015-06/dociim.en.ts/g5-final.pdf
39
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
4.0 VOLATILE ORGANIC COMPOUNDS BY AUTO-GC
Each agency is to prescribe in an appropriate quality systems document, such as an SOP, or
equivalent, its procedures for sampling and analysis of speciated VOCs by auto-GC. Various
requirements and best practices for such are given in this section. Note that regardless of the
specific procedures adopted, the MQOs in Table 3-1 and the QC specifications as given in Table
4-4 should be met.
VOCs are defined as organic compounds having a vapor pressure greater than 10"1 Torr at 25°C.1
PAMS VOCs consist of hydrocarbons, compounds consisting solely of carbon and hydrogen
with the exception of several optional compounds which also contain chlorine or oxygen. These
compounds, containing between 2 to 12 carbon atoms, noted as C2 to C12, are listed in Table 4-1.
PAMS Required Sites are to monitor for those VOCs listed in Table 4-1 as priority compounds
and are encouraged to monitor the concentrations of those VOCs listed as optional compounds,
particularly when they are considered to contribute to the formation of ozone in the CBS A
represented by the PAMS Required Site. Concentrations of monitored VOCs are to be reported
to AQS.
Table 4-1. PAMS Priority and Optional VOCs Measured by Auto-GC
Prioriit C ompoiinds
Oplioiiiil Compounds
1,2,3-trimethylbenzene
m-ethyltoluene
1,3,5-trimethylbenzene
isopropylbenzene
1,2,4-trimethylbenzene
n-butane
1-pentene
m-diethlybenzene
1-butene
n-hexane
2,2-dimethylbutane
methylcyclohexane
2,2,4-trimethylpentane
n-pentane
2,3,4-trimethylpentane
methylcyclopentane
benzene
o-ethyltoluene
2,3 -dimethylbutane
n-decane
cis-2-butene
0-xylene
2,3 -dimethylpentane
n-heptane
ethane
p-ethyltoluene
2,4-dimethylpentane
n-nonane
ethylbenzene
propane
2-methylheptane
n-octane
ethylene
propylene
2-methylhexane
n-propylbenzene
isobutane
styrene
2-methylpentane
n-undecane
isopentane
toluene
3-methylheptane
p-diethylbenzene
isoprene
trans-2-butene
3-methylhexane
trans-2-pentene
m&p-xylenes
total non-methane
organic carbon
(TNMOC)
3-methylpentane
a/P-pinene
acetylene
1,3 butadiene
cis-2-pentene
carbon tetrachloride
cyclohexane
ethanol
cyclopentane
tetrachloroethylene
4.1 Priority and Optional Volatile Organic Compounds
Target VOCs are emitted from a variety of sources including mobile sources, biogenic sources
(pine tree forests), energy production (natural gas extraction), and industry (petroleum
refineries), among others, and are typically measured in ambient air at concentrations ranging
from single parts per trillion (ppt) to hundreds of parts per billion (ppb) by volume (ppbV). Their
concentrations will be expressed on a carbon concentration basis as ppbC, which is the
concentration in ppbV multiplied by the total number of carbon atoms in the compound. For
40
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
example, if a sample contains both propane (3 carbon) and hexane (six carbon) at 6 ppbV, the
concentrations of these two compounds are 18 ppbC and 36 ppbC, respectively.
Based on the evaluation EPA conducted in 2011 to determine the relative importance of the
existing PAMS target compound list at that time, EPA created a two-tiered list of compounds
detailed in a 2013 memo.2 Compounds were assigned to the 'priority' category based on having
high overall reactivity-adjusted average concentrations, high reactivity-adjusted average
concentrations during 9 a.m. rush hour on high ozone days, high reactivity-adjusted average
concentrations based on geography, and on being as hazardous air pollutant and/or a high
priority secondary organic aerosol precursor. An additional seven compounds, alpha-pinene,
beta-pinene, 1,3-butadiene, benzaldehyde, carbon tetrachloride, ethanol, and tetrachloroethylene,
were assigned as optional. However, none of these seven compounds were ultimately included in
the final October 2, 2017 PAMS priority compound list:
https://www3.epa.gov/ttnamtil/files/ambient/pams/targetlist.pdf
Although they have been identified as important precursors in ozone formation, moisture
management techniques for some of the currently available auto-GC instruments do not permit
accurate quantitation of the pinene isomers. This limitation also impacts the ability to analyze for
ethanol, which was added as an optional compound in 2017 to provide data on the potential
impact on ozone formation attributable to changes in fuel usage (e.g., biofuels). Moreover,
retention time standard gas mixtures typically provided by EPA do not include the halogenated
compounds, which are difficult for monitoring agencies to identify without purchasing additional
standards and increasing the complexity involved in instrument calibration.
Each PAMS Required Site is to measure and report TNMOC for each sampled hour. TNMOC is
defined as the sum of the concentration of all identified and unidentified compounds in the auto-
GC chromatograms for the hourly sample. TNMOC is different than the PAMS hydrocarbon
parameter (PAMSHC), which is defined as the sum of the identified PAMS target (priority and
optional) compounds for the hour and is not required to be reported to AQS. PAMSHC can
readily be derived by summing the measured concentrations of the target compounds and can be
easily added to data analysis routines by data users polling data from AQS.
TNMOC is calculated by determining the total concentration of chromatographic peaks in the
light and heavy hydrocarbon chromatograms for the hour. The concentration of light
hydrocarbons in ppbC is determined by summing the total peak area in the C2 to C(,
chromatogram for all peaks eluting between the first and last eluting target compounds (typically
ethane and 1-hexene, respectively) and multiplying this area by the response factor for propane
or butane, depending on which compound is the calibrant gas for the specific flame ionization
detector (FID). The concentration of heavy hydrocarbons in ppbC is determined by summing the
total peak area in the C(, to C12 chromatogram for all peaks eluting between the first and last
eluting target compounds (typically hexane and dodecane, respectively) and multiplying this area
by the response factor for benzene. These light and heavy concentrations are summed to
determine the TNMOC for the hourly sample.
41
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
Determination of TNMOC for auto-GC systems employing Deans switching for analysis operate
with one preconcentrator and chromatographically separate one gas stream. This Deans switch
convention makes the TNMOC determination straightforward as the collected sample is
separated and detected by one FID discretely and partway through the separation, the gas stream
is rerouted to another FID such that compounds are detected on only one FID channel at a time.
Auto-GCs that employ more than one preconcentration module will have target analytes detected
on more than one FID channel and thus require a procedure to subtract target and unknown peaks
from one of the chromatogram channels to ensure they are not counted twice in the TNMOC
determination. Monitoring agencies should ensure that TNMOC determinations for such
systems correctly include such peak responses only once.
Monitoring agencies should periodically compare the TNMOC and PAMSHC values to
determine the percentage of TNMOC consisting of unidentified compounds. If the TNMOC
exceeds the total PAMSHC by more than 20%, monitoring agencies should attempt to identify
the unknown hydrocarbon species. Such an investigation may identify hydrocarbons that are of
importance to ozone formation at the site and monitoring agencies may consider measuring these
compounds at the site on an ongoing basis.
4.2 Instrumentation - Measuring VOCs with an Auto Gas Chromatograph with Flame
Ionization Detection
4.2.1 Summary of Method
PAMS VOCs cover a wide range of volatility (vapor pressures of approximately 0.003 to 44 atm
at 20°C) and molecular weight, making their collection and analysis challenging to perform with
a single instrument. The auto-GC has proven effective for measuring the hydrocarbon VOCs of
interest to PAMS and is specified for installation at all PAMS Required Sites. Note that while the
carbonyl compounds are considered VOCs, they are measured by EPA Compendium Method
TO-11A as described in Section 5.0 or by continuous formaldehyde measurement instruments
(which are not covered in this TAD).
The 1998 PAMS TAD described a general method for operation of auto-GCs for measurement of
PAMS VOCs based on EPA Compendium Method TO-123 and elaborated on the method to
perform speciated identification and quantitation. Instrument manufacturers have updated the
auto-GCs and developed technologies to improve the measurement of PAMS VOCs; however,
the general technique has not changed substantially from the technology available in 1998.
Several auto-GCs are commercially available that were evaluated in EPA studies conducted in
20134and 2015.5 The instruments evaluated in these studies that demonstrated meeting
acceptable performance for PAMS are configured differently; however, they operate with the
same basic convention. A general description of the method follows. Ambient air is drawn
through the sampling inlet by a vacuum pump. Sample flow is regulated by a mass flow
controller (MFC) and moisture in the sampled air stream (or a portion of the air stream) is
removed. The sample stream is then drawn through a preconcentrator trap (or traps) typically
cooled to -10°C or less with a Peltier cooling device (some systems trap at ambient temperature
or colder than -10°C). The traps typically contain one or more sorbents targeted to efficiently
retain the compounds of interest and allow nitrogen, oxygen, carbon dioxide, and argon to pass
42
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
through the trap. Once the sampling period has completed, the sample flow to the trap is stopped
and the trap is isolated and typically flushed with carrier gas to remove any residual moisture
(dry purge). The trap is then rapidly heated for several minutes to liberate the trapped
hydrocarbons which are introduced to the GC portion of the instrument for separation and
detection with an FID (or FIDs). FID response is established by analysis of known concentration
standards of target hydrocarbons. Target compounds are identified by their retention time
established by analysis of a known standard(s).
The auto-GC employs two separate columns and two discrete FIDs to cover the molecular
weight range of target compounds. Lighter hydrocarbons with lower boiling points (C2 to C6) are
separated with a porous layer open tubular (PLOT) Ah03-Na2S04 column and detected with a
dedicated FID. Heavier hydrocarbons with higher boiling points (C6 to C12) are separated with a
polydimethylsiloxane (PDMS)-coated capillary column (this column is also commonly referred
to as the "BP-1" channel - "BP-1" is a proprietary term) and detected with a separate dedicated
FID.
Quantitation of the target analytes is (typically) based on the carbon response of either propane
(for compounds containing 2 to 6 carbon atoms) or benzene (for compounds containing 6 or
more carbon atoms). Ongoing analysis of QC samples such as calibration check standards and
blanks demonstrate the instrument calibration remains valid and the instrument is free from
carryover or memory effects. The concentration of each target analyte and TNMOC is reported
in units of ppbC.
4.2.2 Sample Introduction and Collection
4.2.2.1 Probe Inlet
Sampling probe location and siting criteria are described in Section 3.3.1.2. The inlet probe and
inlet line must consist of borosilicate glass or chromatographic-grade stainless steel. 40 CFR Part
58 Appendix E Section 9(b) allows for an equivalent to these materials; however, the use of
other materials such as polytetrafluoroethylene (PTFE) or perfluoroalkoxy alkane (PFA) Teflon®
have shown to be problematic for quantitative transfer (for example, due to memory effects of
higher molecular weight (lower boiling point) compounds) and their use is strongly discouraged.
Experience has shown that inlets constructed of borosilicate glass or chromatographic stainless
steel, particularly steel with a silicon ceramic coating, provide the best performance. VOCs
experience adsorption to and from fluorinated ethylene propylene (FEP) Teflon®, therefore it is
not appropriate for use in sampling inlets. Heating of the glass and stainless steel inlet pathways
aids in the quantitative transfer of higher boiling point VOCs, and is recommended where
possible.
The flow rate through the sampling inlet probe to the auto-GC inlet must be selected so that the
residence time in this line is 20 seconds or less, as required by 40 CFR Part 58 Appendix E
Section 9(b). Typical air sampling flow rates for auto-GCs are approximately 10 to 30
mL/minute, which are insufficient to meet the 20-second residence time unless the inlet tubing
run is very short (e.g., 1 meter). At such low flow rates, the inlet tubing requires a small internal
flow path volume that may be achieved by narrow bore (e.g., 1/16th inch) tubing; however, such
narrow diameter tubing causes unacceptably elevated pressure drop which pumps onboard the
43
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
auto-GC may be unable to overcome. Alternatively, including an inlet manifold or single larger
bore inlet tube in the flow path and utilizing a constant high flow pump or fan to move ambient
air through this tubing can ensure a constant supply of fresh air closer to the auto-GC inlet where
a short run of 1/8- or 1/16-inch chromatographic stainless steel can be connected. For inlet
configurations requiring vacuum blowers, if the manifold blower fails, the auto-GC will sample
stagnant air which is not representative of the ambient air as the residence time will exceed 20
seconds. Particular attention should be given to ensuring manifolds and wider bore inlets with
higher flows are cleaned at a regular interval to avoid buildup of particulate matter that can bias
VOC concentrations. Particulate matter residue can function as a sorbent and adsorb VOCs
during periods of high concentration and release them during times of lower concentration.
Whether employing a standalone or manifold inlet, it is recommended that the inlet be heated. In
humid environments, when warm, moist air enters the inlet and is drawn into the shelter, the
lower temperature inside the shelter causes the water vapor in the sampled atmosphere to
condense within the inlet tubing. This liquid water can act as a sink for polar compounds and is
generally problematic for sample collection and analysis as the pooled water can be drawn into
analyzers. This is particularly the case when reactive gas species analyzers (such as NO2, ozone,
and SO2) are attached to a manifold. Heating the inlet manifold and inlet lines reduce the
likelihood of condensation and its related complications.
If connected to a glass manifold, the auto-GC inlet should be connected to a manifold port
corresponding to its flow demand. Instruments having lower flow demand, such as auto-GCs,
should be connected to ports closer to the inlet end of the manifold.
Guidance on suggested manifold inlet flows (whether one or multiple instruments are connected)
cover a range of approximately three- to five-fold above the total airflow draw of the instruments
connected to the manifold, or at a rate equal to the total sampling requirement plus 140 L/minute
(EPA QA Handbook Volume II, Revision 1 December 2008 Section 7.3.3).6
The air flow through the manifold (or wide bore tubing inlet) should not be so great as to cause
the pressure inside the manifold to be more than 1 inch of water below ambient pressure. The
pressure inside the manifold can be assessed as follows. Construct the manifold and insert a
pitot tube with an attached manometer into the center of the manifold where air flow is expected
to be highest. Commence air flow in the manifold and measure the flow of the air inside the
manifold according to the manometer measurement. The pitot tube measures air flow by
measuring the pressure differential across the probe which has a dynamic port (exposed directly
to the flowing air) and a static port (exposed to the air but shielded from direct pressure by
flowing air). At the same time, attach a water manometer to a sampling port to measure the
pressure drop in the manifold compared to the ambient pressure. Measure the flow rate with the
pitot probe and measure the vacuum with the water manometer. Adjust the flow rate to be
between three to five times the total instrument airflow demand while keeping the pressure inside
the manifold less than 1 inch of water below ambient. If this is impossible, the diameter of the
manifold or wide bore tubing is too small.6
44
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
4.2.2.2 Sample Collection Requirements
Auto-GC sample collection for PAMS requires that ambient air be collected for a minimum of a
40-minute period for each hour (most systems will be configured to sample for 40 minutes -
longer sampling periods are not advised as the instrument may not have sufficient time to
properly cool and reset for the subsequent sampling event). Sample collection is to commence at
the beginning of the hour. Due to discrepancies between the instrument computer clock and the
true time or extended cooling times to ready the GC or thermal desorber to the proper
temperature for the next sample, the sample start time may deviate from the beginning of the
hour. In such cases, for the sample hour to be valid, the sample collection must start between 10
minutes before the beginning of the hour and 30 minutes after the top of the hour to ensure that
minimally 30 minutes of the 40-minute sampling period (75%) are sampled within the hour. For
example, for a sample to represent the hour between 8:00 and 9:00 a.m., sampling would begin
between 7:50 a.m. and 8:30 a.m. Samples with start times outside of this window will be
invalidated as "AG" when reported to AQS.
Clocks controlling sampling timing for some auto-GC systems may gain (run faster) or lose (run
slower) time when compared to the true time. If the amount of time gained or lost is significant
such that the change is several minutes over the course of a week, it is recommended that clocks
be adjusted regularly (e.g., weekly) to ensure the clock discrepancy does not result in sample
collection hours where there is insufficient sample collected during the designated hour. Clocks
should remain as close to the true time as practical. Clock adjustments and errors should be
documented and adjustments to sample collection times made for reported data. Monitoring
agencies should closely review sample hours that are invalidated for sampling start times outside
the appropriate window due to clock time gain or loss. Several PAMS agencies have reported
instances of instrument clock inaccuracy and instances of sampling time delay due to extended
cooling or instrument readying times. The combination of these two aspects may result in
sampling start times outside the prescribed acceptance window.
As described in Section 3.3.1.1, sampling is to occur for the duration of PAMS season, defined
as June 1 to August 31 each calendar year. Monitoring sites are encouraged to conduct sampling
outside this season, as well, when peak ozone concentration "season" in the CBSA extends
beyond this three-month period. Some locations that experience year-round elevated ozone
concentrations may extend sample collection to occur throughout the year to capture seasonal
changes in the precursor mix as such concentrations could provide additional information useful
in developing mitigation strategies.
4.2.3 Automatic Gas Chromatograph (Auto-GC)
PAMS VOCs are to be measured at each PAMS site via an auto-GC, the method for which is
described generally in EPA Compendium Method TO-12.3 Monitoring agencies are responsible
for selecting instruments and support equipment that will meet the PAMS MQOs. Due to the
variety of instruments and associated support equipment, the technical information provided in
subsequent sections is purposely not instrument-specific. Technical details tailored to the specific
auto-GC instruments are described within the national PAMS auto-GC SOPs developed for the
PAMS Required Site network.
45
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
4.2.3.1 Instrument Sensitivity
Depending on the instalment configuration, the auto-GC samples ambient air for approximately
10 to 30 mL/minute for a minimum of 40 minutes for a total collected volume of approximately
400 to 1200 mL. The instrument designated flow rates and durations are suggested or set by the
manufacturer to maximize the collected mass of each target compound while minimizing
breakthrough. At the designated flow rates suggested by the instrument manufacturer,
breakthrough should not be an issue for new, properly conditioned preconcentrator traps. In
general, the greater volume of sample that can be collected onto the sorbent trap (by increasing
the flow rate or duration, or both), the lower the concentration the instrument is able to measure
as the mass delivered to the FID is directly proportional to the air volume collected. However,
the potential of breakthrough increases with increasing flow rate and/or collection time.
Breakthrough is more likely to occur for light (C2 and C3) compounds, particularly for acetylene,
which requires strong sorbents to retain. The auto-GC instruments should be operated at their
recommended conditions to ensure optimum performance (quantitative capture and subsequent
desorption) of the target compounds.
The instrument's concentration at which it can positively detect a given target compound is
established by determining the MDL, which is discussed in Section 4.3 for auto-GC.
4.2.3.2 Moisture Management
Water vapor in humid ambient air samples is problematic for VOCs analysis by thermal
desorption (TD) auto-GC for several reasons. Foremost among these is that preconcentrator traps
are typically cooled to temperatures of -10°C or less to aid in the effective retention of low
boiling point compounds on the sorbent substrate. If not removed, the water vapor would freeze
in the trap, reducing and eventually halting the gas flow through the trap. Secondly, moisture, if
allowed to pass through to the PLOT column used to separate light hydrocarbons, changes the
characteristics of the porous layer Ab03-Na2S04 stationary phase, resulting in poor
chromatographic peak shapes and negative bias in observed retention times for the C2 to C(,
hydrocarbons. Such changes in retention times and poor chromatography can render compound
identification difficult or impossible. Note that heated inlet manifolds and lines do not remove
moisture from the sampled air stream but prevent condensation in the inlet flow path.
Effective and reproducible moisture removal from the sample stream is important for achieving
consistent chromatographic performance. Moisture management on auto-GCs is currently
addressed with one of two different techniques.
Nafion™ Dryers: As of the time of publication of this TAD, drying of gas streams with
Nafion™ tubing is available for the auto-GC instruments approved for use in the EPA auto-GC
evaluation studies. Nafion™, manufactured by PermaPure, is a semi-permeable fluoroelastomer
polymer membrane that allows water to pass through the membrane due to a concentration
gradient.7 For the PAMS auto-GC application, the humid sampled atmosphere containing water
vapor enters the drying tube contained in an outer sheath. Dry sheath (purge) gas (zero air or
nitrogen) with a low dewpoint is supplied to the outside of the drying tube and flows in a
direction opposite that of the sample gas flow. Water molecules in the humid sample stream pass
by gradient through the membrane into the dry sheath gas, drying the sample stream. Variables
affecting the drying efficiency include the humidity of the sample stream, membrane surface
46
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
area, membrane temperature,7 residence time in the dryer, pressure of the sample stream,
dewpoint of the purge gas, and flow of the purge gas. In general, Nafion™ dryers are most
effective when sample residence time, purge gas flow, and membrane surface area are
maximized and membrane temperature and purge gas dewpoint are minimized. Previous studies
have shown that drying efficiencies of greater than 80% 8'9>10'n can be attained with Nafion™
dryers. Users should note that if the sample stream humidity is increased (due to a rain event or
similar) and the other variables are kept constant, the drying will be less effective and permit
more water to be passed to the preconcentrator. While the additional water may not result in trap
freezing, the additional water may interfere with chromatographic separation resulting in
shortened retention times for target analytes on the light hydrocarbon channel (PLOT column
analytes). Such retention time shifting may be transient and retention times will usually stabilize
and revert to their means once the ambient relative humidity levels decrease.
Nafion™ dryers have several drawbacks when employed for drying gas streams analyzed for
VOCs by auto-GC. A study performed in 199212 demonstrated that VOCs' recoveries were lower
when passed through a Nafion™ dryer, collected into a canister, and analyzed by GC/FID. For
example, total non-methane hydrocarbons were reduced by 9 to 23%, olefins by 6 to 19%,
paraffins by 8 to 26%, and aromatics by 3 to 21% reduction when compared to the same analysis
omitting a Nafion™ dryer. Furthermore, the Nafion™ polymer contains sulfonyl moieties that
function to permit the transport of the water molecules across the polymer membrane. These
sulfonyl groups act as strong acids and catalyze the conversion of ketones and aldehydes to
alcohols which may then be transported through the membrane.7 Note that the membrane is not
directional and that alcohols and carbonyls (once converted to alcohols) in the sample stream
may be lost to the purge gas or those in the purge gas may be introduced (as alcohols) to the
sample stream. These alcohols may subsequently interfere with target compound analysis or
result in unidentified peaks that will be incorporated into the TNMOC result. Finally, Nafion™
has been documented to interfere with the analysis of monoterpenes such as alpha-pinene and
beta-pinene (two biogenic compounds of interest in ozone formation); however, Nafion™ does
not appear to have a similar impact on isoprene.9
After several weeks or months of use, Nafion™ dryers may act as a source of small chain
alkenes such as ethylene or propylene. For this reason, the dryer should be replaced at the
beginning of each season, or more frequently as indicated by the changes in the recovery of the
target compounds caused by carryover or memory effects in system blanks or by the loss of
target compounds in the CCV checks and/or retention time standard (RTS) checks. Previous
studies13 have shown that attempts to "regenerate" the Nafion™ by heated purging causes loss
and rearrangement of the C4-C6 alkenes. For this reason, regeneration of the drying tube by
heated purging is not recommended.
Electronic Cooling to Remove Moisture: At the time of this TAD's publication, several of the
auto-GC manufacturers developed or were developing dryers that employed other drying
technologies such as Peltier cooling to remove water by freezing it out of the sample stream prior
to its introduction to the preconcentrator trap(s). EPA plans to evaluate available drying systems
to ensure data quality is acceptable and comparable to, or better than, that obtained with auto-
GCs equipped with Nafion™ dryers. These evaluation studies had not been completed before
47
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
this document's publication. When completed, a report of the evaluation study will be available
on the following AMTIC webpage:
https://www3.epa.eov/ttnamtil/pamsreeng.html
Examination of auto-GC data from a PAMS monitoring agency employing an electronic cooling
dryer14 demonstrated virtually no RT shifts over the course of several weeks for the C2 to C(,
channel, indicating that the electronically cooled dryer sufficiently removed water from the
sampling stream.
4.2.3.3 Thermal Desorption
The preconcentration step involves trapping the compounds of interest from the dried sample
stream within sorbent bed(s) held at ambient temperature or cooled to approximately -10°C or
less. The dried sample stream is passed through the sorbent bed(s) in the preconcentrator trap(s)
and the target compounds are retained within the preconcentrator trap(s) during the sample
collection period. At the end of the sample collection period, the preconcentrator trap is purged
with inert gas, isolated and heated rapidly to > 300°C and the target compounds are liberated and
sent by backflushing to the GC for separation, detection, and subsequent quantification.
Auto-GC Preconcentrator Traps: Auto-GC preconcentrator traps consist of a quartz or
stainless steel tube containing a sorbent or series of sorbents arranged in "beds." These beds are
maintained within the quartz or stainless steel tube, typically retained by a glass frit or glass wool
plug that holds the small granules of sorbent in place. Beds are separated by these frits or plugs
within the tube, and a similar plug is installed on the trap outlet and held in place with a spring to
provide constant tension to the trap beds.
Sorbents: Sorbents effective at trapping and releasing very volatile compounds (such as
ethylene, propane, and propylene) will retain and not release less volatile compounds (such as
decane, 1,2,4-trimethylbenzene). Weaker sorbents may perform well for the less volatile
compounds but will not effectively retain the very volatile compounds. Due to these differences
in strength, instrument manufacturers have specified combinations of sorbents to effectively trap
and release (absorb and desorb) the variety of compounds of interest. Gas samples (whether
ambient air, a blank, or a standard) are first introduced to the weakest sorbent bed which retains
the less volatile compounds and some fraction of the highly volatile compounds. Some of the
highly volatile compounds pass through the weak sorbent bed and are then retained on the
stronger sorbent(s) bed(s). Upon trap desorption, the trap(s) is heated quickly to > 300°C and
backflushed (gas flows in the direction opposite of sampling) and the compounds of interest are
desorbed from the sorbent beds.
Trap Conditioning: When an auto-GC system is new or when replacing the trap, it is
recommended to condition the trap prior to use to remove contaminants and interferences on the
sorbent(s). This may be done by performing a prolonged baking of the trap at an elevated
temperature (e.g., 200 to 300°C) while flowing dry, inert carrier gas (hydrogen or helium)
through the trap. The conditioning temperature is dependent on the sorbent and is typically
recommended by the sorbent or trap manufacturer. Note that preconcentrator traps with multiple
sorbent beds should be conditioned at the lowest tolerated temperature of the sorbents contained
48
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
in the trap.15 For example, if a sorbent trap contains both Tenax®-TA (recommended
conditioning temperature 320°C) and Carbopack™ (recommended conditioning temperature
350°C), the trap conditioning temperature should not exceed 320°C to avoid damaging the
Tenax. If possible, the temperature during conditioning should be raised slowly in a stepwise
manner (e.g., 20°C/hour) until the conditioning temperature is achieved. Bakeout periods of
approximately 48 hours at the conditioning temperature have shown to be effective; however,
manufacturer recommendations should be followed. After this 48-hour period, most of the trap
contamination will have been removed; however, lower concentration (sub-ppbC) levels of target
compounds may still evolve from the trap for an extended period of several weeks. Operators are
encouraged to install the traps well in advance of the monitoring season and to analyze
humidified blanks and ambient air for several weeks to ensure trap contaminants have been
reduced to acceptable levels (this is particularly important if traps cannot be conditioned prior to
use). In general, this conditioning period is essential to ensure that support equipment (e.g., zero
air generators, connecting lines, and the sampling flow path) are sufficiently clean. If an
extended conditioning period is infeasible or impractical, the monitoring agency should analyze
humidified zero air blanks and/or ambient air for minimally 24 hours to minimize interferences
and contaminants, and longer periods are recommended. Auto-GCs should be recalibrated after
any trap replacement.
Trap Lifespan: It is important to follow the instrument manufacturer recommendations for
preconcentrator trap replacement. Unless recommended to be more frequent, preconcentrator
traps should be replaced prior to each PAMS season. Sorbents can degrade over time due to the
hundreds of heating and cooling cycles experienced (there are approximately 2200 cycles for the
three-month PAMS season) and eventually require replacement. Decreases in the GC-FID
carbon response factors for propane and/or benzene may indicate sorbents have degraded and the
trap(s) is due for replacement. Likewise, decreases in recoveries of light (C2-C3) or heavy (C9-
C10) hydrocarbons without concomitant decreases in C4-C7 hydrocarbons may also signal
degradation of sorbents.
Trap Failure: Due to manufacturer defects or prolonged use, preconcentrator traps may fail.
Traps are subjected to numerous aggressive heating and cooling cycles and the sorbents
experience pressure differentials (e.g., 40 pounds per square inch [psi]) when the systems switch
from cooled sample adsorption to heated sample desorption. Trap housings may crack and cause
a leak in the system. Glass frits or plugs retaining sorbent beds may migrate, permitting sorbent
granules to leak into the other sorbent bed(s) or into the system flow path. A cracked trap or leak
of sorbent into another sorbent bed may be a simple repair requiring trap replacement; however,
a leak of sorbent out of the trap entirely and into the flow path may result in significant system
downtime requiring cleaning and potential component replacement within portions of the auto-
GC. Sudden trap failure typically causes a marked change in the response of the very light (C2-
C4) or very heavy (C8-C10) hydrocarbons. Other signs of trap failure include sudden retention
time shifts and poor chromatography (broad, split, and/or tailing peaks, or unresolved unknown
peaks or "blobs").
49
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
4.2.3.4 Separation of Compounds
In order to effectively separate the target compounds of interest, auto-GCs employ two separate
analytical columns. Auto-GC columns are narrow bore tubing with an interior coating selected to
separate the compounds in the gas mixture.
The polydimethylsiloxane (PDMS) column efficiently and effectively separates the heavier (C6
to C12) hydrocarbons but does not effectively separate the lighter hydrocarbons at ambient
temperature. Lighter hydrocarbons (C2 to C6) are separated within a PLOT column coated with
Ah03-Na2S04 stationary phase. This stationary phase is effective at separating compounds by
their degree of hydrogen saturation (in addition to their volatility) due to the affinity compounds
exhibit to the inorganic oxides. Unsaturated compounds have a higher affinity for the inorganic
oxide stationary phase and will be retained in the column longer than their saturated
counterparts. For example, when separating propane (C3H8) and propylene (C3H6), both C3
compounds, the more highly saturated propane has a lower affinity for the column stationary
phase and will elute from the column before propylene.
Manufacturers have developed instrument methods that optimize the chromatography and GC
run time for separation of the PAMS target VOCs. For example, each auto-GC manufacturer has
specified the carrier gas (either hydrogen or helium), column characteristics (diameter, length,
and stationary phase), oven temperature program, GC injection split, and carrier gas flows.
4.2.3.5 Flame Ionization Detection
The auto-GC instruments evaluated in the EPA field and laboratory studies and selected for use
at PAMS Required Site employ FIDs (note that one GC with a mass spectrometer detector was
approved; however, its use is outside the scope of this document). FIDs are advantageous for use
in auto-GCs for PAMS as they are robust and require little maintenance, have a predictable linear
and stable carbon response, and provide sufficient sensitivity to measure the target compounds at
the desired concentration levels.
FIDs operate by creating and measuring the ions created during the combustion of the organic
molecules in the column eluent. The ions create a current within the FID and the current is
measured. This measured current is approximately proportional to the number of carbon atoms in
the combusted gas stream, which is compared to the FID response to a standard of known
concentration to determine the concentration of the unknown sample. FIDs require a fuel
(hydrogen) and an oxidizer (oxygen in zero air) to create and maintain the ionizing flame.
To maintain consistency with hydrocarbon precursor data historically collected for PAMS, the
concentration calculation convention for PAMS target analytes will be unchanged from the
longstanding convention. That is, target compound concentrations will be reported assuming that
each compound's carbon response is the same for propane or butane (for the C2 to C(,
compounds) or benzene (for the C(, to C12 VOCs). It is important to note that this is an
assumption, and that FIDs respond differently to hydrocarbons based on their amount of
hydrogen saturation and the presence of oxygen and/or halogen atoms.16'17
The following paragraph is informational; concentration data generated by auto-GC should not
be adjusted to correct for theoretical FID responses prior to reporting to AQS. Note that one
50
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
auto-GC manufacturer has incorporated a "response factor "for target compounds other than
butane and benzene based on analysis of a calibration standard containing the suite of PAMS
target analytes. These "response factors " may be employedfor reporting measured
concentrations; however, such data should be identified according to the associated method
code. Experimental studies have been performed1819 to determine the effective carbon number
(ECN) which provides a concentration correction based on the FID response of each compound
and the relationship to the concentration response of a saturated hydrocarbon such as propane,
butane, or benzene. For example, aliphatic compounds which substitute a chlorine atom for a
hydrogen atom show a lower FID response for that carbon by approximately 12%. After data
are reported to AQS, PAMS data users can apply the ECNfactors to the concentrations of PAMS
target compounds reported to AQS to correct for the ECN to determine more accurate in-air
concentrations of target analytes. Unless data users are familiar with the impact of ECN or
similar adjustments, they should utilize concentration data as reported without adjustment.
While FIDs are robust and relatively maintenance-free, the flame in the detector may be
extinguished (for example, due to temporary disruption in the supply of fuel, oxidizer, etc.; see
below) which will result in no signal for that FID. In order to verify the FID flame is lit, a cold
piece of metal or glass (chrome plated wrench or small dental/inspection mirror) can be placed
by the FID outlet. If lit, water vapor from the outlet will condense on the surface. If no
condensation is observed, the FID has likely been extinguished. Some auto-GC systems
automatically ignite the FID when power is supplied; however, operators may still need to take
action to correct problems which resulted in failure of the FID to light or the FID to be
extinguished. Such problems can include:
• excess moisture in the FID (can be removed by allowing dry zero air to flush the FID for
several minutes before attempting to ignite the FID)
• incorrect mixing ratio of hydrogen to air (which may include lack of hydrogen)
• leaks or tripped leak alarms on hydrogen generators and power failures resulting in the
safety shutoff of hydrogen
• significant column leak or flow disruption, and leaks at the Deans switch resulting from
poorly cut column ends or improperly installed columns
4.2.4 Compound Identification
Identification of compounds by auto-GC with FID requires maintaining consistent
chromatography. Target compounds are identified based on their retention time and the
associated chromatographic peaks must meet a minimum S:N ratio to be positively identified.
4.2.4.1 Compound Retention Time
Assignments of compound identity are established based on analysis of an RTS which contains
the compounds of interest. Operators assign retention time (RT) windows to each target analyte
peak in the auto-GC chromatography data systems (CDS) software. For analysis of ambient or
unknown samples, the software identifies target compounds by "looking for" a chromatographic
peak in the RT window. The target analyte peak is to show a S:N >3:1, preferably >5:1. Details
regarding calculation of S:N are provided in Section 4.2.4.2. CDS algorithms generally
51
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
distinguish chromatographic peaks from baseline based on pre-programmed parameters to
discern changes in baseline's slope, inflection, or rates of change. These settings for peak
identification also typically take into account the area of the peak, which can be set by the user to
allow the CDS to ignore "peaks" with area response less than a specified value. This allows the
user to avoid identification of instrumental noise as chromatographic peaks. Parameters and
their adjustability differ widely among CDS manufacturers with some allowing a large degree of
user adjustment to many variables and others only permitting adjustment of a few variables.
Numerous variables impact the target compound RT. These variables include purge times,
column pressures and flows, and preconcentrator and GC temperatures and timing, and are to be
set identically on the auto-GC whether analyzing a calibration standard, retention time standard,
blank, or ambient air. Even with the improvements in auto-GC technology since the beginning of
the PAMS program, compound RTs can still shift substantially enough to fall outside assigned
retention windows, leading to false negatives (missed identifications) or incorrectly identifying a
different target compound or an unknown compound within the assigned window. RT shifts can
occur due to a number of factors which include, but are not limited to:
Insufficient moisture removal from the sampled air stream or from carrier gas. This
typically impacts the C2 to C(, hydrocarbons and generally acetylene will be the most
impacted of the light hydrocarbons.
o Increases in moisture in the sampled air stream or carrier gas may be due to
several factors, among which include:
¦ Failure of the zero air generator to deliver sufficiently dry gas (e.g.,
dewpoint < -100°C) for dehydration of the sampled gas in the Nafion™
dryer
¦ Failure of carrier gas dryers to remove moisture (depleted desiccants, etc.)
¦ Rain events resulting in atypically moist ambient air from which the
Nafion™ dryer cannot sufficiently remove water
¦ Degradation of the Nafion™ dryer's ability to sufficiently dehydrate the
sampled gas stream
¦ Insufficient purge of Peltier cooler drying systems between samples
System leaks. Such leads to poor chromatography and shifting RTs
o Cracked, broken, or incorrectly installed preconcentration trap
o Poorly seated or incorrectly installed transfer line or column
o Poorly seated connections to the Deans switch or other internal flow pathways
Preconcentrator trap failure
Some CDSs permit the instrument operator to assign reference compounds that the CDS uses as
anchors in a chromatogram to more reliably identify other target analytes by comparison of
retention times. These reference compounds will ideally be those that:
• are always present in the chromatogram (system blanks being an exception)
• have a relatively stable RT
• exhibit sharp peak shape and are chromatographically well-resolved from other target
peaks
52
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
• elute in an area of the chromatogram that is relatively unaffected by typical
chromatographic interferences (moisture, unknown VOCs).
The software systems automatically adjust RT windows based on observed shifts in the RTs of
the reference compound(s). If the CDS permits, it is recommended that the instrument operator
assign minimally one reference compound on each of the two FID channels. Pentane and toluene
typically meet the above criteria for the light hydrocarbon (HC) and heavy HC channels,
respectively. More information on reference peaks can be found in PAMSGRAM Volume 14
from January 1999 available at the following link on AMTIC:
https://www3.epa. gov/ttn/amtic/files/ambient/pams/eram 14.pdf
4.2.4.2 Signal-to-Noise Ratio
Chromatographic peaks will minimally show a S:N that is > 3:1, and preferably >5:1.
Determination of the S:N is somewhat subjective based on the individual analyst and his/her
characterization of the noise and analyte peak. Some chromatography systems include S:N
functions that require the analyst to assign the noise and target peak. For well-resolved, sharp
peaks, the S:N will exceed 5:1, and does not need to be measured. For peaks with low S:N that
are questionable as to whether they meet the 3:1 S:N criterion, the criterion is a guideline; it is
unnecessary to measure each peak, but rather the experienced analyst's opinion should weigh
heavily on whether the peak meets the S:N criterion.
Refer to Figure 4-1 for the following example for determining the S:N. To determine S:N, the
characteristic height of the noise of the baseline (A) just before the peak and the height of the
analyte peak (B) are measured. The ratio of the analyte peak height (B) is divided by the noise
height (A) to calculate the S:N. In the example below, the peak at 17.0 minutes is discernable
from the noise, but is not well differentiated from the noise and is very close to a S:N of 3. In the
example, the peak heights of the noise and analyte peak are approximately 700 units and 1700
units, respectively, for a S:N of 2.4.
If the 3:1 S:N criterion is not met, the compound should not be positively identified. The only
exception to this is when, in the opinion of an experienced analyst, the compound is positively
identified despite not having a S:N < 3. Such instances should not occur typically and the
rationale for such an exception should be documented.
53
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
TIC: D1231503.DVi3ta.ms
I 1 1 1 1 I 1 1 1 1 I 1
5.5C 16.OD 16.50
17. DO 17.50
Figure 4-1. Determination of Chromatographic Peak Signal-to-Noise Ratio
4.2.5 Auto-GCData File Naming
Auto-GCs sampling hourly for 59 compounds will generate 1400 individual concentration data
points in 24 or 48 raw data results files (where the latter depends on if a separate file is generated
for each of the two chromatographic columns). The CDS will also generate files containing
chromatograms for each sample hour for each FID, totaling approximately 100 data files each
day. The monitoring agency should carefully plan a convention for file naming and file
organization that permits ready identification of files by the file name. Such a convention should
permit the user, from the file name without needing to open the file, to discern the FID channel,
date, sample hour, sample type, and whether the data file is original or has been reprocessed.
Note that some CDSs may limit the number of characters permitted in file names or may force
the user to utilize a pre-determined file naming convention. An example convention using 11
characters follows in Table 4-2.
For an ambient sample collected for the 6:00 p.m. hour on July 31, 2020, on the light
hydrocarbon channel that had not been processed, the file name would be:
L20073118AU.dat
Filename Format = @12345678&#.dat
54
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
Table 4-2. Example Auto-GC File Naming Convention
Position
Detail
Character or Character Combination
@
Hydrocarbon channel
L = light HC channel
H = heavy HC channel
1 & 2
2-digit year (YY)
2-digit numeric
3 & 4
2-digit month (MM)
00 through 12
5 & 6
2-digit day (DD)
00 through 31
7 & 8
2-digit hour from 24-
hour clock
00 through 23
&
Sample type
A = ambient sample
B = blank sample
C = continuing calibration verification (CCV)
I = initial calibration standard
P = precision check standard
S = second source calibration verification
(SSCV)
X = exploratory or experimental
(troubleshooting, conditioning, etc.)
#
Processing status
U = unprocessed
R = reprocessed
4.3 Method Detection Limits for Auto-GC
Annually prior to each PAMS sampling season, PAMS Required Site monitoring agencies
should determine the MDL of each priority compound and optional compound.
MDLs are needed to provide method sensitivity information to the data user. The EPA auto-GC
studies conducted to determine instruments suitable for PAMS Required Site deployment
demonstrated that an MDL of 0.5 ppbC or less could be achieved for most target compounds.4'5
Discussions with auto-GC operators have indicated that the selected auto-GC instruments are
capable of detecting lower concentrations, in some cases as low as 0.1 ppbC.
Briefly, MDLs are determined per the MUR of the MDL process defined in 40 CFR Part 136
Appendix B,20 which prescribes that a minimum of seven low-level standards and seven blanks
be prepared in the matrix and analyzed. The average concentration and the standard deviation of
the standard analyses and blank analyses are calculated separately and used to generate the MDL
values. The instrument should be calibrated and pass all relevant quality control criteria before
analysis of MDL samples.
When conducting the MDL study, auto-GC operators are encouraged to ensure that the
instrument and support equipment have been operating and stable for several weeks. This
includes ensuring that the preconcentrator trap has been conditioned to ensure quantitative
adsorption and desorption and to minimize and preferably eliminate interferences, and that the
humidified matrix blank background levels are sufficiently low. This warm-up and conditioning
period further ensures that support equipment such as zero air generators, hydrogen generators,
55
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
and gas purifiers are functioning properly and that gas delivery pathways have been flushed and
properly conditioned. Once the instrument conditioning and stabilization are complete,
calibration can be established for the two FIDs. Matrix blanks and spikes for determining the
MDL are analyzed over the course of three or more different non-consecutive days. Distributing
the blank and standard analyses over time incorporates a component of temporal variability in
the MDL determination and increases the representativeness of the auto-GCs routine
performance. While the conduct of the MDL itself is different, the example calculation shown in
Section 5.6.2 for determining carbonyls MDLs is analogous to determination of MDLs for
speciated VOCs by auto-GC.
4.3.1 MDL Blank Component, MDLb
For the blank component of the MDL, operators should include in the calculation those
humidified blanks analyzed with the instrument after completing shakedown and achieving
stabilization. While operators should make reasonable efforts to ensure the instrument is clean
and target analyte responses are as low as possible, it is normal to see some background of target
analytes in chromatograms. Blanks that show typical concentrations of target compounds should
not be excluded from the MDL calculation; however, if the concentrations of these compounds
appear to be decreasing over time and continue to decrease, the system may not be completely
stable with respect to contamination levels, and may require a longer stabilization period before
generating blank data for use in the MDL determination. Inclusion of only typical blank
background levels is important to determine a realistic MDL. Each auto-GC performs uniquely,
and levels of background for one instrument may be different for another separate but identical
auto-GC. For contamination in blanks which is excessive (-0.3 ppbC or above, analyte
dependent), operators should take corrective actions to eliminate the source and root cause of the
contamination where possible before analyzing blanks to include in the MDL calculation. It is
acceptable to exclude blanks from the MDL calculation for technical reasons, such as blanks that
are known to be problematic due to instrument issues such as uncharacteristic contamination or
memory effects (e.g., blanks analyzed following a high concentration calibration standard). A
minimum of seven humidified blank analyses is needed to calculate the MDLb; however,
operators should include as much typical blank data as possible to provide a realistic average
background concentration (and its associated variability) to generate the MDLb.
4.3.2 MDL Standard Spike Component, MDLsp
To determine the standard spike component of the MDL, matrix standard concentrations of the
target analytes are chosen to be analyzed minimally seven times. If too low of a spiking level is
chosen, the analyte may not be reliably detected. If too high of a spiking level is chosen, the
variability of the method near the actual limits of detection may not be properly characterized.
An appropriate spiking level may be selected by considering the following (in order of
importance):
1. The concentration at which the instrument S:N is three- to five-fold for the analyte.
Auto-GC operators in November 2017 indicated that diluting the stock RTS (15 to 60
ppbC) to 100-fold (0.15 to 0.6ppbC) and 200-fold (0.075 and 0.3 ppbC),
56
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
respectively, provided analyte responses at approximately a S:N of 5:1 for the heavy
and light HCs, respectively.
2. The concentration at which qualitative identification criteria for the analyte are lost
(note that this will be a S:N below approximately 5:1 and will be approximately the
concentration determined from the MDL process absent of blank contamination).
3. Analysis of a suite of matrix blank samples; calculate the standard deviation of the
measured concentration and multiply by 3.
4. Previously acceptable MDL studies and related experience.
Note that the MDL spiking level should not be within the calibration curve (which will typically
span approximately 1 to 25 ppbC); rather, the MDL spiking level should be less than the lowest
calibration standard in order to determine the most realistic MDL. Concentrations within the
calibration curve are required to meet method precision and bias acceptance criteria and are of a
sufficient concentration that qualitative identification is certain, which practical experience has
shown to be approximately three- to five-fold the expected MDL concentration.
MDL standard spikes are typically analyzed from the RTS (or a similar standard) which contains
most of the target analytes. Concentrations of target analytes in the RTS typically range from
tens of ppbC to approximately 50 ppbC, therefore, as described above, the RTS may require two
or more different dilution ratios to achieve a concentration of 0.5 ppbC for each target analyte.
Alternatively, a standard gas prepared with all target compounds at the same concentration in
ppbC (e.g., 100 ppbC) can be readily diluted to approximately 0.5 ppbC for determining MDLs.
Preparation of the chosen concentrations and introduction of the standards to the instrument may
be performed by one of several conventions, depending on the equipment available at the site. In
all cases, the standards analyzed to determine the MDL will need to be humidified to ensure
proper performance of the higher molecular weight (C9 and C10) target compounds. Analysis of
dry gas standards will result in MDLs that are not representative of those expected to be achieved
in ambient air (which always contains some amount of moisture during the PAMS season) and
may result in depressed target compound response for some analytes. To generate representative
MDL standards, they should be prepared in a similar convention to the instrument calibration
standards, which may include:
1. Preparation of a standard canister at the desired concentration's): The RTS is diluted to
the desired concentration with humidified zero air into a clean evacuated stainless steel
canister. Dilution may be performed by static dilution (refer to Section 4.5.2.1) or
dynamic dilution (refer to Section 4.5.2.2).
Dilution into a canister has several drawbacks as the inclusion of the canister in the MDL
process can impart bias (positive or negative) depending on the target compound under
evaluation and the condition or cleanliness of the canister. Recovering target
hydrocarbon analytes from canisters can be problematic, particularly C9 and C10
compounds with higher boiling points. Standards prepared in canisters should be properly
humidified (~ 40 to 50 % RH) to ensure recovery of higher molecular weight compounds
from the canister. Humidification is discussed further in Section 4.5.2.4.
57
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
Monitoring agencies should note that the absolute accuracy of the standard gas dilution
for MDL determination is not critical, as the MDL procedure is designed to characterize
the variability of replicate analysis at approximately the MDL concentration. Prior to the
use of canisters, users are encouraged to verify that the canister(s) impart acceptably low
bias by performing a bias check on the canister (further information on performing bias
checks on canisters can be found in Revision 3 of the NATTS TAD, available at the
following link on AMTIC:
https://www3.epa.gov/ttnamtil/files/ambient/airtox/ S%20TAD%20Revision%203
i T \l %20Qctober%202i«s a>
Note that the determination of canister bias is not critical if all MDL standards are taken
from the same canister. However, if a number of canisters is employed, use of
uncharacterized canisters may impart additional variability to the MDL process and result
in artificially elevated determined MDLs.
Active sites within canisters can result in the low recovery of labile analytes such as
acetylene, olefins, and terpenes. An additional drawback to preparing standards in
canisters is that there is a limited number of samples that can be drawn from the canister
before the canister pressure is insufficient for the auto-GC to remove a sample aliquot. In
such instances, another standard canister will need to be prepared. In general, use of
canisters for preparing standards requires a support laboratory with the capability to
perform heated canister cleaning and evacuation.
Note: Unlike canister collection methods, preparation of MDLs for auto-GCs does not
require that a minimum of seven separate canisters be prepared. The requirement for
canister collection methods to include a minimum of seven separate canister samples
does not apply to auto-GCs as the measurement method does not utilize a canister for
sample collection and therefore does not require characterization of the variability
inherent in the canister fleet.
2. Pulsed standard delivery: The RTS is connected directly to the instrument inlet and
"pulsed" to deliver the standard for a known portion of the 40-minute sampling duration.
Instrument operators should note that reproducible delivery is more difficult to achieve
when employing short pulse times (e.g., less than 4 minutes) due to the increased
variability in MFC performance during startup and shutdown, and that MDLs determined
in this manner may be biased high as a result. The remaining duration of the 40-minute
sampling period should consist of sampling humidified zero air to best replicate the
humidity exposure to the system during routine analyses. Refer to Section 4.5.2.3 for
further information and an example calculation for pulsed standard delivery.
3. Delivery of a dynamic dilution gas stream to the instrument: The RTS is diluted with
humidified zero air to achieve the desired concentration and this diluted gas is delivered
to the auto-GC for analysis as described in Section 4.5.2.2. The auto-GC samples the gas
stream for the entire 40-minute sampling period as is done for calibration standards,
ambient samples, and blanks. This convention replicates the manner in which ambient air
58
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
is introduced to the auto-GC and is likely to result in the most representative and accurate
MDLs. The main drawbacks to this convention are that it requires additional equipment
(a dynamic dilution gas calibrator system) and is less conservative of standard gas
consumption than either delivery of MDL standard aliquots by canister or by pulsed
standard delivery.
The EPA anticipates development of a subscription service through which the RTS will be
available at an approximate 10-fold dilution (approximately 1.5 to 6 ppbC per target compound)
with which MDL standards may be prepared. Instrument operators will analyze the standard in a
pulsed fashion (as described in paragraph 2 above) to provide an effective dilution to the auto-
GC and complete the sampling with humidified zero air.
4.3.3 Redetermination of MDLs
MDLs are to be determined annually prior to the beginning of each PAMS season and when
there are changes to the auto-GC that would reasonably result in changes in its sensitivity. For
MDL determination prior to PAMS season, it is recommended that MDLs be determined
following the typical annual instrument maintenance which should include replacing the
preconcentrator trap and Nafion™ dryer (if so equipped). As discussed in Section 4.2.3.3, the
auto-GC should be conditioned for several weeks to ensure that the instrument is sufficiently
clean and that its performance is stable. Once sufficiently conditioned, FID response is typically
stable over time, and a decrease in propane, butane, or benzene response factor that exceeds 10%
from the initial calibration indicates a decrease in sensitivity that may require preconcentrator
trap replacement or other system maintenance. Such a decrease in response would typically
occur over a longer period than the three-month PAMS season, even including the several weeks
of conditioning prior to beginning monitoring for the year. If, after such trap replacement or
maintenance, the response factor returns to within 5% of that level determined in the initial
calibration (by analysis of a standard with a concentration in the lower third of the calibration
curve), the MDL would not require redetermination. If the response factor remains depressed as
compared to the initial calibration, the MDL should be determined anew following recalibration
of the instrument.
4.4 Auto-GC Interferences
Approved auto-GCs for use at PAMS Required Sites range from instruments specifically
designed for field use to laboratory instruments configured to operate in a monitoring shelter. For
any of the instrument systems, they can be subject to interferences which impact measurement
quality. The most common interferences result from the presence of ozone and/or moisture and
fluctuations in shelter temperature.
4.4.1 Ozone Interference
During sample preconcentration, co-collected ozone may react with target analytes (particularly
unsaturated hydrocarbons) within the air stream, with target analytes trapped on the sorbent beds,
and with the sorbents themselves. Such reactions result in the formation of oxidized organic
byproducts that may appear as unknown peaks in the chromatogram. To date, ozone scrubbers
are prescribed for carbonyl sampling (as discussed in Section 5.4) but have not been formally
59
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
evaluated with the approved auto-GC systems and have not been widely adopted for removing
ozone interferences from the analysis of VOCs by auto-GC. Use of an ozone denuder on auto-
GC inlets is not recommended unless the monitoring agency has performed an appropriate
collocation study to evaluate effects of the ozone denuder on the ambient measurements. Such a
study would require passing calibration standards and QC samples through the ozone denuder.
Further detail of such a collocation study is outside the scope of this TAD.
4.4.2 Moisture
As discussed in Section 4.2.3.2, water vapor can adversely impact auto-GC performance.
Sampled gas streams that are insufficiently dehydrated prior to reaching the preconcentrator traps
can result in trap icing and/or poor chromatography and RT shifts of target analytes. Moisture
from insufficiently dried carrier gases can have similar effects on chromatography and RTs. Such
changes in chromatography can persist in the light hydrocarbon column for several consecutive
runs. Moisture remaining on preconcentrator traps when the thermal desorption step begins can
degrade sorbents and water introduced onto the separation columns can damage stationary phase
linings. Preconcentration systems typically perform a purge of the preconcentrator trap with dry
carrier gas at the end of the sample collection prior to trap heating for desorption. The flow rate
and volume of this dry purge is selected to provide sufficient dehydration while minimizing the
loss of target analytes retained on the trap.
4.4.3 Temperature
Auto-GCs are sensitive to temperature fluctuations within the monitoring shelter. Monitoring
shelters for auto-GCs require heating, ventilation, and air conditioning (HVAC) systems to
maintain environmental conditions suitable for instrument operation. During the June 1 through
August 31 PAMS season, shelters will typically require air conditioning to maintain shelter
temperatures below ambient. Auto-GC moisture management systems employing Nafion™
dryers are more effective at lower temperatures. Temperature fluctuations in the monitoring
shelter can cause insufficient drying of the ambient air and/or humidified check standard gases
and may result in excess moisture passing through to the preconcentrator trap and/or GC
column(s). As PAMS Required Sites are typically located at NCore sites, PAMS instruments
may be installed with criteria gas monitoring instruments in shelters that are required to be
maintained at 20 to 30°C with a standard deviation of < 2% over 24 hours. These conditions are
sufficient for auto-GC installation; however, existing HVAC systems may be of a capacity
insufficient to maintain conditions with the additional heat burden from the PAMS instruments.
In such cases where criteria pollutant monitors and PAMS instruments can be installed in the
same shelter, additional cooling capacity may be required or partitions can be installed either to
maintain environmental conditions for the criteria pollutant monitors or to stabilize the
environmental conditions for the auto-GC. The auto-GC, its support equipment, and its exhausts
should be placed sufficiently away from HVAC thermostats to ensure the latter do not artificially
influence HVAC operation. If auto-GCs are installed in separate shelters from the criteria
pollutant monitors, recommendations in the following paragraphs should be considered.
Efficient chromatographic separation is accomplished by ramping the temperature of the GC
column from approximately 30°C to 200°C. Following the completion of the temperature ramp,
60
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
the oven temperature is reduced through venting to achieve the initial temperature for analysis of
the next sample. Achieving reproducible and predictable compound RTs requires precise control
of the oven temperature; however, shelter temperature fluctuations can result in changes in the
GC oven temperature profile during analysis with consequent changes in observed RTs. It may
be difficult for auto-GCs in monitoring shelters with insufficient cooling capacity to return to the
starting oven temperature or for the Peltier cooler in the thermal desorber or dryer to reach its
initial temperature setpoint. In such cases, the thermal desorber may not be ready to collect the
sample at the proper time or the GC portion of the instrument may not be ready to begin
separation of the collected sample when the thermal desorber is programmed to introduce the
sample to the GC, ultimately causing a cascade of delays within the instrument and preventing
sample analysis per the intended hourly schedule. Additionally, auto-GCs in shelters that
experience higher temperatures may experience shortened Peltier cooler life due to the additional
cooling burden required to reach its initial temperature setpoint.
It is recommended that monitoring agencies consult with an HVAC professional to ensure the
monitoring shelter's cooling capacity is sufficient. Several of the auto-GC instruments
incorporate connections on the GC oven vent to permit routing oven heat exhaust to the exterior
of the shelter, reducing cooling demand for the HVAC system. Additionally, locating
compressors for zero air generation systems outside the monitoring shelter (protected from the
elements) will reduce both heat burden and noise levels inside the shelter. Installation of
compressors outdoors will require more frequent maintenance due to increased humidity in the
source air and the build-up of water in the compressor ballast tank(s). To reduce temperature
fluctuations at the instrument to the extent possible, avoid direct impingement of conditioned air
onto the auto-GC. This can be accomplished by installing baffles to diffuse flows of conditioned
air and/or redistributing HVAC outlets to reduce temperature gradients within the monitoring
shelter.
4.4.4 Source-Oriented Interferences
Monitoring sites that are impacted by industrial sources, such as refineries, refueling stations, or
other similar sources, may contribute unknown hydrocarbon artifacts to the GC chromatogram,
particularly on the heavy hydrocarbon (PDMS column) channel. These additional hydrocarbons
may produce interfering peaks in chromatograms that coelute with target compounds and
complicate the integration and identification of target analytes. Where chromatographic
resolution is inadequate, the amount of area integrated for the target peak will typically be
defined by a vertical to the baseline and the resulting area will be less than that for a completely
resolved peak. In such instances, the instrument operator should indicate the reported
concentration is biased low and qualify the concentration as "LL" when reporting it to AQS. In
some cases coeluting peaks will be larger than the target analyte peak, especially when the target
peak is a small shoulder on the coeluting compound. In such cases, the instrument operator and
technical reviewers should evaluate whether the value should be estimated or invalidated.
61
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
4.4.5 Problematic Compounds for Auto-GC
Target compounds that are particularly problematic with respect to recovery are acetylene
(C2H2), styrene (CxHx), and alpha- and beta-pinene (C10H16).
Acetylene typically shows lower recovery (70% or less) when compared to the other light
hydrocarbons when analyzed as part of a CCV. While the low recovery problem is well-
documented for acetylene21 the reasons for this lower recovery appear to be due to a combination
of instability of acetylene within the high-pressure stock standard cylinders and the inability to
quantitatively trap and desorb acetylene.
Styrene is suspected to be unstable in the high-pressure stock standard cylinder; however,
experience has shown that trapping and desorbing styrene is not particularly problematic. Lower
recoveries of styrene have been reported when establishing calibrations based on the carbon
response of benzene and when analyzing CCVs. These lower recoveries are typically consistent
with concentrations relative to other species in the standard mixture, even over the course of
several months to a year, indicating that the concentration of styrene has decreased then
stabilized within the standard cylinder.
For alpha- and beta-pinene, the beta isomer is suspected to be unstable in the high-pressure stock
standard cylinder, with the latter typically isomerizing to form camphene, d-limonene and/or
alpha-pinene. While alpha-pinene appears to be more stable than the beta isomer, the former
may also isomerize or decompose. Losses of both pinenes will manifest as changes in their
response relative to the other target compounds in the standard cylinder, particularly the
compounds propane and benzene employed to establish the FID carbon responses. Monitoring
agencies should be aware that acetylene, styrene, alpha-pinene, and beta-pinene may not meet
the established bias criteria (±30% difference from the theoretical concentration) in routine QC
samples even on recently calibrated and stabilized auto-GC instruments with new
preconcentrator traps. In such instances, ambient concentration data for these monoterpenes
should be appropriately qualified as an estimate (refer to Table 11-2). More information on the
stability of monoterpenes in cylinders may be found elsewhere.22
Monitoring agencies should also note that three compounds listed in Table 4-1 (optional
compounds) are not included in the RTS typically sourced by EPA: carbon tetrachloride
(halogenated), tetrachloroethylene (halogenated), and ethanol (alcohol). Each of these
compounds responds differently (less intensely) to an FID than hydrocarbons and each will
likely not demonstrate expected accuracy (±30% of nominal) using a carbon-based response
factor. In addition to the decreased FID response relative to hydrocarbons, monitoring agencies
operating auto-GCs with Nafion™ sample stream drying systems will be unable to accurately
determine ethanol concentration as the fluoroelastomer polymer membrane removes a fraction of
ethanol from the sampled air stream.
62
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
4.5 Calibration of Auto-GCs
4.5.1 Standard Materials
Stock standards for calibration of the auto-GC are commercially available. Standard gases will
preferably be National Institute of Standards and Technology (NIST)-certified or NIST-traceable
certified and are to be accompanied by a certificate of analysis stating the certified concentration
and associated uncertainty for each component. Expiration dates are typically one year or more
and several vendors offer a recertification service that verifies the component concentrations and
extends the useful life of the standard cylinder beyond the original expiration date.
Recertification of standard gases is often more cost effective than purchasing new standards and
can be performed during the non-sampling season (as applicable to the specific PAMS Required
Site). Note that manufacturers may exclude the compounds note in Section 4.4.5 from their
guarantee of stability.
4.5.1.1 Primary Calibration Standard
The standard chosen for calibration, the primary calibration source, is to minimally be NIST-
traceably certified for propane or butane and benzene. The primary calibration source will
typically be a high-pressure cylinder minimally containing two compounds, propane and
benzene, but if the auto-GC permits, the source may be from NIST-traceably certified
permeation tubes of target compounds (one for each FID channel), such as butane and benzene.
It may be useful for troubleshooting purposes to include other target compounds in the primary
calibration stock gas, such as a suite of compounds representing the C2 to C10 range, the priority
compounds, or other combination of desired target compounds. Several gas vendors offer an 18-
component blend, mimicking the previously available NIST 1800b standard, and containing the
following analytes:
ethane
n-hexane
propane
n-heptane
propene
benzene
iso-butane
iso-octane
n-butane
n-octane
iso-butene
toluene
iso-pentane
nonane
n-pentane
o-xylene
1-pentene
decane
Selection of standard concentrations depends on the capability of the operator to dilute the
standard with humidified zero air to generate concentrations in the desired range calibration
range of 1 to 25 ppbC.
4.5.1.2 Secondary Source Calibration Verification Standard
The calibration established by analysis of the primary calibration standard will be verified by
analysis of second source calibration verification standard (SSCV). The SSCV stock standard
will typically be the PAMS RTS and will contain all of the priority and many of the optional
compounds {Note: The current PAMS RTS contains 59 compounds and includes all priority
63
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
speciated VOCs and most of the optional speciated VOCs with the exceptions of carbon
tetrachloride, tetrachloroethylene, and ethanol; Refer to Table 4-1. The RTS is prepared to
contain the target analytes at concentrations covering approximately 15 to 60ppbC.) The SSCV
stock will be sourced from a different provider than the primary calibration standard or will
minimally be from a different lot of source material from the same provider, if unavailable from
an independent supplier. The SSCV stock standard is to minimally be NIST-traceably certified
for propane or butane and benzene to verify the calibration established with the primary
standard. The recoveries of the other target VOCs will be evaluated by analysis of the SSCV.
The SSCV is prepared from this standard at a concentration within the lower third of the
calibration curve (approximately 5 ppbC for propane or butane and benzene). Acceptance criteria
are listed in Table 4-4.
4.5.1.3 Retention Time Standard
In order to establish RT windows for each of the target analytes, a known standard is to be
analyzed on the auto-GC to determine the RT of the individual target compounds. The RTS
described in Section 4.5.1.2 serves this purpose. Two compounds included in the RTS blend but
not listed among the target analytes in Table 4-1 are 1-hexene and dodecane, which are used to
set Deans switch timing and end the GC run, respectively. EPA supplies the RTS to PAMS sites
and has provided for the RTS to be NIST-traceable certified for propane and benzene. As such,
when certified for these two compounds, the RTS may also serve as the SSCV (see Section
4.5.1.2). If not employed as the CCV or SSCV, there are no recovery criteria associated with the
RTS since its purpose is to verify and adjust target analyte RTs, and not to evaluate the
calibration response.
4.5.1.4 Zero Air
Zero air is not a standard gas; however, it is critical for the proper performance of the auto-GC,
specifically to demonstrate lack of interferences in system blanks and to prepare non-biased and
reproducible dilutions of stock standards. Zero air is also typically employed as the oxidizer for
the hydrogen fuel in FIDs, as the purge gas for dehydrating sample streams with Nafion™
dryers, and/or as the purge gas to prevent moisture from freezing on the Peltier coolers in
electronic sample dryers and preconcentrators. The zero air should be provided by an online
zero air generator (ZAG) that can generate approximately 5 L/minute flow of nominally HC-free
air (total hydrocarbon concentration of < 10 ppbC, and preferably less). When employed as dry
gas for sample stream dehydration or Peltier purging, zero air should have a dewpoint of
< -100°C. Zero air employed for diluting standard gases and/or as zero air blanks should be
humidified to approximately 40-50% RH. For zero air generation systems providing both gas for
dry purging and for the QC sample matrix (standards dilution or blank), the humidification for
QC samples should occur downstream from the dry gas split.
Dry nitrogen, such as available in high-pressure cylinders, may be substituted for zero air
employed as the dry purge gas. Such a substitution requires the monitoring agency to replace
nitrogen cylinders routinely and does not eliminate the need for zero air for FIDs and for use as a
QC sample matrix. For sites employing zero air as a purge gas, dry nitrogen or dry zero air in a
high-pressure cylinder may be useful for troubleshooting a ZAG's water removal capability.
64
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
ZAGs providing zero air for blank analysis and as a diluent should be evaluated to ensure the
generated gas is sufficiently clean. ZAGs should be set up, conditioned, and operated per the
manufacturer's instructions prior to connection to the auto-GC sample inlet or to a dilution
system. This permits flushing of any contaminants in the ZAG system that could otherwise
contaminate the instrument(s). Once the proper conditioning has been completed, an aliquot of
the zero air should be analyzed and evaluated to ensure the concentrations of target analytes and
interferences are sufficiently low (minimally <0.5 ppbC for each compound to be reported to
AQS, and preferably very little chromatographic peak response on either channel - total
hydrocarbons <10 ppbC). In general, a chromatogram of zero air should show little or no
response for target analytes. Where concentrations of reportable compounds exceed >0.5 ppbC,
the ZAG should be disconnected from sampling inlets and dilution systems and permitted longer
periods of flushing, or subjected to maintenance prior to redeployment.
4.5.2 Retention Time Establishment and Calibration Convention and Procedure
Prior to attempting to calibrate the auto-GC prior to the beginning of PAMS season, the operator
should first verify that instrument maintenance has been performed to replace the
preconcentrator trap and the Nafion™ dryer (if so equipped) and to verify the instrument is
functioning properly (e.g., no error messages, the instrument progresses through a sequence
normally, etc.). It is also recommended that the operator assess the need to replace carrier gas
scrubbers and ZAG consumables as this ensures optimum auto-GC performance and reduces the
likelihood that future troubleshooting may be required.
The instrument should be powered on and analyzing ambient air or humidified zero air for
minimally two weeks (preferably longer) to ensure the instrument is leak-free and operating
properly and that the support equipment (compressors, zero air generator(s), hydrogen generator,
etc.) is functioning properly. During this time, it is suggested that the trap conditioning described
in Section 4.2.3.3 be conducted. For monitoring agencies operating year-round that cannot afford
several weeks of instrument down time, the conditioning period may be shortened, as needed, so
long as ongoing demonstration of instrument performance is attained.
Once it is established that the instrument is operating properly, the operator should analyze a
series of humidified blanks to ensure there are no chromatographic artifacts and that the
instrument is sufficiently clean. If the preconcentrator trap has recently been replaced, levels of
contaminant target compounds in the humidified zero air blank may still be desorbing from the
trap, and the conditioning should continue until target compound concentrations are acceptably
low. The instrument is sufficiently clean when there are few and very small chromatographic
peaks in the blank chromatograms and the target compound responses in the humidified zero air
blanks are stable and do not exceed 0.5 ppbC. Following successful demonstration that the
instrument is appropriately clean, the operator should analyze the humidified RTS a minimum of
five times successively to establish and assign retention windows for the target analytes. The
mean, standard deviation, and percent relative standard deviation (RSD) of the RTs for each
compound should be calculated for the five aliquots. If the percent RSD of the mean RT exceeds
5% for any compound, the RT may not stable for that compound and/or the peak identification
may not be correct. If peak identifications are correct, the instrument operator should analyze
additional RTS aliquots until the RTs stabilize (RT RSDs < 5%).
65
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
RSD is calculated by dividing the standard deviation of the population of measurements
population by the arithmetic mean of the population, expressed as a percentage:
%RSD = ¦ 100%
x
where: o = standard deviation of population
x = population arithmetic mean
Once the RT windows have been established and are stable for the target analytes, a series of
humidified zero air blanks is again analyzed and the operator again verifies the instrument is
acceptably free of contaminants and carryover. Once the auto-GC is demonstrated to be
acceptably clean, the operator can calibrate the instrument.
The instrument calibration is to include a minimum of three standard levels at approximately 1,
5, and 25 ppbC for propane or butane and benzene. The propane or butane responses will be used
to establish the carbon-based response calibration for the light HC "PLOT" FID channel and the
benzene responses will be used to establish the carbon-based response for the heavy HC
"PDMS" FID channel.
As discussed above in Section 4.5.1.1, the primary calibration source gas(es) are to be NIST-
traceably certified for minimally propane or butane and benzene. If the primary calibration gases
are sourced from high pressure cylinders, the standards may be introduced by the following
conventions (as described in Section 4.3.2):
• dilution of stock standard gas into a canister or canisters for introduction to the auto-GC
{Note: canisters used for this purpose will need to be appropriately qualified and cleaned
for this use.)
• dynamically diluting a stock gas using differential flow control to actively deliver a
known concentration gas to the auto-GC
• pulsing delivery of a stock standard gas to introduce known masses of standard
(equivalent to that contained in the typically-collected volume of gas at the three
challenge concentrations) to the preconcentrator
Auto-GC systems employing permeation tubes for generating calibration standard levels will
establish calibration by delivering known masses to the preconcentrator based on the dilution of
the gas evolved from the permeation tube, where the masses are determined as with the pulsed
delivery method. Permeation tubes are maintained in a temperature-controlled oven and emit a
known mass of the target compound at a known rate at a given constant temperature.
4.5.2.1 Static Dilution
Static dilution of a gas standard can be performed by adding known amounts of a standard gas
and diluent gas to a fixed volume vessel (such as a canister) and measuring their partial pressures
with a calibrated pressure gauge or transducer. The technician should establish a desired final
66
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
pressure and desired final concentration of the target compounds as well as the corresponding
volumes of standard and diluent gas prior to beginning the dilution process.
Starting with an evacuated canister (initial vacuum of < 0.5 psia), the pressure is measured with a
calibrated pressure gauge or transducer and recorded. Next a standard gas is connected to the
canister and the gas slowly added to the canister to the desired pressure and the pressure
permitted to stabilize. This pressure is measured and recorded with a calibrated gauge. Finally,
humidified diluent gas is added to the canister to the desired final pressure to complete the
dilution. The following is an example for static standard preparation of a 25-ppbC propane and
benzene:
Example Static Dilution:
Primary Stock Gas: 507 ppbC propane and 495 ppbC benzene
Desired Working Standard Concentration: 25 ppbC
Fixed Vessel: 6-L stainless steel canister
Final Desired Absolute Pressure: 30 psia
1. Calculate the effective dilution factor (DF) needed by dividing the stock gas
concentration by the desired working standard concentration (these concentrations
can be approximate to determine the DF):
500 ppbC/25 ppbC = 20
2. For a 20-fold dilution and a final pressure of 30 psia, it is assumed the canister has
been evacuated to hard vacuum of 0 psia and the required partial pressures of the
stock gas can be calculated by dividing the final desired absolute pressure by the DF:
30 psia/20 =1.5 psia
3. The evacuated canister pressure is measured to be 0.50 psia. Note that to minimize
contamination from previous canister contents, the evacuated canister pressure will
ideally be <0.1 psia (> 29.7 in Hg vacuum). If the intended diluent gas is not
humidified, the canister can be humidified at this time. Refer to Section 4.5.2.4 for
further information on humidification.
4. The stock calibration gas is connected via a regulator to the canister. The stock gas is
then bled into the canister until 1.5 psia has been added to the canister. Since the
canister starting pressure was 0.50 psia, the pressure gauge will read 2.0 psia. The
canister pressure reading is allowed to stabilize, the pressure recorded, the valve is
closed, and the calibration gas is disconnected.
5. The diluent gas (humidified zero air) is then connected to the canister and bled into
the canister to the desired final pressure (30 psia) and the pressure measured once the
pressure stabilizes as 30.4 psia. The canister valve is then closed and the standard
67
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
preparation is complete. The canister is then held overnight prior to use to stabilize.
6. The final nominal concentration of the diluted standard is calculated as follows:
^stock ' Pstock
Cfinal ~ p
I 1
total
where:
Cfmai = final nominal concentration of the target compound (ppbC)
Cstock = stock gas concentration of target compound from certificate of
analysis (COA) (ppbC)
Pstock = pressure of stock gas added to the vessel (psia)
Ptotal = final total pressure of the vessel (psia)
7. For this example:
Cstock = 507 ppbC propane and 495 ppbC benzene
Pstock = 1.5 psia
Ptotal = 30.4 psia
Propane Cfmai = 507 ppbC • 1.5 psia = 25.0 ppbC
30.4 psia
Benzene Cfmai = 495 ppbC -1.5 psia = 24.4 ppbC
30.4 psia
4.5.2.2 Dynamic Dilution
Dynamic dilution is a method of preparing a known standard gas by admitting known volumetric
flows of stock standard gas(es) and a diluent gas into a mixing chamber. This method typically
employs calibrated mass flow controllers or mechanical valves or restrictors with associated
calibrated pressure transducers to control the standard and diluent gas flows. The standard and
diluent gases should be flowing at the desired rates for a sufficient time to permit complete
passivation of the dilution apparatus flow paths. Passivation timing is dependent on the mixing
chamber volume and total gas flows, but systems should be passivated for minimally five
minutes when diluting gases with lower boiling points (such as a two-component propane and
benzene standard). This passivation period is of particular importance when diluting high boiling
point VOCs (such as C9 and C10 compounds) and should be minimally 30 minutes in such cases.
Exhausts should be directed to a fume hood or snorkel.
As with static dilution, the technician should establish the desired final concentrations of the
target compounds and be familiar with the flow range limitations of the flow control devices
prior to beginning the dilution process. For example, most mass flow controllers operate
optimally within 10 to 90% of their full-scale rating (e.g., a 1000 cc/minute MFC should be
operated between 100 to 900 cc/minute) and may not control flow well outside this range. The
flows of the stock standard gas and the diluent gas are set to provide the desired DF. If a
68
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
minimum flow rate is required for delivery of the diluted gas to the auto-GC inlet, this should be
considered when setting the flow rates. Dynamically diluted standards may be delivered directly
to the auto-GC inlet or may be delivered for capture in an evacuated canister. Diluent gases
should be humidified to ensure proper quantitative transfer of all target species. Refer to Section
4.5.2.4 for information on humidification.
The following is an example for dynamically preparing a 25 ppbC standard of propane and
benzene:
Example Dynamic Dilution:
Primary Stock Gas: 512 ppbC propane and 492 ppbC benzene
Desired Working Standard Concentration: 25 ppbC
Minimum Required Flow Rate: 45 mL/minute
Standard Gas Flow Controller Range: 0 to 100 mL/minute
Diluent Gas Flow Controller Range: 0 to 5000 mL/minute
1. Calculate the effective DF needed by dividing the stock gas concentration by the
desired working standard concentration (these concentrations can be approximate
to determine the DF):
500 ppbC/25 ppbC = 20
2. For a 20-fold DF, the flow controllers will need to be within their recommended
operating ranges. To determine the total flow rate needed, multiply the standard
flow controller range minimum (10% full scale, or 10 mL/min) and maximum
(90% full scale, or 90 mL/min) by the dilution factor.
10 mL/minute ¦ 20 = 200 mL/minute
90 mL/minute ¦ 20 = 1800 mL/minute
3. Verify that the total flow exceeds the required minimum flow (preferably this
would be exceeded by a factor of 2). The minimum total flow rate of diluted gas
of 200 mL/minute exceeds 45 mL/minute, so the total flow rate will be adequate
for any selected rate of delivery of standard gas.
4. Calculate the flow rate range of the diluent channel by subtracting the standard
flow needed at the minimum and maximum potential flow rates for the standard
gas from the minimum and maximum total flows required to achieve the DF
(from step 2).
200 mL/minute - 10 mL/minute =190 mL/minute
1800 mL/minute - 90 mL/minute =1710 mL/minute
69
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
The minimum diluent flow is 500 mL/minute (10% full scale), therefore the
diluent flow must be between 500 and 1710 mL/minute. To conserve standard
gas, choose a total flow close to, but above the minimum, e.g., 600 mL/minute.
5. Determine the approximate needed standard flow corresponding to a diluent flow
of 600 mL/minute, by dividing the diluent flow by the DF:
600 mL/minute /DF = 600 mL/minute /20 = 30 mL/minute
6. Determine the needed diluent flow for a standard flow of 30 mL/minute by
subtracting the standard flow from the total flow:
600 mL/minute - 30 mL/minute = 570 mL/minute
7. The final nominal concentration of the diluted standard is calculated as follows:
n _ ^stock ' ^stock
Cfinai ~ T~^
rstock rdiluent
where:
Cfinai = final nominal concentration of the target compound (ppbC)
Cstock = stock gas concentration of target compound from COA (ppbC)
Fstock = calibrated flow of standard gas (mL/minute)
Fdiluent = calibrated flow of diluent gas (mL/minute)
8. F or thi s exampl e:
Cstock = 512 ppbC propane and 492 ppbC benzene
Fstock = 30 mL/minute
Fdiluent = 570 mL/minute
Propane Cfinai = 512 ppbC • 30 mL/minute = 25.6 ppbC
570 mL/minute + 30 mL/minute
Benzene Cfinai = 492 ppbC • 30 mL/minute = 24.6 ppbC
570 mL/minute + 30 mL/minute
4.5.2.3 Pulsed Standard Delivery
The humidified standard gas is connected directly to the instrument inlet and "pulsed" to deliver
the standard to the instrument for a known portion of the 40-minute sampling duration for an
effective dilution. This provides the instrument with the mass of target analytes desired without
employing a dynamic dilution system or preparing a standard dilution in a canister. This
convention can be problematic for delivering the desired mass to the instrument, particularly for
heavier hydrocarbons which require passivation of flow path surfaces to facilitate complete
70
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
quantitative transfer to instrument. Note that the standard gas should be humidified to ensure
proper transfer of higher molecular weight HCs.
To calculate the pulse time for the desired concentration, the following formula can be
used:
T _ Cdc ' Ts
v ~ r
lstd
where: TP = pulse time (minutes)
Cdc = desired diluted concentration (ppbC)
Ts = typical sampling time in minutes (assumed to be 40 minutes)
Cstd = concentration of stock standard gas
For example, if a benzene standard is to be prepared at 1.0 ppbC from the RTS standard
at 40 ppbC and the sampling period is 40 minutes (to achieve an 80-fold dilution):
Cdc = l.OppbC
Ts = 40 minutes
Cstd = 40 ppbC
TP = 1.0 ppbC 40 minutes/40 ppbC = 1.0 minute
To best mimic the instrument conditions during collection of ambient air, the remaining
sampling time should consist of sampling humidified zero air, where possible. Note that it may
be difficult to accurately pulse the standard delivery for short durations (less than approximately
10 minutes). Instrument manuals should be consulted to determine the minimum acceptable
sampling time when performing effective dilutions in this manner.
4.5.2.4 Humidification
Reagent water for humidification of gases should be ASTM Type I (> 18 MQcm). Additional
purifying steps, such as sonication, sparging with helium or nitrogen, or boiling may be
necessary to reduce or eliminate dissolved gases potentially present in the water. Purified water
should be stored in a sealed container to reduce the dissolution of ambient gases. Boiled water
should be loosely capped and not be sealed in a container during cooling, as the container may be
difficult to open once cooled.
Humidification of diluent gas streams is most efficiently performed by bubbling the gas to be
humidified through a bubbler via a diptube submerged in the reagent water or passing the gas
across the surface of the reagent water with an impinger. Analysts should be aware of the
potential for water to enter the bubbler tube and be introduced into the inlet gas supply tubing if
the pressure downstream of the bubbler becomes greater than the upstream pressure. Passing of
the gas to be humidified through the headspace of a vessel containing water typically achieves a
RH of 20 to 30%, which is lower than the desired moisture level of approximately 40 to 50% for
diluting standards or for use as a humidified blank matrix. Analysts should measure the RH of
the resulting humidified gas stream to ensure it reaches approximately 40 to 50%.
71
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
If this RH level cannot be achieved with an inline humidification system when employing
canisters, liquid water should be added to the canister. Approximately 75 |iL of deionized water
can be added to an evacuated 6 L canister to increase the RH to approximately 40 to 50% at
room temperature once filled to 30 psia with dry zero air. Adding water to canisters with a
syringe via rubber septum is not recommended, as the syringe needle can core the septum,
resulting in deposits of rubber into the canister and valve, leading to leaks in the canister valve or
later bias problems with the canister. For direct injection of water into a canister with a syringe, a
high-pressure Teflon® sealed septum (such as a Merlin Microseal®) should be installed on the
canister. For canisters that are connected to a gas source for pressurization via a dynamic or
static dilution system, the water can be added to the canister valve opening prior to connecting
the gas source, after which the valve is opened and the water is pulled into the canister along
with the diluted standard gas or diluent gas.
4.5.3 Auto-GC Calibration Curves
This TAD focuses on auto-GC calibration by establishing the carbon-based response with a
single hydrocarbon compound for each FID, propane or butane for the C2 to C6 PLOT column
FID channel and benzene for the C6 to C12 PDMS FID channel.
Monitoring agencies with expertise in PAMS hydrocarbon analysis may establish instrument
calibration responses for each target analyte; however, users should note that such a convention
involves a significant increase in analyst and reviewer time to input and verify known standard
concentrations into instrument software systems. Such an approach is generally not
recommended unless the monitoring agency has sufficient staff resources and experience with
such methods (such as an analyst with experience operating laboratory GCs where numerous
parameters are simultaneously measured) and understands the impacts and limitations of the
convention. Therefore, such an approach is not covered in this TAD.
Calibration standards for propane or butane and benzene should be prepared at concentrations of
approximately 1, 5 and 25 ppbC. Note that 25 ppbC was selected as the recommended high
concentration standard as ambient concentrations of most target compounds are typically no
higher than 25 ppbC. However, operators may select a concentration greater than 25 ppbC to
match the anticipated concentrations of target compounds. The three levels are analyzed on the
auto-GC and the resulting area response factors (RFs) for propane or butane and benzene are fit
to a linear regression equation to determine the carbon response for each FID. The regression
model will not be forced through the origin (y-intercept must not be set to 0) and the quality of
the curve fit will be assessed by inspection of the observed coefficient of determination (r2), |y-
intercept/slope|, RF precision (as %RSD), and accuracy of each standard's back-calculated
concentration.
The RF at each concentration level is calculated as:
72
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
where: As = area response of the calibration level
Cs = nominal concentration of compound (in ppbC)
The coefficient of determination should be > 0.99. The absolute value of the y-intercept/slope
should be < 0.5 ppbC or the < MDL, whichever is lower. The precision of each calibrant
compound's RF is determined as the RSD across the three concentrations, and should be < 10%.
The accuracy of each standard is evaluated as the back-calculated concentration determined from
the area response of each standard level inserted into the regression equation. Concentrations so
determined should be within ±20% of the nominal concentration. The acceptance criteria for
these calibration parameters are also listed in Table 4-4. Attainment of these quality metrics
verifies the instrument response is linear, accurate, and precise over the calibration range, and
that unacceptable contamination is absent. When these criteria are met, the analyst can input the
average RF into the CDS for the calibration, essentially utilizing the regression equation without
the intercept.
Some auto-GC CDSs do not include a linear regression function; however, this can be
accomplished external to the CDS with spreadsheet programs. Failure to meet one or more of the
calibration acceptance criteria suggests an inability to quantitatively deliver calibration gas or
indicates the presence of contamination or carryover in the system resulting in an elevated
carbon response factor for one or more calibration levels.
4.5.4 Second Source Calibration Verification
Once the auto-GC calibration is established for the two FIDs, the instrument calibration is to be
verified by analysis of a SSCV standard. The SSCV will minimally contain propane (or butane)
and benzene for verifying the carbon-based calibration response and preferably will also contain
additional compounds covering the C2-C10 range to verify acceptable recoveries across the
spectrum of target compound volatilities. Such a standard should include the suite of priority
compounds where possible. The SSCV concentration analyzed to verify the calibration should be
in the lower third of the calibration range. Using the RTS as the source for the SSCV is highly
recommended.
4.6 Auto-GC Operation and Quality Control
Discussed in this section are the QC samples and auto-GC analytical sequence as well as other
pertinent information for routine operation of the auto-GC for PAMS monitoring.
QC samples for auto-GC are needed during routine monitoring to ensure the instrument remains
within calibration, interferences and contamination remain acceptably low, and established RT
windows remain stable to ensure target compound identification. The QC samples, their
purpose, and their acceptance criteria are detailed in the sections below and in Table 4-4.
EPA data analysts requested that daily auto-GC QC samples be analyzed such that they do not
occur at the same hour each day (e.g., 10:00 p.m.), as such would eliminate all ambient
concentration data for that hour. In order to accommodate ambient data acquisition for each
hourly period, a schedule was developed to rotate the daily QC samples through different hours.
73
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
Note that monitoring agencies are not required to stagger the QC samples, as this may be
burdensome in preparing sample sequences and in subsequent data evaluation; however, such is
recommended where possible.
An example analysis sequence is given in Figure 4-2 for a six-week period. This example
sequence rotates the QC samples through the overnight hours of 20:00 (10:00 p.m.) through
02:00 (2:00 a.m.) local time. A CCV (the RTS serves as the CCV in this example) followed by a
system blank (SYSB) is analyzed nightly during these hours and a CCV, precision check
(replicate CCV), and SYSB are analyzed during the overnight hours from Saturday to Sunday
weekly. This sequence provides for collection of ambient samples for each hour and day of the
week, including weekend days. Analyzing the RTS as the CCV provides nightly bias check data
in addition to regular (daily) RT information for the target compounds. In the example scheme
shown, the auto-GC will be collecting 920 ambient hours of data for the 1008 hours during the
six days, equivalent to a maximum potential completeness of 91%.
Monitoring agencies have flexibility to meet the QC criteria listed in Table 4-4. For example,
agencies operating historical PAMS auto-GCs may possess established QC conventions for their
instruments and associated preferred standard gas mixtures specific to their network of sites.
EPA and PAMS stakeholders collaborated to develop the QC paradigm presented herein to
minimize the number of standard gas cylinders and sampling hours allocated to QC analysis, and
to provide data of sufficient quantity and quality to meet the re-engineered PAMS network's
quality objectives. Examples of several conventions developed to satisfy the QC requirements
are shown below in Table 4-3.
As of the publication of this document, EPA planned to offer two separate gas standards (in high
pressure cylinders) to PAMS monitoring agencies, the RTS and propane benzene mix (PBM).
EPA has historically procured a 56-component, and in recent years a 59-component, VOCs RTS
for PAMS monitoring agencies, which is NIST-traceably certified for propane and benzene.
Additionally, EPA has arranged to provide a two-component PBM standard gas to PAMS
monitoring agencies for establishing calibration for auto-GCs. With these two standard gases
(the RTS and PBM) sourced by EPA, the carbon-based calibration and all positive control QC
samples can be prepared. System blanks (negative controls) involve analysis of humidified zero
air, and do not require a standard gas. PAMS monitoring agencies may purchase additional
standard gases to satisfy in-house QC requirements and these standards may also serve to satisfy
some of the QC requirements in Table 4-4. Examples of such standards include a 15- to 18-
component standard containing VOCs across the molecular weight range and standards
containing most of the suite of priority and optional target VOCs. EPA does not intend to
provide these additional (e.g., 15-, 18-, or other) standards and are noted as "independent" below
in Table 4-3. The conventions in Table 4-3 assume that all PAMS sites will be acquiring and
utilizing the 59-component RTS. Generally, monitoring agencies are expected to calibrate the
auto-GC with a standard other than the RTS. The primary standard employed for calibration of
the auto-GC's two FIDs may be the PBM, may be an independent standard containing minimally
two compounds, or may be a combination of certified permeation tubes. Table 4-3 is generally
divided between instruments with and without onboard permeation tube capabilities (permeation
tube ovens and supporting valving and flow control to provide a known standard concentration).
74
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
As can be seen from Table 4-3 for Conventions A, D, and E, once the calibration is established,
the remaining positive QC checks can be satisfied by analysis of the RTS. Monitoring agencies
are required to verify RTs for target analytes weekly; however, nightly analysis of the RTS as the
CCV provides RT information more frequently and would permit assessment of and recovery
from observed RT drift. Similarly, if the RTS is the SSCV and analyzed daily as the CCV, the
requirement to analyze the SSCV weekly is satisfied. Thus, the RTS for these conventions
satisfies the SSCV, CCV, RTS, and precision checks.
Table 4-3. Auto-GC Quality Control Standard Conventions
Coii\ciilion
Auln-(>(' S\stems Without Onhnnrd
(¦•is Siiindiii'ds
Auln-(>(' S\stems with Onhoiii'd
PcniH-iifinn l uho Si;ind;ird
(;u);ihilit\
U<"
Piiriimclcr
l'lT(|IK'llC>
A
K
<
1)
1.
1
Initial
calibration
(ICAL)
Initially and
when failure
of CCV
necessitates
recalibration
PBM
Independent
primary 15-
component
std
Independent
primary 59-
component
std
Permeation
tubes with
two
compounds
PBM
Independent
primary 15-
, 18-, or 59-
component
std
Second
Source
Calibration
Verification
(SSCV)
Immediately
after ICAL
and weekly
thereafter
RTS
Independent
secondary
18-
component
std
RTS
RTS
RTS
RTS
Continuing
Calibration
Verification
(CCV)
Nightly
RTS
Independent
15-
component
std
Independent
59-
component
std
RTS
RTS
Independent
primary 15-
, 18-, or 59-
component
std
Retention
Time
Standard
(RTS)
Weekly
RTS
RTS
RTS
RTS
RTS
RTS
Precision
Weekly
RTS
Independent
15-
component
std
Independent
59-
component
std
RTS
RTS
Independent
primary 15-
, 18-, or 59-
component
std
75
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
week 1
week 2
sample start
time
Su
M
T
.
Th
F
Sa
Su
M
T
nr
Th
F
Sa
0:00
liali
A
A
a Hb 1 ccv
A
A
A
A
A
A
1:00
A
A
A
A CCV
A
A
A
A
E21
A
2:00
A
A
A
A
A
A
A
A
A
A
A
3:00
A
A
A
A
A
A
A
A
A
A
A
A
A
A
400
A
A
A
A
A
A
A
A
A
A
A
A
A
A
5:00
A
A
A
A
A
A
A
A
A
A
A
A
A
A
6:00
A
A
A
A
A
A
A
A
A
A
A
A
A
A
7:00
A
A
A
A
A
A
A
A
A
A
A
A
A
A
8:00
A
A
A
A
A
A
A
A
A
A
A
A
A
A
9:00
A
A
A
A
A
A
A
A
A
A
A
A
A
A
I COO
A
A
A
A
A
A
A
A
A
A
A
A
A
A
1100
A
A
A
A
A
A
A
A
A
A
A
A
A
A
12:00
A
A
A
A
A
A
A
A
A
A
A
A
A
A
13:00
A
A
A
A
A
A
A
A
A
A
A
A
A
A
14 XX)
A
A
A
A
A
A
A
A
A
A
A
A
A
A
15.00
A
A
A
A
A
A
A
A
A
A
A
A
A
A
16.00
A
A
A
A
A
A
A
A
A
A
A
A
A
A
1700
A
A
A
A
A
A
A
A
A
A
A
A
A
A
1800
A
A
A
A
A
A
A
A
A
A
A
A
A
A
19:00
A
A
A
A
A
A
A
A
A
A
A
A
A
A
20:00
CCV
A
A
A
A
A
' A
A
A
A
A
CCV
A
21:00
CCV 1
A
A
A
A
i CCV I A
A
A
A
A
22:00
A
MM
A
A
1 ccv
A
A
A
A
A
23:00
A
, A
l££i
A
A
|a!
[W
-mr\
A
A
1 A
A
week 4
sample start
time
SU
*
W
Th
F
Sa
0:00
A
¦
CCV
A
A
A
A
1:00
A
A
SYSB
CCV
A
A
A
2:00
A
A
A
SYSB
A
A
A
3:00
A
A
A
A
A
A
A
4:00
A
A
A
A
A
A
A
5:00
A
A
A
A
A
A
A
6:00
A
A
A
A
A
A
A
7:00
A
A
A
A
A
A
A
8:00
A
A
A
A
A
A
A
9:00
A
A
A
A
A
A
A
10:00
A
A
A
A
A
A
A
1100
A
A
A
A
A
A
A
12.00
A
A
A
A
A
A
A
1300
A
A
A
A
A
A
A
1400
A
A
A
A
A
A
A
1500
A
A
A
A
A
A
A
1600
A
A
A
A
A
A
A
1700
A
A
A
A
A
A
A
1800
A
A
A
A
A
A
A
1900
A
A
A
A
A
A
A
A
A
A
A
A
A
21:00
A
A
A
A
A
22:00
A
A
A
A Hni
A
23:00 [ CCV
A
A
A
A
A
week 5
Su
M
T
W
Th
F
Sa
A
A
A
A
A
A
^mTWM a
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A I CCV 1
A
A
A
A
A
A
A
A
B ambient sample
hours for possible QC analysis
continuing calibration verification
system blank
precision check, replicate CCV
week 3
Su
M
T W
Th
F
Sa
EsaJ
KB En
A
A
A
[_a_^^BEBH|
A
A
A
A
a^BaJ
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
E53
A
A
A
A
A
A
CCV
fcBI
| |
A
A
•-•B I P
1SYSB1 LLV
A
A
week 6
Su
M
T
W
Th
F
Sa
A
A
A
A
A
A
A
CCV
A
A
A
A
A
A
SYSB
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
Figure 4-2. Example Auto-GC Sampling Sequence for Ambient and QC Samples
4.6.1 System Blanks
SYSBs consisting of humidified zero air are to be analyzed daily to demonstrate the instrument
is free of interferences, carryover, and contamination. The SYSB may be analyzed prior to or
following the nightly CCV, per analyst discretion. If analyzed prior to the CCV, the blank
should show the instrument is sufficiently clean after ambient analysis and prior to the CCV. If
analyzed following the CCV, the blank will demonstrate the instrument is sufficiently clean prior
to the next ambient sample. The response for each target analyte in the SYSB should be as low
as possible, and will be less than or equal to 0.5 ppbC or the MDL, whichever is lower. TNMOC
should be less than 10 ppbC.
76
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
SYSBs may also be analyzed for instrument conditioning such as for conditioning new
preconcentrator traps, Nafion™ dryers, or GC columns, or for cleaning or purging the instrument
following the analysis of high concentration standards or samples. In such instances, the
measured concentrations do not need to meet blank acceptance criteria except for blanks that
precede ambient samples measured with the intent to report to AQS. For such blanks, only the
blank immediately preceding the sample to be reported will need to meet acceptance criteria.
Note that SYSBs analyzed during the course of routine sample analysis will be reported to AQS
as discussed in Section 11.2. It is recommended that a blank be analyzed following positive
control (known standard) QC samples (such as CCV, RTS, and SSCV) to ensure the auto-GC is
sufficiently clean before analyzing ambient samples.
4.6.2 Continuing Calibration Verification (CCV)
Once the calibration is established and verified by the SSCV, it is verified on an ongoing basis
by daily analysis of a known standard at a concentration in the lower third of the calibration
curve. This is recommended to be a humidified dilution of the RTS to approximately 2 to 5
ppbC. The CCV should be analyzed during overnight hours, preferably between 22:00 and 03:00
to avoid missing sampling hours during the day when photochemistry characterization is most
important.
The concentrations of propane and benzene in the CCV will verify the ongoing acceptability of
the initial calibration and the responses of additional compounds covering the entirety of the
molecular weight range (C2 to C10) will assess that instrument performance remains stable for the
suite of target compounds. Measured propane and benzene concentrations should be within
±30% of their nominal concentrations and other compounds in the CCV should also be within
±30%. Failures of the propane and/or benzene response require corrective action which may
include recalibration. Failures of additional target compounds should prompt corrective action,
where appropriate. For example, if compounds such as acetylene, styrene, or the pinene isomers
show low recovery, the analyst may have little recourse for corrective action, particularly if other
compounds are not demonstrating recovery problems. However, if several typically well-
behaved compounds fail criteria, root cause analysis should be performed to investigate the
problem. In all cases, ambient data for the affected compounds failing acceptance criteria should
be appropriately qualified back to the most recent acceptable CCV.
4.6.3 Precision Check
The precision check consists of a second consecutive (back-to-back) CCV analysis that is
analyzed weekly. The precision check should meet the CCV requirements for recovery and
additionally should show RPD < 25% from the immediately preceding CCV.
4.6.4 Retention Time Standard
The RTS is analyzed routinely (minimally weekly) to evaluate the ongoing acceptability of the
RT windows established in the CDS. There are no acceptance criteria for the RTS; instrument
operators will verify the target compound RT windows or may make adjustments based on
shifting of RTs.
77
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
Table 4-4. Speciated VOCs Quality Control Parameters Summary
0<
Description
Kcipiiml
Acccphincc
Recommended ( oiTcc(i\c
Piii'iiiiKMcr
l"lV(|IICIIC\
( rilcrhi
Action
Initial
Multi-point calibration
Initially at the
Linear regression
Analysis cannot commence if
calibration
of the auto-GC with
beginning of
with non-zero y-
propane (or butane) or benzene
(ICAL)
minimally a
PAMS season,
intercept should
fail to demonstrate appropriate
representative
after
show r2 of > 0.99.
criteria for the ICAL.
hydrocarbon for each
maintenance
Also
Investigate chromatogram for
GC column-FID
(such that
intercept/slope <
retention time shifts which may
combination (e.g.
response is
0.5 ppbC or<
result in peak misidentification.
propane and benzene).
impacted),
MDL, whichever
Investigate for instrument
Minimum of three
following
is lower. RSD of
contamination resulting in co-
concentrations
failing
determined RFs
eluting peaks. Investigate for
covering
continuing
must be <10%.
system leaks or trap
approximately 1.0 to
calibration
Each standard
malfunction resulting in low
25 ppbC. At their
checks, and at
level evaluated
recovery. Investigate standard
discretion, agencies
the conclusion
against the
introduction system for errors
may use other higher-
of monitoring
calibration curve
or malfunction. Unless
level concentrations
each PAMS
should be within
technical justification is
(e.g., 50 or 80 ppbC).
season.
20% of the
provided to explain
Agencies may
nominal
nonconformance, minimally
analyze the
concentration. If
qualify as "QX" and potentially
primary
all of the above
invalidate as "AS" samples for
calibration
criteria (r2, y-
which acceptable carbon-based
standard
intercept/slope,
calibration could not be
weekly as an
RF RSD, and
established.
additional
standard ±20% of
check to
nominal) are met,
monitor
the calibration
system
may utilize the
performance -
average RF.
not required.
Measurements
exceeding the
calibration range
will be qualified
as "EH".
System blank
Analysis of humidified
Prior to ICAL,
All target VOCs
Analyze another blank, if
(SYSB)
zero air to ensure the
and every 24 ±
should be < the
possible, to investigate
system is sufficiently
4 hours of
determined MDL
potential carryover from high
clean for continued
operation
or 0.5 ppbC,
concentration sample.
analysis.
thereafter, to
whichever is
Investigate system for
follow or
lower.
contamination. Unless
precede the
technical justification is
CCV
provided to explain
(preference is
nonconformance, qualify as
to follow the
"LB" in AQS all samples for
CCV to ensure
affected compounds since the
absence of
last passing SYSB.
carryover
before
continuing to
analyze
ambient
samples).
78
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
Table 4-4 (continued). Speciated VOCs C
•uality Control Parameters Summary
Piii'iiiiKMcr
Description
Kc(|iiiml
l'lV(|IK'IIC\
Acccpliiiicc
(rik-rhi
UiTnmiiH'mk'ri (onvcli\ e
Acl inn
Second
Source
Calibration
Verification
(SSCV)
Analysis of a known
standard prepared
from a stock gas
including target
analytes across the
molecular weight
range from a supplier
different from the
stock gas (primary
standard) for preparing
the ICAL. This check
independently verifies
the quality of the
ICAL for compounds
across the molecular
weight range.
Immediately
following
ICAL and
minimally
weekly
thereafter -
may serve as
the CCV
All target VOCs
should recover
within ±30% of
the expected
nominal
concentration.
Analysis cannot commence if
propane (or butane) or benzene
fail in the SSCV immediately
following the ICAL.
Investigate for discrepancy
between ICAL and SSCV.
Investigate chromatogram for
retention time shifts which may
result in peak misidentification.
Investigate for instrument
contamination resulting in co-
eluting peaks. Investigate for
system leaks or trap
malfunction resulting in low
recovery. Unless technical
justification is provided to
explain nonconformance,
minimally qualify as "QX" and
potentially invalidate as "AS"
samples for affected
compounds since the last
acceptable SSCV.
Continuing
Calibration
Verification
(CCV)
Analysis of a known
standard containing
compounds
representing the
molecular weight
range prepared within
the calibration curve
to demonstrate the
instrument calibration
remains within
tolerance.
Concentration of CCV
should be
approximately 2-5
ppbC for target
analytes.
Every 24 ± 4
hours of
operation
All target VOCs
should recover
within ±30% of
the expected
nominal
concentration.
Investigate chromatogram for
retention time shifts which may
result in peak misidentification.
Investigate for instrument
contamination resulting in co-
eluting peaks. Investigate for
system leaks or trap
malfunction resulting in low
recovery. Unless technical
justification is provided to
explain nonconformance,
qualify as "QX" in AQS all
samples for affected
compounds since the most
recent passing CCV.
Invalidation as "AS" may be
required at analyst discretion if
compound recovery is
exceptionally high or low.
Retention
Time
Standard
(RTS)
Analysis of a multi-
component (e.g. 59-
compound blend)
blend of each target
VOC (minimally all
priority compounds
and any reported
optional compounds)
in the ~2 to 60 ppbC
range to verify
established retention
time windows
Minimally
weekly
All target VOCs
should be within
the established
retention time
windows.
Review previous week's
ambient and QC check sample
data to evaluate events
resulting in retention time shift.
May require reassignment or
adjustment of retention time
windows and reprocessing of
data collected since the most
recent CCV or RTS. Unless
technical justification is
provided to explain
nonconformance, associated
79
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
Table 4-4 (continued). Speciated VOCs C
•uality Control Parameters Summary
Piii'iiiiKMcr
Description
Kc(|iiiml
l'lV(|IK'IIC\
Acccpliiiicc
(rik-rhi
UiTnmiiH'mk'ri (onvcli\ e
Acl inn
ambient sample data will be
invalidated as "BH" for
compounds whose identities
cannot be confirmed.
Precision
check
Replicate analysis of
the CCV to evaluate
the reproducibility of
the analysis -
replicates are analyzed
sequentially (back to
back)
Weekly
All target VOCs
should recover
within ±30% of
the expected
nominal
concentration.
Absolute relative
percent difference
of each target
VOCs
concentration
should be
< 25% on a week-
to-week basis.
Investigate system for
carryover, contamination,
leaks, or suppression, as
indicated by trends in
compound behavior. Qualify
ambient sample data for
affected compounds since the
last passing precision check as
"QX" in AQS.
Clock
Accuracy
Verify clock accuracy
against a known
accurate time standard
Weekly,
recommended
to check each
site visit
Within ±5
minutes of the
time standard
Reset clock to correct time.
Adjust data timestamp
accordingly where possible.
Ensure adjusted sampling start
times are no earlier than 10
minutes before the hour and no
later than 30 minutes after the
hour. Invalidate sample hours
that do not conform.
4.7 References
1. Spicer, C. W., S. M. Gordon, et al. (2002). Hazardous Air Pollutants Handbook:
Measurement Properties, and Fate in Ambient Air. Boca Raton, Lewis Publishers.
2. Cavender, Kevin A., Weinstock, Lewis; Memorandum, Revisions to the Photochemical
Assessment Monitoring Stations Compound Target List, EPA Office of Air Quality Planning
and Standards, November 20, 2013. Available at (accessed March 2018):
https://www3.epa.eov/ttnamtil/files/ambient/pams/tareetli; 13.pdf
3. Methodfor the Determination of Non-methane Organic Compounds (NMOC) in Ambient Air
Using Cryogenic Preconcentration and Direct Flame Ionization Detection (PDFID); EPA
Compendium Method TO-12; U.S. Environmental Protection Agency: 1999. Available at
(accessed March 2018): https://www3.epa.eov/ttnamtil/files/ambient/airtox/tocom.p99.pdf
4. RTI International and EC/R Incorporated, Gas Chromatograph (GC) Evaluation Study,
Laboratory Evaluation Phase Report. October 3, 2014. Available at (accessed March 2018):
https://www3.epa.eov/ttnamtil/files/ambient/pams/labevalreport.pdf
5. EC/R Incorporated and RTI International, Gas Chromatograph (GC) Evaluation Study, Field
Deployment Evaluation Report. February 28, 2017. Available at (accessed March 2018):
80
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
¦epa.gov/ttnamtil/files/ambient/pams/FINAL-AutoGC-Field-Evaluation-
Report.pdf
6. Quality Assurance Handbook for Air Pollution Measurement Systems. Volume II - Ambient
Air Quality Monitoring Program, EPA-454/B-17-001, U. S. Environmental Protection
Agency, January 2017.
7. Perma Pure® LLC, "All About Nation" http://www.permapure.com/resources/all-about-
nafion-and-faq/
8. Foulger, B. E. and P. G. Simmonds. Drier for Field Use in Determination of Trace
Atmospheric Gases. Anal. Chem., 51: 1089-1090 (1979).
9. Burns, W. F., D. T. Tingey, R. C. Evans, and E. H. Bates. Problems with a Nafion®
Membrane Dryer for Drying Chromatographic Samples. J. Chromatogr. 269: 1-9 (1983).
10. Pleil, J. D., K. D. Oliver, and W. A. McClenny. Enhanced Performance of Nafion® Dryers in
Removing Water from Air Samples Prior to Gas Chromatographic Analysis. J. Air Pollut.
Control Assoc. 37: 244-248 (1987).
11. Leckrone, K. J. and Hayes, J. M. Efficiency and Temperature Dependence of Water
Removal by Membrane Dryers. Anal. Chem. 69: 911-918 (1997).
12. Zielinska, B., Sagebiel, J. C., Harshfield, G., Gertler, A. W., Pierson, W. R. Volatile Organic
Compounds up to C20 Emitted From Motor Vehicles; Measurement Methods. Atmospheric
Environment, Vol. 30: No. 12. 2269-2286. 1996.
13. Gong, Qing, and K. L. Demeijian. Hydrocarbon Losses on a Regenerated Nafion® Drier. J.
Air Waste Manage. Assoc. 45: 490-493 (1995).
14. Furdyna, P., Perry, J., Shipley, E., and Felton, D. NYSDEC PAMS Air Monitoring with a
Markes Agilent Auto GC System. Poster presented at the National Ambient Air Monitoring
Conference, Portland Oregon, August 13-16, 2018.
15. Brown, Jamie, Supelco. Choosing the Right Adsorbent for your Thermal Desorption Gas
Chromatography Applications. Presented at the Separation Science Webinar, October 22,
2013. Available at (accessed March 2018):
https://www.sigmaaldrich.com/content/dam/sigma-
aldrich/docs/Supelco/Posters/l/Adsorbent-Selection-TD-GC-Apps.pdf
16. Dietz, W. A. (1967). Response factors for gas chromatographic analyses. Journal of Gas
Chromatography, February 1967, 68-71.
17. Perkins, G., Rouayheb, G. M., Lively, L. D., & Hamilton, W. C. (1962). Response of the Gas
Chromatographic Flame Ionization Detector to Different Functional Groups. In N. Brenner,
J. E. Callen, & M. D. Weiss (Eds.), Gas Chromatography. New York: Academic Press.
18. Scanlon, J. T., Willis, D. E., Calculation of Flame Ionization Detector Relative Response
Factors Using the Effective Carbon Number Concept. Journal of Chromatographic Science,
Vol. 23, August 1985. pp 334-340.
19. Faiola, C. L., Erickson, M. H., Fricaud, V. L., Jobson, B. T., and VanReken, T. M.,
Quantification of Biogenic Volatile Organic Compounds with a Flame Ionization Detector
81
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
Using the Effective Carbon Number Concept. Atmospheric Measurement Techniques, Vol. 5,
August 2012. pp 1911-1923.
20. Definition and Procedure for the Determination of the Method Detection Limit, Revision.
EPA Office of Water, EPA 821-R-16-006, December 2016. Available at (accessed March
2018): https://www.epa. gov/sites/production/files/2016-12/docum ents/m dl -
procedui >df
21. U.S. EPA. Photochemical Assessment Monitoring Stations (PAMS): Data Assessment.
January 23, 2001.
22. Allen, N. D. C., Worton, D. R., Brewer, P. J., Pascale, C., & Niederhauser, B. (2018). The
importance of cylinder passivation chemistry for preparation and long-term stability of
multicomponent monoterpene primary reference materials. Atmospheric Measurement
Techniques, 2018, 1-18. doi:10.5194/amt-2018-165
82
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
5.0 CARBONYL COMPOUNDS VIA EPA COMPENDIUM METHOD TO-11A
In 2006, due to concerns regarding the accuracy of method TO-11A for carbonyls measurement,
EPA eliminated the requirement to measure carbonyls at PAMS sites with the exception of sites
in areas designated as severe or extreme non-attainment for the 8-hour ozone standard (July 1997
standard - annual fourth highest daily maximum 8-hour concentration averaged over 3 years
could not exceed 0.08 ppm). EPA's evaluation of the target compound list identified that
carbonyl compounds, specifically formaldehyde and acetaldehyde, are ubiquitous in the ambient
atmosphere and have a very high maximal incremental reactivity.1 These aldehydes were added
to the priority compound list. Acetone and benzaldehyde were added as optional compounds.
EPA has recently evaluated method TO-11A to better characterize the performance of the
collection and analysis methods. Part of this evaluation has been to quantify and characterize the
limitations of the collection and analysis to determine optimized parameters. As of publication of
this document, EPA has begun revision of TO-11A based on work performed to optimize and
modernize the method. Guidance in this section reflects many of the outcomes of the studies
performed to update the method. EPA plans to announce the revised method when published
and monitoring agencies should expect communication regarding updates to method
performance and acceptance criteria as they relate to the PAMS program.
Each agency is to prescribe in an appropriate quality systems document, such as an SOP, or
equivalent, its procedures for collection of airborne carbonyls onto cartridges, extraction of the
cartridges, and analysis of the extracts. Various requirements and best practices for such are
given in this section. Note that regardless of the specific procedures adopted, method
performance QC criteria given in Table 5-5 are to be met.
5.1 General Description of Sampling Method and Analytical Method
Carbonyl compounds such as aldehydes and ketones may be collected and analyzed via EPA
Compendium Method TO-11 A. As priority compounds, all PAMS sites are to measure and
report acetaldehyde and formaldehyde. Sites are encouraged to additionally measure and report
acetone and benzaldehyde, which are optional compounds, as well as additional carbonyls listed
in Table 5-1. EPA recognizes that additional resources are required to provide quality-assured
data for these additional optional analytes; however, given that this method is already conducted
to measure the priority compounds, data for many of the optional PAMS compounds and
additional hazardous air pollutant (HAP) analytes can be reported with modest additional
resource input.
The atmosphere to be characterized is drawn at a known flow rate for a known duration of time
through an ozone denuder and through a sorbent cartridge coated with DNPH, where the
carbonyl compounds react with the DNPH and are derivatized to form carbonyl-hydrazones.
These carbonyl-hydrazones are solids at typical ambient temperatures and are retained on the
cartridge sorbent bed until eluted with acetonitrile (ACN). Eluted extracts are analyzed by HPLC
(or ultra-high performance liquid chromatograph [UHPLC]) with a UV detector at a wavelength
of 360 nm.2
83
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
The carbonyls including, but not limited to, those in Table 5-1 may be determined by this
method.
Table 5-1. Carbonyl Target Compounds Measured by Method TO-11A
Tariiel ( arhninl
CAS#
AQS Parameter II)
acetaldehydea
75-07-0
43503
acetone b
67-64-1
43551
benzaldehyde b
100-52-7
45501
formaldehydea
50-00-0
43502
Additional Carbonyls That May Be Quantitated by TO-11A and Reported toAQS
butyraldehyde
123-72-8
43510
crotonaldehyde
4170-30-3
43528
2,5 -dimethy lbenzaldehy de
5779-94-2
45503
heptaldehyde
111-71-7
43950
hexaldehyde
66-25-1
43517
isovaleraldehyde
590-86-3
43513
m&p-tolualdehyde
(m) 620-23-5/(p) 104-87-0
45506
methyl ethyl ketone
78-93-3
43552
methyl isobutyl ketone
108-10-1
43560
o-tolualdehyde
529-20-4
45505
propionaldehyde
123-38-6
43504
valeraldehyde
110-62-3
43518
" PAMS priority compound
b PAMS optional compound
5.2 Minimizing Bias
The sampling of airborne carbonyls onto DNPH cartridges is potentially affected by a variety of
interferences. For example, nitrogen oxides react with the DNPH derivatives to form compounds
which may coelute with carbonyl-hydrazone derivatives. Moreover, ozone reacts with DNPH to
form possible coeluting interferences and also reacts with and causes negative bias in the
measurement of various carbonyl-hydrazones. To minimize introduction of contamination and to
keep bias to a minimum, manage ozone per Section 5.4 and handle cartridges as in Section 5.5.2.
Clean labware and select high-purity reagents as in Section 5.9.1.2.
The cartridge inlet and outlet caps are to be installed when the cartridge is not in use so as to
isolate it from the ambient atmosphere where carbonyl compounds and interfering compounds
may be passively sampled. Further, cartridges are to be stored sealed in the foil pouch or similar
opaque container, as light may degrade the DNPH derivatives. Finally, DNPH cartridges are to
be stored at < 4°C after sampling to slow the reaction of contaminants. Cartridges should only be
handled while wearing powder-free nitrile or vinyl gloves.
5.3 Carbonyls Precision
EPA's goal is to have minimally 10% of PAMS Required Sites conduct precision sample
collection at a frequency of 10% of the primary sample collection. EPA encourages all PAMS
84
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
sites to conduct precision sampling, if possible, and report the concentration data to AQS.
Monitoring agencies are to describe precision sampling in the agency ANP and/or PAMS QAPP.
5.3.1 Sampling Precision
Depending on the configuration of the sampling unit or units at the monitoring site, sampling
precision may be assessed by way of the collection and analysis of collocated or duplicate
sample cartridges. Duplicate and collocated samples are collected over the same duration and
sample the same air mass as the primary sample. Sampling precision is a measure of the
reproducibility in the sampling, handling, extraction, and analysis procedures. Monitoring
agencies are encouraged to collect duplicate samples, where possible, and to collect collocated
samples where equipment is available to do so. For monitoring agencies collecting collocated
and/or duplicate samples, it is recommended that the frequency be 10% of the primary samples.
At publishing of this document, there were two commercially-available carbonyls sampling unit
models capable of collecting three sequential 8-hour samples. The instrument configurations are
such that one of the models can collect up to three sequential samples and does not include a
separate channel for collection of duplicate samples, and the other instrument can collect up to
eight sequential samples and may be purchased in configurations that include one or two separate
channels for collection of duplicate samples.
5.3.1.1 Collocated Sample Collection
A collocated sample is a sample for which air is drawn through a co-collected cartridge from an
independent inlet probe through a separate discrete sampling unit or with a single sampling
instrument configured with two separate flow paths and flow controls (at publishing of this
document, no such instrument with sequential sampling capability and separate inlets for
collocated sampling is commercially available). Collocated sampling typically involves a
completely separate sampling unit, inlet probe, and flow path (Figure 5-1). If two cartridges are
collected together with such a single sampling instrument, to be collocated the air passing onto
each cartridge is to flow through wholly separate channels, where each channel is to have a
discrete inlet probe, plumbing (including ozone denuder), pump, and flow controller such as an
MFC or rotameter. Collocated sampling provides for an estimate of variability of the complete
measurement system.
More information on collocated samples is given in Section 5.8.2.3.
85
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
COLLOCATED
standalone inlet probes
manifold B
manifold A
sampling unit
sampling unit
sampling unit
sampling unit
DUPLICATE
standalone inlet probe
sampling unit
A
u
sampling unit
manifold inlet
probe
manifold inlet
probe
sampling unit
sampling unit
Figure 5-1. Collocated and Duplicate Carbonyls Sample Collection
5.3.1.2 Duplicate Sample Collection
Duplicate sampling assumes that both the primary and duplicate sampling inlets are connected to
the same inlet probe to the atmosphere whether connected to a manifold or a standalone inlet
probe.
86
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
A duplicate sample can be collected, for example, by splitting (with a tee, or similar) the primary
sample flow path onto two separate cartridges, where each cartridge has its own discrete and
separate flow channel and/or flow control device (MFC, orifice, or rotameter) located within a
single sampling unit. Duplicate sample collection can provide for estimates of variability of some
field aspects and serves to distinguish field from laboratory variability.
More information on duplicate samples is given in Section 5.8.2.4.
5.3.2 Laboratory Precision
Laboratory precision for field-collected carbonyls cartridges is limited to replicate analysis of a
single extract. A replicate analysis is a second discrete analysis of one sample extract. Each
DNPH cartridge is extracted as a discrete sample that does not permit assessing precision
through the extraction process. Replicate analysis of a given extract is required with each
analysis sequence and is to show < 10% RPD for concentrations > 0.5 |ig/cartridge.
Precision incorporating both the extraction and analysis procedures may be assessed by
preparation, extraction, and analysis of duplicate LCSs. An LCS and LCSD are to be prepared
minimally quarterly and are recommended with each extraction batch at a concentration in the
lower third of the calibration range. The LCS/LCSD pair is to show precision of < 20% RPD.
5.4 Managing Ozone
Ozone is present in the atmosphere at various concentrations ranging from approximately 20 ppb
at non-source impacted locations to as much as 150 ppb at peak times in urban environments.
Ozone is a strong oxidant and may impact the sampling and analysis in various ways. Ozone that
is not removed from the sampled air stream may react directly with the DNPH reagent, thereby
making the DNPH unavailable for derivatizing carbonyl compounds. Ozone may also react with
carbonyl-hydrazones on the sampled cartridge to degrade these compounds, leading to
underestimation of carbonyl concentrations. These degradation byproducts may also be difficult
or impossible to separate chromatographically from desired target compounds, resulting in
overestimation or false positive detection of target compounds.
In order to mitigate the impact of ozone on carbonyl measurements, an ozone denuder/scrubber
is to be installed in the sampling unit flow path upstream of the DNPH cartridge(s). Typically,
the removal of ozone by potassium iodide (KI) is affected by the oxidation of the iodide ion to
iodine in the presence of water, as follows:
O3 —* 02 + 0
+ 2KI + H20 + 0 -> 2KOH + I2
03 + 2KI + H20 -> I2 + 02 + 2KOH
Several different KI ozone scrubbers are described in the following sections. However, to ensure
comparability and consistency across the PAMS network, ozone is to be removed during the
collection of carbonyls with the denuder in Section 5.4.1.
87
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
5.4.1 Copper Tubing Denuder/Scrubber
Method TO-11A describes an ozone denuder/scrubber and this is the preferred ozone removal
method for the PAMS program. The scrubber is made from coiled copper or stainless steel
tubing where the interior has been coated with a saturated KI solution and is heated to
approximately 50°C or above to eliminate condensation. Heating prevents the deposition of
liquid water to the denuder walls which may both dissolve the KI coating and clog the silica gel
pores in the DNPH cartridge with KI as it recrystallizes. As this type of scrubber/denuder
operates via titration, its efficacy over time is related to the amount of deposited KI, the total
volume of sampled air, and the average ozone concentration of the sampled air. In general, it is
presumed that this type of denuder/scrubber should be effective for up to 100,000 ppb-hours at
flow rates of less than 1 L/minute.2 Ongoing EPA-funded work has confirmed that such ozone
scrubbers are effective for the 100,000 ppb-hours cited in TO-11 A; they were able to efficiently
remove 150 ppb O3 over 30 consecutive days when operated at a flow rate of 1 L/min at RHs
ranging from 10 to 85% at a nominal temperature of 25°C.3 Given an average ozone
concentration of approximately 70 ppb, this type of denuder/scrubber should effectively scrub
ozone from the sampled air stream for roughly double the minimum 30 collection days of three
consecutive 8-hour samples minimally required by the PAMS Program without depleting the KI
reagent. If the average concentration of ozone is greater than 70 ppb over the course of the
sampling season or the sampling frequency is increased from one in three days, or if duplicate
sampling is performed more frequently than every other month such that the flow rate through
the denuder is doubled during most sampling events (thereby exposing the scrubber to twice the
burden of ozone), the life span of the KI denuder/scrubber will be proportionately reduced.
The denuder/scrubber should be replaced or recharged with KI minimally every other PAMS
season to ensure there is sufficient KI substrate to eliminate co-sampled ozone; they should also
be recharged if ozone breakthrough is observed as decomposition products of O3 attacking the
DNPH and the formaldehyde hydrazone derivative. Denuders are commercially available or they
can be recharged by recoating the copper tubing with a saturated solution of KI in deionized
water (144 grams KI in 100 mL deionized water). The solution is maintained inside the copper
tubing for minimally 15 minutes (some agencies suggest 24 hours or more), then the solution
drained. The emptied tubing is then dried by a gentle stream of dry ultra-high purity (UHP)
nitrogen for minimally one hour.
When a sampling instrument is removed from service for recharging the KI denuder/scrubber
and/or for calibration/maintenance, a best practice is to challenge the denuder with ozone at
120% of the maximum site-measured ozone concentration for several hours and measure the
resultant downstream ozone concentration. Such will demonstrate the ozone scrubber's efficacy
prior to removal from the field. For denuders shown to be less than fully effective upon removal
from the field, defined as downstream ozone concentration >10 ppb or a breakthrough > 5% of
the challenged concentration, chromatograms from recent sampling events should be examined
for indications of ozone interference. Following recharge/replacement of the KI
denuder/scrubber, the 120% ozone concentration challenge should be repeated to demonstrate
effective ozone removal prior to its deployment for field use. The zero challenge of the sampling
unit prescribed in Section 5.7.1.1 is to be performed following recharging of the
denuder/scrubber.
88
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
5.4.2 Sorbent Cartridge Scrubbers
Sorbent cartridges, such as silica gel, coated with KI are commercially available, but their use is
strongly advised against due to the sorbent bed becoming saturated with water, resulting in
clogging of the cartridge substrate which substantially reduces or eliminates sample flow. While
inexpensive and convenient for use, sorbent bed KI cartridges should not be employed for PAMS
sampling.
5.4.3 Other Ozone Scrubbers
Agencies may opt to develop custom-made KI ozone scrubber/denuders. The efficiency of ozone
removal is to be demonstrated for such custom systems. To demonstrate efficiency of ozone
removal, the homemade scrubber/denuder is to be challenged over a contiguous 24-hour period
with a minimum of 100 ppb ozone at the flow rate for the carbonyl instrument sampler (typically
approximately 1 L/min) and demonstrate breakthrough of < 5%. Agencies are also to quantify
the capacity of such scrubbers (for example, in ppb-hours) and with such data they should
determine and codify in their quality system the minimum required recharge/replacement
frequency of the scrubbers. Again, to ensure comparability and consistency across the PAMS
network, the ozone denuder described in Section 5.4.1 is strongly encouraged.
5.5 Collection Media
EPA Compendium Method TO-11A specifies DNPH-coated silica gel sorbent cartridges for the
collection of carbonyl compounds from ambient air. These DNPH cartridges may be prepared in-
house or purchased from commercial suppliers. PAMS sites will typically utilize one of two
commercial brands of media, specifically the Waters WAT037500 or Supelco S-10 cartridges.
These cartridges are specified to meet the background criteria of TO-11A and typically exhibit
proper flow characteristics. Examination of background concentrations and proficiency test data
do not indicate an obvious difference in the performance between the two brands of cartridges.
Laboratories may prepare DNPH cartridges in-house; however, preparation is a time- and labor-
intensive process that requires meticulous detail to cleanliness to ensure the resulting media are
contaminant-free. The expense and resources involved in preparation of DNPH media in-house
are generally greater than the cost of purchasing commercially-available DNPH cartridge media.
Regardless of the type of cartridge selected, the method performance specifications in Section
5.5.1 should be met.
5.5.1 Lot Evaluation and Acceptance Criteria
For each lot or batch of purchased or prepared DNPH cartridge, a representative number of
cartridges should be analyzed to demonstrate that the lot or batch is sufficiently free of
contamination. Most commercially-available DNPH cartridges are accompanied by a COA
indicating the lot or batch background of various carbonyls. While a COA provides a level of
confidence that the lot or batch is sufficiently clean, laboratories should verify the background
levels of carbonyls in each batch or lot of cartridges.
89
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
For commercially-purchased cartridges, a minimum of three cartridges, or 1% of the total lot,
whichever is greater from each lot or batch, should be extracted and analyzed. For cartridges
prepared in-house, a minimum of three cartridges per each preparation batch should be extracted
and analyzed. Each cartridge tested in the lot or batch should meet the criteria listed in
Table 5-2. Ongoing analysis of method blanks permits continual assessment of the lot's
contamination levels.
Additionally, agencies may elect to perform flow evaluations of the lot(s) to ensure cartridges do
not overly restrict sampling flows.
Table 5-2. Maximum Background per Lot of DNPH Cartridge
( iirhmnl Compound
Acit|>I;iikt l imit (fig/csirlririgi')
Acetaldehyde
<0.10
Formaldehyde
<0.15
Acetone
<0.30
Other Individual Target Carbonyl Compounds
<0.10
If any cartridge tested exceeds these criteria, an additional three cartridges, or 1% of the total lot,
whichever is greater, should be tested to evaluate the lot. If the additional cartridges meet the
criteria, the lot or batch is acceptable for sampling. If any of the additional cartridges fail criteria,
the lot or batch should not be used for PAMS sampling and should be returned to the provider or
discarded.
5.5.2 Cartridge Handling and Storage
DNPH sampling cartridge media are typically shipped unrefrigerated by the supplier. DNPH
cartridges should be stored refrigerated at < 4°C upon receipt (note that freezing temperatures are
permitted and do not affect the cartridge integrity). Users are cautioned that storage at extreme
cold temperatures (e.g., -80°C as is typical for biological tissue storage) are not recommended
and may cause cartridge housings to separate; however, the DNPH and sorbent beds will not be
affected. Unsampled cartridges should be maintained sealed in their original packaging and
protected from light (foil pouch or similar opaque container) until installed for sample collection
or use as QC samples as light may degrade the DNPH derivatives. Cartridges that are not stored
appropriately may suffer from degradation of the DNPH reagent and may show increased levels
of contaminants from passive sampling of target compounds and interferants.
DNPH cartridges should only be handled by staff wearing powder-free nitrile or vinyl gloves or
equivalent. Measures are to be taken to avoid exposure of DNPH cartridges (unsampled or
collected samples) to exhaust fumes, sunlight, elevated temperatures, and laboratory
environments where carbonyl compounds such as acetone may contaminate sampling media.
As soon as possible after sample collection, cartridges are capped (if caps are provided), sealed
in the foil pouch (to protect from light and the ambient atmosphere), and transported (shipped)
and stored refrigerated at < 4°C. Cartridges are transported in coolers with ice, freezer packs, or
an equivalent method for providing refrigeration during transport to and from the laboratory.
Monitoring the shipping temperature with a calibrated min-max type thermometer is a best
90
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
practice. Sample cartridge temperature should be recorded at the laboratory upon receipt. This
can be accomplished by recording the temperature from the calibrated min-max thermometer
included in the shipment or measuring the temperature with a calibrated infrared thermometer.
5.5.3 Damaged Cartridges
DNPH cartridges are susceptible to water damage and to physical damage. Unused or sampled
cartridges, including blanks, should not indicate clumping of the silica gel sorbent which is
indicative of water condensation inside the cartridge sorbent bed. Physical damage to cartridges
such as cracks in the housing, broken inlet or outlet fittings, or openings into the sorbent bed are
pathways for the ingress of contamination. Cartridges that indicate such damage cannot be used
in the PAMS Program, or if already used for sample collection, are to be voided and a make-up
sample should be collected per Section 3.3.2.1, where possible.
5.5.4 Cartridge Shelf Life
DNPH cartridges that are commercially purchased typically are provided with an expiration from
the manufacturer specifying storage conditions. Agencies are to comply with the manufacturer
expiration, if given. Degradation of the DNPH reagent or silica gel sorbent bed which may
reduce collection efficiency to unacceptable levels may occur after the assigned expiration date.
Additionally, as DNPH cartridge media age, their levels of background contamination are likely
to have increased, perhaps to unacceptable levels, due to passive sampling and uptake from the
ambient atmosphere. For cartridges not assigned an expiration date or assigned an arbitrary
expiration date (i.e., six months from time of receipt) by the manufacturer, agencies should work
within this expiration period as practical. For such cartridges which have exceeded the arbitrary
expiration period, they may be shown to be acceptable if, by performing another lot background
assessment as described in Section 5.5.1, levels of contaminants meet the criteria in Table 5-2
and there remains sufficient DNPH to conduct sampling and ensure excess DNPH levels remain
following sample collection. This level of DNPH on unsampled cartridges is recommended to be
a reduction of DNPH area counts of no more than -15% from the original lot acceptance
analysis. Note that both the Supelco S10 and Waters WAT037500 cartridges are impregnated
with 1 mg of DNPH per cartridge, therefore a 15% reduction would indicate that 0.85 mg of
DNPH are still available for sample collection.
5.6 Carbonyls Method Detection Limits
MDLs for carbonyls are to be determined prior to the use of the method for reporting PAMS
Required Site data and minimally annually thereafter by following the procedures in this section.
To ensure that the variability of the media and the extraction process are characterized in the
MDL procedure, separate cartridges are to be spiked and extracted (it does not suffice to simply
analyze a low-concentration solution of derivatized carbonyls). The following section provides
specific details on selecting a spiking concentration, procedures, and calculations for determining
MDLs per the MUR. 4
91
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
5.6.1 Carbonyls MDL Procedure
All steps performed in the preparation and analysis of field sample cartridges (such as dilution of
extracts) are to be included in the MDL procedure. Cartridges should be spiked and the solvent
permitted to dry prior to extraction. The MDL process includes the following steps:
• Determining the spiking concentration for each target analyte
• Preparing spiking solutions and acquiring cartridge media
• Preparing and extracting minimally three separate batches of spikes and blanks
• Analyzing the spike and blank extracts in minimally three separate batches
• Calculating a separate MDL each for the blanks and spikes
• Assigning the laboratory MDL for each target analyte
5.6.1.1 Selecting a Spiking Level
The first step is to select a spiking level for each target analyte for preparing the MDL spiked
samples. If too low of a spiking level is chosen, the analyte may not be reliably detected by the
instrument. If too high of a spiking level is chosen, the variability of the method near the actual
limits of detection may not be properly characterized. An appropriate spiking level may be
selected by considering the following (in order of importance):
1. The concentration at which the instrument S:N ratio is three- to five-fold for the
analyte.
2. The concentration at which qualitative identification criteria for the analyte are lost
(note that this will be approximately the concentration determined from the MDL
process absent of blank contamination and is typically when the S:N ratio falls below
3:1).
3. Analysis of a suite of blank samples - calculate the standard deviation of the
measured concentration and multiply by 3.
4. Previously acceptable MDL studies and related experience.
Note that the MDL spiking level should not be within the calibration curve; rather, the MDL
spiking level should be less than the lowest calibration standard to best approximate the MDL.
Concentrations within the calibration curve are required to meet method precision and bias
acceptance criteria and are of a high enough concentration that qualitative identification is
certain.
The MDL procedure involves spiking of standards, which imparts additional variability to the
MDL determination. Laboratories should be employing pipettes or syringes which have been
calibrated and demonstrated to meet the accuracy and precision specifications for volumetric
deliveries, which minimizes the variability in the spiking portion of the MDL determination.
Prepare the analytes to be spiked at the appropriate concentration in a single cocktail in
acetonitrile to target a spike volume of approximately 50 |iL.
92
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
5.6.1.2 Preparing MDL Spikes and Blanks
A minimum of seven separate spiked samples and seven separate method blanks are to be
prepared in matrix over the course of a minimum of three different preparation batches. A batch
is defined as a group of standard samples prepared on one day, therefore three different
preparation batches would require preparation on three separate days. To properly characterize
the variability in preparation, the dates of preparation should be spread out such that they are not
consecutive.
The following is to be taken into consideration during preparation of the MDL samples for
carbonyls:
1. Spiked samples are to be prepared in matrix (DNPH cartridge).
2. Selection of media should include as much variety as possible (e.g., individual DNPH
cartridges selected from different boxes or lots) to best characterize the variability of
the method attributable to the use of media representative of field-collected samples.
3. Blanks or blank media which do not meet cleanliness criteria for a given analyte
should trigger root cause analysis to determine the source of the contamination and
should not be used to determine the method blank portion of the MDL. For DNPH
cartridges, media background levels should meet the criteria specified in Method
TO-11A (duplicated in Table 5-2).
5.6.1.3 Extraction and Analysis of MDL Spikes and Blanks
Extraction of the blanks and spikes is to be similarly conducted over the course of three different
extraction batches where each batch occurs on a separate day. Once extracted, each spiked and
blank sample is analyzed only once over minimally three separate analytical batches, where an
analytical batch is defined as a group of samples analyzed on one day. Spreading the preparation
and analysis over multiple preparation batches and across analysis days is intended to
incorporate the variability of both sample preparation and analytical instrumentation that occurs
over time. QC criteria for the analysis are to be met (blanks, continuing calibration checks,
secondary source quality control standards, LCS, calibration checks, etc.). It is preferable to
determine an MDL that is representative of the laboratory's capability than to have an
unrealistically low MDL determined by selecting the best sampling media and attempting to
generate the lowest MDL value possible.
5.6.1.4 MDL Calculation
After all spikes and blanks are analyzed, two MDL values are calculated, one MDL for the
spiked samples according to the convention in 40 CFR Part 136 Appendix B (MDLsp) and one
MDL for the method blanks which includes the media background contribution (MDLb).
Perform all MDL calculations in the final units applicable to the method (e.g., ppbC or |ig/m3).
To calculate the MDL of the spiked samples, MDLsp:
1. Following acquisition of the concentration data for each of the seven or more spiked
samples, calculate the standard deviation of the calculated concentrations for the spiked
samples (%>). Include all replicates unless a technically justified reason can be cited
93
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
(faulty injection, power glitch, etc.), or if a result can be statistically excluded as an
outlier.
2. Calculate the MDL for the spiked samples (MDLsp) by multiplying .vsp by the one-sided
99th percentile Student's t value corresponding to the number of spikes analyzed
according to Table 3-3. Other values of the t-statistic for additional samples (n > 34) may
be found in standard statistical tables.
MDLsp = Ssp-t
3. Compare the resulting calculated MDLsp value to the nominal spiked amount. The
nominal spiked level should be greater than MDLsp and less than 10-fold MDLsp,
otherwise the determination of MDLsp should be repeated with an adjusted spiking
concentration. For MDLsp values greater than the nominal spike level, the MDL spiking
level should be adjusted higher by approximately two or three-fold. For nominal spike
levels which are greater than 10-fold the MDLsp, the MDL spiking level should be
adjusted lower by approximately two or three-fold.
To calculate the MDL of the method blanks, MDLb:
• If none of the method blanks provide a numerical result for the analyte, the MDLb does
not apply. A numerical result includes both positive and negative results for analytes
which are positively identified. Non-numeric values such as "ND" would result when the
analyte is not positively identified. Only method blanks that meet the specified qualitative
criteria for identification (S:N, etc.) are to be given a numerical result.
• If the method blank pool includes a combination of non-numeric (ND) and numeric
values, set the MDLb to equal the highest of the method blank results. If more than 100
method blank results are available for the analyte, set the MDLb to the level that is no less
than the 99th percentile of the method blanks. In other words, for n method blanks where
n> 100, rank order the concentrations. The value of the 99th percentile concentration
(n-0.99) is the MDLb. For example, to determine MDLb from a set of 129 method blanks
where the highest ranked method blank concentrations are ... 1.10, 1.15, 1.62, 1.63, and
2.16, the 99th percentile concentration is the 128th value (129 0.99 = 127.7, which rounds
to 128), or 1.63. Alternatively, spreadsheet programs may be employed to interpolate the
MDLb more precisely.
• If all concentration values for the method blank pool are numeric values, calculate the
MDLb as follows:
a. Calculate the average concentration of the method blanks (xb). If Xb < 0, let Xb = 0.
b. Calculate the standard deviation of the method blank concentrations, 5b.
c. Multiply sb by the one-sided 99th percentile Student's t value corresponding to the
number of blanks analyzed according to Table 3-3. Other values of T for
additional samples (n > 34) may be found in standard statistical tables.
94
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
d. Calculate MDLb as the sum of Xb and the product of sb and the associated
Student's t value:
MDLb = Xb + svt
Compare MDLsp and MDLb. The higher of the two values is reported as the MDL for the given
analyte.
If the MDL is determined as the MDLsp, the determined MDL should be verified by:
a. Preparing one or more spiked samples at one- to five-fold the determined MDL
and analyzing the sample per the method to ensure the determined MDL is
reasonable. Recall that at the MDLsp concentration there is a 50% chance that the
analyte will not be detected; however, the analyte should be detected at two- to
five-fold the determined MDL.
b. Comparing the measured values to reasonable acceptance criteria for the MDL
verification. For example, an MDL verification that recovers 2% of the nominal
amount is not realistic, nor is one that recovers 300%. Appropriate acceptance
limits are to double the acceptance window prescribed by the method for the
given analyte. For example, TO-11A normally permits formaldehyde LCS
recoveries to be 80 to 120% (± 20% error), therefore doubling the MDL
verification acceptance limits would permit 60 to 140% recovery. Note that
agencies may develop alternate acceptance criteria through control charts or other
similar tools. For methods with a significant background contamination, blank
subtraction may be necessary to evaluate the recovery of the MDL verification
sample.
c. Examining the MDL procedure for reasonableness if the verification sample is
outside of the laboratory-defined acceptance criteria. Such an examination might
include investigating the S:N ratio of the analyte response in the spiked samples,
comparing the MDL to existing instrument detection limits (if known - discussed
below), and relying on analyst experience and expertise to evaluate the MDL
procedure and select a different spiking level. The MDL study should then be
repeated with a different spiking level.
Note that the following instrument detection limit (IDL) procedure is not required.
Troubleshooting may include determination of the IDL to evaluate whether the
poor or elevated recovery is due to the instrument. The IDL is determined by
analyzing seven or more aliquots of a standard, bypassing sample preparation or
conditioning (such as spiking on media and subsequent extraction), calculating
the standard deviation of the measurements, and multiplying the standard
deviation by the appropriate student's T value. IDL samples are to be prepared in
the same matrix as calibration standards and are not processed through sample
collection media as is done for MDL spiked samples.
5.6.1.5 Ongoing Determination of MDLs
An efficient method to determine the MDL (once the MDL is initially established) following this
convention is to measure an MDL sample on a continuous basis over the course of several weeks
95
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
or months. In this scenario, one spike (or up to three) would be measured with each
extraction/analysis batch periodically and after seven or more data points have been collected for
the MDL spikes and for the associated method blanks (which are analyzed routinely as ongoing
QC), the MDL could be calculated. This alleviates the need to dedicate a significant contiguous
block of time to preparing and analyzing MDL samples and method blanks.
5.6.2 Example Carbonyls MDL Scenario and Calculation
A laboratory is determining the MDL for formaldehyde by TO-11A by spiking commercially-
prepared DNPH cartridges. The analyst spiked eight cartridges with formaldehyde-DNPH at
0.030 |ig/cartridge (in terms of the amount of the free formaldehyde) over three separate
preparation batches. These eight spiked cartridges and eight additional method blank cartridges
were extracted over three different dates. Results were analyzed over three different analysis
batches per Method TO-11A yielding the results in Table 5-3.
Table 5-3. Example Carbonyls MDL Determination
Mel hod
('si rl ridge
IVepiii'iilion liiitch
Aiiiilvsis liiitch
Spikes
lihinks
Nil 111 her
iiiul Diite
:ind l):tte
(fig/csirlriilge)
(fig/csirlridge)
1
A - September 12, 2015
QR9 - September 13
0.1685
0.1412
2
A - September 12, 2015
QR9 - September 13
0.1651
0.1399
3
A - September 12, 2015
QR9 - September 13
0.1701
0.1402
4
B - September 19, 2015
QR12 - September 21
0.1673
0.1405
5
B - September 19, 2015
QR12 - September 21
0.1692
0.1408
6
C - September 28, 2015
QR16 - September 29
0.1686
0.1403
7
C - September 28, 2015
QR16 - September 29
0.1705
0.1402
8
C - September 28, 2015
QR16 - September 29
0.1696
0.1410
The average (x) and standard deviation (5) of measured formaldehyde mass were determined for
both the spikes and the method blanks (all in units of |ig/cartridge):
Xsp = 0.1686
Xb = 0.1405
= 0.0017
5b = 0.0004
To calculate the MDLsp, the standard deviation of the spiked aliquots is multiplied by the
associated Student's t-statistic. The 99th percentile Student's t value for eight aliquots is 2.998,
corresponding to seven degrees of freedom (8-1=7):
MDLsp = 0.0017 |ig/cartridge • 2.998
= 0.0051 |ig/cartridge
96
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
The MDLsp is subsequently verified to be less than the spike level, and the spike level is
confirmed to be less than 10-fold the MDLsp:
MDLsp < spike level < 10-fold MDLsp
0.0051 |ig/cartridge < 0.030 |ig/cartridge < 0.051 |ig/cartridge
Observe that the determined MDLsp is less than the background level offormaldehyde (xb =
0.1405 ng/cartridge) attributed to the DNPH cartridge media; such indicates that the MDLsp is
biased low and that background levels MUST be taken into account.
To calculate the MDLb, the standard deviation of the blank measurements is multiplied by the
associated student's T and this product is added to the average blank value, Xb:
MDLb = 0.0004 |ig/cartridge • 2.998 + 0.1405 |ig/cartridge
= 0.1417 |ig/cartridge
The MDLsp and MDLb are compared to determine which is greater, and the greater of the two
values is reported as the laboratory MDL for the specific analyte.
0.1417 |ig/cartridge > 0.0051 |ig/cartridge
In this case, the formaldehyde MDLb of 0.1417 |ig/cartridge is greater than the MDLsp of 0.0051
|ig/cartridge, and is reported as the laboratory MDL for formaldehyde as measured by Method
TO-11A.
This value of 0.1417 |ig/cartridge is then normalized per the collected air sample volume. For
sites collecting 8-hour samples at 1 L/minute, the collected sample volume is 480 L or 0.480 m3.
0.1417 ug/cartridge = 0.295 |ig/m3
0.480 mVcartridge
5.7 Carbonyls Sample Collection Equipment, Certification, and Maintenance
Carbonyls are collected by drawing the ambient atmosphere through a DNPH cartridge at a
known flow rate of approximately 0.25 to 1.25 L/minute over the 8-hour collection period. An
ongoing EPA funded study has found that collection efficiency (= measured
concentration/challenge concentration * 100%) did not appreciably vary across this flow rate
range and was greater than 65% over an 8-hour collection interval at aldehyde concentrations of
-1.25 ppbv. (Carbonyls whose performance was assessed included formaldehyde, acetaldehyde,
propionaldehyde, and benzaldehyde. Collection efficiencies were found to be most strongly
associated with the relative humidity of the sampled atmosphere.)5 Collection of samples with
flow rates of approximately 1 L/minute represents an appropriate compromise between
maximizing collection efficiency and sensitivity.
97
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
5.7.1 Sampling Equipment
The sampling units specified for PAMS control flow rate with an MFC. MFCs provide real-time
control of a specified flow, adjusting for changes in backpressure and sampling conditions.
Additionally, MFC flow data may be continuously captured and recorded so as to permit
calculation of a total sampled volume.
At the time of this document's publication, there was a limited selection of carbonyls sampling
units capable of collecting three sequential samples unattended. The sampling unit chosen should
minimally include the following options:
• Elapsed time indicator
• Multi-day event control device (timer)
• Sequential sampling for a minimum of three consecutive samples
• MFC to control sampling flow
• Ozone denuder
Each sampling unit should be flow calibrated annually bracketing the sample collection flows
and shown to be free of positive bias.
5.7.1.1 Sampling Unit Zero Check (Positive Bias Check)
It is strongly recommended that prior to field deployment and minimally annually thereafter each
carbonyl sampling unit be certified to be free of positive bias by collection over 8 to 24 hours of
a sample of humidified hydrocarbon-free (HCF) zero air (or equivalent carbonyl- and oxidant-
free air) or UHP nitrogen. Each channel of each carbonyl sampling instrument employed for
sample collection should be so verified. A best practice is to perform this procedure TTP where
the entire in-situ sampling train is tested. As many agencies do not possess the resources to
perform TTP procedures, the zero check may be performed in the laboratory where as much of
the flow path as possible should be included. Minimally the portion of the flow path comprising
the ozone denuder/scrubber and sampling unit into which the DNPH cartridge is installed should
be verified as non-biasing. The positive bias check should be performed following the recharge
or replacement of the ozone scrubber/denuder, is ideally performed following the annual
recalibration of the flow control device, and ideally includes the length of tubing that connects
the instrument to the manifold or the entire new or cleaned inlet probe.
A recommended zero check procedure is described below. For agencies that cannot perform the
annual maintenance (ozone scrubber/denuder recharge, flow control calibration) and challenge
in-house, manufacturers, the national contract laboratory, or third-party laboratories can perform
this service. Regardless of the exact procedure adopted, when performed, the performance
specifications listed below should be met.
The zero check is performed by simultaneously providing humidified (50 to 70% RH)
hydrocarbon- and oxidant-free zero air or UHP nitrogen to the sampling unit for collection onto a
cartridge and to a separate reference cartridge connected directly to the supplied zero gas source.
98
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
As closely as possible, sample collection parameters for the ozone scrubber/denuder, flow rate,
etc., should mimic those for field sample collections.
The humidified zero gas flow is provided to a challenge manifold constructed of
chromatographic stainless steel (which may include portions of PTFE Teflon®). The manifold
should include three additional ports for connections to the sampling unit inlet, reference sample,
and a rotameter which serves as a vent to ensure that the manifold remains at ambient pressure
during sample collection. The reference sampling flow is set to approximate the flow rate of the
sampling unit with an MFC, mechanical flow device, or needle valve downstream from the
reference cartridge. Humidified zero gas is supplied such that there is excess flow to the
manifold as indicated by the rotameter on the vent port. Sampling is performed for minimally 8
hours to simulate real world sample collection conditions, into the reference cartridge and
through the sampling unit and into the zero challenge cartridge.
Another method to provide the sampling unit with carbonyl-free gas is to install a DNPH
sampling cartridge on the inlet to the sampling unit. This cartridge traps the carbonyl compounds
and effectively replaces the zero gas source. A zero challenge cartridge collected in this manner
should be compared to a field blank (FB) as the reference cartridge.
Analysis for target compounds in the zero challenge cartridge should show that each compound
is < 0.2 |ig/cartridge greater than the reference cartridge. Comparison to the reference cartridge
permits evaluating the contribution of the sampling unit irrespective of cartridge background
contamination. Where exceedances are noted for the zero challenge cartridge, corrective action
should be taken to remove the contamination attributable to the sampling unit and the sampling
unit zero challenge should be repeated to ensure criteria are met before sampling may be
conducted.
5.7.1.2 Carbonyls Sampling Unit Flow Calibration
Prior to field deployment and whenever an independent flow verification indicates the flow
tolerance has been exceeded, the MFC should be calibrated against a calibrated flow transfer
standard and the flow control device adjusted to match the transfer standard (or the regression
characterizing its response is to be reset to match the transfer standard).
Note that manufacturer procedures for calibration may be followed if flows can be calibrated at
standard conditions. A suitable calibration procedure for MFCs is as follows. The sampling unit
pump(s) and MFC should be warmed up and run for approximately five minutes to ensure the
MFC is stable. A blank DNPH cartridge should be installed into the air sampler to provide a
pressure drop to the pump, and airflow through the cartridge commenced. The calibrated flow
transfer standard should be connected at the upstream end of the sampling unit so as much of the
flow path is included as possible in order to identify potential leaks in the flow path that may not
otherwise be evident. MFC calibration should be performed at minimally three flow rates: the
typical flow rate for sample collection, approximately 30% less than the typical flow of sample
collection, and approximately 30% higher than the typical flow of sample collection. Particular
attention should be paid to ensure that the correct calibration conditions are compared - that both
the reading on the flow transfer standard and MFC are in standard (25°C and 760 mm Hg)
conditions.
99
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
5.7.1.3 Moisture Management
Humidity plays several roles with regard to sample collection. Water vapor can condense on
interior portions of the sample flow path, potentially resulting in a low measurement bias due to
carbonyls dissolving in the liquid water. To minimize the condensation of liquid water onto the
interior surfaces of the flow path, the ozone scrubber is maintained at a minimum of 50°C.
Additionally, connecting tubing comprising the flow path may be insulated to maintain the
elevated temperature and discourage condensation. High humidity in sampled atmospheres may
also lead to somewhat lower carbonyl collection efficiencies due to the possible back reaction of
the DNPH-carbonyl derivative with water to form the free carbonyl. The reverse reaction is less
likely for aldehydes due to their higher reactivity, however, can lead to lower collection
efficiencies for ketones.6
5.7.2 Sampling Train Configuration and Presample Purge
The carbonyl sampling inlet probe may be standalone or connected to a manifold inlet. For either
configuration, components comprising the wetted surfaces of the flow path must be constructed
of borosilicate glass, PTFE Teflon®, or chromatographic grade stainless steel. Chromatographic
stainless steel tubing includes that with interiors coated with a fused silica lining. Due to the
reactivity of materials such as copper or adsorptive/desorptive properties of materials such as
FEP Teflon®, rubber, or plastic tubing, these materials should not be utilized within the flow
path. It should be noted that the ozone denuder will typically consist of copper tubing coated
with KI. At the time this document was published, there has not been a definitive study to
investigate whether there is a reduction in collection efficiency when sampled air contacts
uncoated portions of the copper tubing. While not required, operators may consider employing
chromatographic grade stainless steel, instead of copper, tubing for preparing a Kl-coated
denuder.
For sites having a common inlet manifold shared with gaseous criteria pollutant monitors and an
auto-GC, the manifold must be constructed of borosilicate glass. 40 CFR Part 58 Appendix E
Section 9(a) states that inlets for these reactive gas parameters can only be Pyrex® (borosilicate
glass) or Teflon®. Since Teflon® is not appropriate for VOCs, borosilicate glass is the only
acceptable manifold material. A bypass pump is connected to the manifold to continuously pull
ambient air though the manifold. The flow rate of the bypass pump should be minimally double
the total maximum sampling load for all sampling units and instruments connected to the
manifold to ensure a constant supply of fresh ambient air is available for sampling. Where the
carbonyls sampling unit has its own standalone inlet probe separate from the manifold, no
additional bypass pump is necessary.
Regardless of how the ambient air is introduced into the sampling instrument, it is strongly
recommended that the inlet line to the sampling unit be purged with ambient air such that the
equivalent of a minimum of 10 air changes is completed just prior to commencing collection of
the day's first sample collection. This purge eliminates stagnant air and flushes the inlet line.
100
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
5.7.3 Carbonyl Sampling Inlet Maintenance
Over time, the carbonyl inlet probe and connecting tubing will become laden with particulate
residue. This particulate residue may scrub target analytes from the gas stream and may act as
sites for adsorption/desorption. Wetted surfaces of inlet probes and connecting tubing are
strongly recommended to be cleaned and/or replaced minimally annually (e.g., prior to PAMS
season), and preferably every six months, particularly if operated in an urban environment where
there is a higher concentration of particulate matter (PM).
Only deionized water should be used to clean inlet lines. If the lines are short enough, a small
brush can be used in concert with the deionized water to effectively clean the interior of the
tubing. It may be more effective to simply replace the tubing on a prescribed basis. Many
carbonyl sampling units utilize Teflon® particulate filters upstream of the denuder to alleviate
particulate loading of internal parts (valves and MFCs) of sampling units. Such particulate filters
are to be replaced periodically, recommended at the beginning of each PAMS season. Sites with
high particulate concentrations should inspect the filter after the first month of use and replace
the filter if heavy particulate loading is evident.
5.8 Sample Collection Procedures and Field Quality Control Samples
5.8.1 Sample Collection Procedures
Prior to beginning sample collection, all DNPH cartridge lot characterization should have been
completed as described in Section 5.5.1. The sampling unit should have passed the zero bias
check in the previous 12 months, the sampling inlet line should have been cleaned or replaced in
the previous 12 months, the flow control device should have been calibrated within the past 12
months, and, if so equipped, the particulate filter should have been changed prior to PAMS
season.
In addition to the procedures described below, all cartridges should be handled as prescribed in
Section 5.5.2.
5.8.1.1 Sample Setup
Blank DNPH cartridge media are transported to the site in a cooler on ice packs where they are
either stored on site in a refrigerator or freezer (with calibrated temperature monitoring) or
installed into the sampling unit for sample collection. Note that freezing does not affect sample
integrity.
Appropriate blank, non-exposed DNPH cartridge(s) are installed into the sampling unit and the
sample collection program verified to comply with Section 5.8.1.3. The flow rate of collection
should be set to a known calibrated flow rate of approximately 0.7 to 1.25 L/minute (at standard
conditions) for a total collection volume of 0.34 to 0.6 m3. Method sensitivity is linearly
proportional to the total collection volume, and the latter should be adjusted within the specified
range so that MDL MQOs are attained.
101
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
For sampling units that permit a leak check function on the sample pathway, a leak check should
be initiated prior to sample collection. At publishing of this document, only one of the known
commercially-available sequential sampling units included a leak check function. This
instrument closes the valves to the sampling port and/or channel and evacuates the flow path and
senses if there is flow for 30 seconds. If the flow exceeds 0.03 L/minute the instrument logs an
error flag to the sampling data. The sampling unit automatically performs a leak check on all
channels or ports programmed for sampling and reports an error flag for the failing channel or
port. The other commercially-available sampling instrument known at the time of publishing of
this document does not include a leak check function. Leak check routines may be performed on
other carbonyls sampling instruments by pressurizing or evacuating the system and observing a
pressure change or flow equivalent to < 0.03 L/minute.
The initial flow rate, date and time of sample initiation, and cartridge identification information
should be recorded on the sample collection form.
5.8.1.2 Sample Retrieval
The collected cartridges are to be retrieved and placed into cold storage as soon as possible after
the conclusion of sampling in order to minimize degradation of the carbonyl-DNPH derivatives.
Sample retrieval should occur the next day, if possible, but should not 72 hours of the end of the
third sequential daily sample (note that to have the sampling unit readied for the next sampling
event, sample cartridges will need to be retrieved within 48 hours of the end of sample
collection). If elevated shelter temperatures are anticipated, samples should be retrieved as soon
as possible to limit the impact to collected samples. The ending flow rate, total flow (if given),
and sample duration is to be documented on the sample collection form for each of the three
sequential samples. The cartridges are removed from the sampling unit, the caps installed on the
inlet and outlet of each cartridge, each cartridge sealed in its separate foil pouch, and the pouches
immediately placed in cold storage. A best practice to minimize contamination is to transport the
sealed foil pouch in an outer zip-lock bag containing activated carbon. The sample is to be kept
cold during shipment such that the temperature remains < 4°C, and the temperature of the
shipment is to be determined upon receipt at the laboratory. Note that samples delivered directly
to a laboratory within a few hours (and not shipped overnight) may not be stored refrigerated a
duration sufficient to reach < 4°C.
Sampling units that incorporate computer control of the sampling event with associated data
logging may provide the sample collection information (flow rates, elapsed sample time, total
collected volume, etc.) which should be transcribed to or printed and attached to the sample
collection form. Note that the sampling unit recorded data may also be downloaded and
maintained; however, due to the potential loss of electronic data on a flash drive, monitoring
agencies are cautioned against relying solely on the electronic data. For such sampling units, the
data logged should be reviewed to ensure the sample was collected appropriately and there are
no flags or other collection problems that may invalidate the collected sample. Electronic sample
collection data should be downloaded and provided to the analytical laboratory, when possible.
The sample custody form is to be completed and accompany the collected sample at all times
until relinquished to the laboratory. COC documentation should comply with Section 5.8.1.4.
102
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
5.8.1.3 Sampling Schedule and Duration
Sample collection is to occur every third day according to the national sampling calendar
(https://www3.epa.gOY/ttn/amtic/calendar.html). Three sequential 8-hour samples are to be
collected on each sampling day, according to the following time schedule, standard local time,
unadjusted for daylight savings time:
04:00 to 12:00 p.m. (noon)
12:00 p.m. (noon) to 20:00
20:00 to 04:00
Valid samples will have been collected for 8 hours ± 20 minutes and commence within 15
minutes of the scheduled start time. For missed or invalidated samples, a make-up sample set
should be scheduled and collected per Section 3.3.2.1. Clock timers controlling sampling unit
operation are to be adjusted so that digital timers are within ±5 minutes of the reference time
(cellular phone, global positioning system, or similar accurate clock).
5.8.1.4 Carbonyls Sample Chain of Custody
Sample custody procedures are required to avoid misplacement of samples or confusion of one
sample with another, and to provide documentation to assist in detection and resolution of COC
problems and instances where data are called into question. A sample is considered to be in
custody if it is in one's actual physical possession or stored in a secured area restricted to
authorized personnel.
Blank cartridge media may originate at the analysis laboratory; therefore, COC procedures may
be prescribed by the analysis laboratory. Regardless of the origin of the new cartridge media,
each sample cartridge, whether an ambient sample or field QC sample (such as a trip blank, field
blank, or exposure blank) will be listed on a COC form documenting the transfer of the sample
cartridges from their origin, through collection, and transport to and receipt by the analysis
laboratory. The following information is to minimally be recorded on the COC form:
• Origin of cartridges (e.g., analysis laboratory or field office)
• Transfer of cartridges between individuals - dates, times, and signatures of individuals
relinquishing and receiving cartridges
o Relinquishing cartridges to site operator (either by handoff or shipment with
courier)
o Receipt of cartridges by site operator
o Relinquishing of sampled cartridges by site operator following retrieval (for
handoff to analysis laboratory or shipment with courier)
Note: Shipping couriers are not expected to sign COC forms. The individual
relinquishing the samples to the shipper/courier will indicate
relinquishment to the shipper/courier on the COC form. Custody is
presumed to be with the courier until received at the laboratory.
o Receipt of sampled cartridges by analysis laboratory
103
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
• Unique identifier(s) for each sample, sample collection date(s), and site(s) location
information
• Storage of cartridges at each point during transfer between individuals, including during
shipment
o Storage at the monitoring site (e.g., stored at < 4°C in onsite refrigerator, etc.)
o Shipping conditions (e.g., on ice packs) and associated information for tracking
or evaluating the shipping conditions - such as thermometers placed in a
shipping cooler
o Upon receipt at the laboratory - document thermometer used for measuring
shipment temperature
Note that the convention for recording custody information for the samples can include recording
transfers and storage on the field collection data sheet; however, it may be more convenient to
include a separate COC form for each shipment that encompasses all samples in the shipment. A
separate dedicated COC form including all associated samples in the shipment reduces the
number of instances where staff transferring cartridge custody are required to sign.
Laboratory sample custodians should ensure that sample custody documentation is complete and
should contact site operators, as appropriate, to complete missing information. A sample COC
for PAMS carbonyls samples is included in Appendix C of the PAMS Required Site National
QAPP.
5.8.2 Field Quality Control Samples
QC samples co-collected with field samples include field, trip, and exposure blanks, collocated
and duplicate samples, field matrix spikes, and breakthrough samples. Blank cartridges provide
information on the potential for field-collected samples to be subjected to positive bias, whereas
spiked cartridges assess the potential for the presence of both positive and negative bias.
5.8.2.1 Field Blanks and Exposure Blanks
Field blanks should be collected twice per month. Field blanks should be handled in the same
manner as all other field-collected samples, transported in the same cooler and stored in the same
refrigerator/freezer storage units. Survey of monitoring agencies has shown that field blanks had
previously been collected by numerous protocols, each characterizing different aspects of
potential contamination. In order to standardize the characterization of associated ambient
sample contamination due to installation in the sampling unit, a field blank is defined as follows:
• Field blanks are installed into one of the sampling positions (in which ambient samples
are installed);
• Field blanks are installed in the sampling unit for approximately 5 to 10 minutes, then the
cartridge is capped, sealed into the foil pouch, and stored refrigerated;
• No air is drawn through the field blank cartridge;
• The field blank travels with the associated ambient samples.
104
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
Collection of the field blank in this manner characterizes the handling of the blank cartridge in a
sampling position in the sampling unit.
An exposure blank is similar to a field blank, but is not required, and may be collected via
several protocols, which may characterize contamination from numerous combinations of
exposure and handling procedures. The exposure blank includes opening the cartridge pouch,
removing the caps exposing the cartridge to the ambient atmosphere briefly, and exposing it to
the temperature conditions of the primary sampling cartridge for the same duration as the co-
collected field samples. Like a field blank, air is not drawn through the exposure blank cartridge.
Some sampling units have a dedicated "field blank" channel for installation of the exposure
blank through which air is not permitted to flow. For multi-channel sampling units, the exposure
blank may be installed in a channel which is not activated for sample flow. For sampling units
which have neither a dedicated blank channel nor unused channel available on the sampling unit,
the exposure blank cartridge may be removed from the foil pouch, installed in the sampling unit
for five to ten minutes, the cartridge uninstalled and the end caps reinstalled, and the cartridge
placed near the sampling unit for the duration the primary sample is installed in the sampling
unit.
Field blanks and exposure blanks may passively sample ambient air throughout the time of
exposure, and as a result may have somewhat higher background levels as compared to lot
blanks, trip blanks, or laboratory method blanks. Field blanks and exposure blanks should meet
the following criteria listed in Table 5-4.
Table 5-4. Carbonyls Field Blank Acceptance Criteria
( iirhmnl Compound
Acit|>I;iikt l imit (fig/csirlririgi')
Acetaldehyde
<0.40
Formaldehyde
<0.30
Acetone
<0.75
Sum of Other Target Carbonyls
<7.0
Failure to meet the field blank criteria indicates a source of contamination and corrective action
should be taken as soon as possible. For agencies which collect associated trip blanks,
comparison of the field blank to trip blank values may provide meaningful insight regarding the
contamination source. Field-collected samples associated with field blanks which do not meet
these criteria are to be flagged/qualified when input to AQS (refer to AQS qualifiers in Section
11). For field blanks which fail criteria and are collected with each sampling event, the co-
collected field sample results are to be flagged/qualified as "FB" when input to AQS. For failing
field blanks which are collected on a less frequent basis (i.e., bi-weekly basis), field collected
samples since the last acceptable field blank should be flagged/qualified when input to AQS.
Field samples should not be corrected for field blank values. Field blank values should be
reported to AQS so that data users may estimate field and/or background contamination.
5.8.2.2 Trip Blanks
Trip blanks are a useful tool to diagnose potential contamination in the sample collection and
transport of carbonyl samples. Trip blanks are not required but are a best practice. A trip blank
105
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
consists of a blank unopened cartridge which accompanies field sample cartridges at all times to
and from the laboratory. The trip blank cartridge is stored in the same refrigerator/freezer,
transported in the same cooler to and from the site, and kept at ambient conditions during sample
collection. The cartridge should remain sealed in the foil pouch and not removed from its pouch
until extracted in the laboratory.
Background levels on the trip blank should be comparable to the lot blank average determined as
in Section 5.5.1 and should not exceed the values listed in Table 5-2. Exceedance of these
thresholds should prompt corrective action and the results of the associated field-collected
samples should be appropriately qualified when input to AQS.
5.8.2.3 Collocated Samples
Collocated sampling is described in detail in Section 5.3.1.1. Where such is performed, it should
minimally be done at a frequency of 10%, meaning approximately one collocated sample every
month.
Following extraction and analysis the collocated cartridge results are compared to evaluate
precision. Precision should be < 20% RPD for results >0.5 |ig/cartridge. Root cause analysis
should be performed for instances in which collocated samples fail this precision specification
and the results for both the primary and collocated samples should be qualified as "QX" when
entered into AQS.
5.8.2.4 Duplicate Samples
Duplicate sampling is described in detail in Section 5.3.1.2. Where such is performed, it should
minimally be done at a frequency of 10%, meaning approximately one duplicate sample every
month.
Following extraction and analysis the duplicate cartridge results are compared to evaluate
precision. Precision should be < 20% RPD for results >0.5 |ig/cartridge. Root cause analysis
should be performed for instances in which duplicate samples fail this precision specification and
the primary and duplicate results should be qualified as "QX" when entered into AQS.
5.8.2.5 Field Matrix Spikes
Performance of field matrix spiked sample collection is a best practice but is not required. Field
matrix spikes are prepared by spiking a blank DNPH cartridge with a known amount of analyte
(either derivatized or underivatized) prior to dispatching to the field for collection. The field
matrix spike is handled identically to field samples; sample storage, transport, and extraction are
identical. Field matrix spiked samples are collected concurrently with a non-spiked primary
sample as a duplicate sample per Section 5.3.1.2 via duplicate channel or split sample flow.
The primary field sample and matrix spiked sample analysis results are evaluated for spike
recovery based on the amount spiked prior to shipment to the field as follows:
%Recovery =
(Field Matrix Spike Result — Primary Sample Result)
¦ 100
Nominal Spiked Amount
106
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
Spike recovery should be within ± 20% (80 to 120% recovery) of the nominal spiked amount. In
the event of an exceedance, root cause analysis should be performed to determine sources of
negative or positive bias, as needed, for example, sources of contamination or reasons for the
loss of analyte. High recoveries may indicate contamination in the matrix spike sample collection
channel or loss in the primary sample collection channel. Low recoveries may indicate a poorly
functioning ozone denuder, which permits ozone to pass through the sample collection flow path
and degrade the spiked analytes.
5.8.2.6 Breakthrough Samples
While not required, collection of breakthrough samples may be performed. A breakthrough
sample is a second DNPH cartridge connected immediately downstream of the primary sample
cartridge. Periodic collection of breakthrough samples provides a level of assurance that the
primary sample cartridge is efficiently trapping target carbonyls. For sites conducting
breakthrough sampling the recommended frequency is once per month which should be
described in the agency PAMS QAPP, SOP, or similar controlled document.
Note that this breakthrough cartridge will increase the pressure drop in the sampling system and
may require an adjustment in the operation of the sampling unit to achieve the desired flow rate.
Breakthrough sample results should meet the field blank criteria listed in Table 5-4.
5.9 Carbonyls Extraction and Analysis
Target carbonyls collected on the DNPH cartridges are extracted and analyzed per EPA
Compendium Method TO-11 A2 according to the following guidance. Note that use of a UHPLC
is acceptable provided the performance and QC criteria listed in Table 5-5 are met.
5.9.1 Analytical Interferences and Contamination
5.9.1.1 Analytical Interferences
The carbonyl-hydrazone derivatives are separated with a HPLC or UHPLC system and are
typically detected at 360 nm with a photodiode array or similar detector operating at UV
wavelengths. Identification is based on retention time matching with known standards. Mass
spectrometer and photodiode array (PDA) detectors are also an option if more definitive
identification and quantification are desired or required. Minimally, analysis by HPLC-UV is
performed.
Interferences from co-eluting peaks may result from hydrazones formed by co-collected
compounds or reactions with co-collected compounds which form artifacts. Such co-eluting
peaks may form as dimers or trimers of acrolein or be the result of chemical reactions with
nitrogen oxides. Target analyte peaks which indicate shoulders, tailing, or inflection points
should be investigated to ensure these chromatographic problems are not related to a co-eluting
interference.
5.9.1.2 Labware Cleaning
Labware is to be thoroughly cleaned prior to use to eliminate potential interferences and
contamination. Regardless of the specific procedures implemented, all method performance
107
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
specifications for cleanliness are to be met. Volumetric labware used for collection of cartridge
eluent can show buildup of silica gel residue over time, requiring aggressive physical cleaning
methods with laboratory detergent and hot water. Clean all associated labware by rinsing with
ACN, washing with laboratory detergent, rinsing with deionized water, rinsing with ACN or
methanol, and air drying or drying in an oven at no more than 80 to 90°C. 8 Heated drying of
volumetric labware at temperatures > 90°C is unnecessary and may void the manufacturer
volumetric certification.
5.9.1.3 Minimizing Sources of Contamination
Several target analytes in this method are typically present in ambient air and may contaminate
solvents and the DNPH reagent if appropriate preventive measures are not in place. ACN used
for sample extraction, standards preparation, and mobile phase preparation should be carbonyl-
free HPLC grade or better (as indicated by the supplier or on the COA) and should be stored
tightly capped away from sources of carbonyls. DNPH cartridges should be handled properly per
Section 5.5.2.
Laboratories that process environmental samples for organic compounds such as pesticides
typically extract with acetone or other solvents which may contaminate DNPH cartridge media
and carbonyl extraction solvents. Laboratory areas in which cartridges are stored, extracted, and
analyzed should be free of contaminating solvent fumes. Carbonyl handling areas should have
HVAC systems separate from such laboratory operations.
5.9.2 Reagents and Standard Materials
5.9.2.1 Solvents
Solvents used for extraction, preparation of standards solutions, and preparation of mobile phase
are to be high-purity carbonyl-free, HPLC grade, and shown by analysis to be free of
contaminants and interferences. Such solvents include ACN, methanol, and deionized water.
Deionized water should be ASTM Type I (18 MQcm).
5.9.2.2 Calibration Stock Materials
Calibration source material should be of known high purity and should be accompanied by a
COA. Calibration materials should be neat high purity solids or sourced as certified single
component or component mixtures of target compounds in an appropriate solvent (i.e., ACN or
methanol).
Neat solid material should be weighed with a calibrated analytical balance with the appropriate
sensitivity for a minimum of three significant figures in the determined standard mass. The
calibration of the balance should be verified on the day of use with certified weights bracketing
the masses to be weighed. Calibration standards diluted from stock standards should be prepared
by delivering stock volumes with mechanical pipettes or calibrated gastight syringes and the
volumes dispensed into Class A volumetric labware to which the diluent (ACN) is added to
establish a known final dilution volume.
108
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
5.9.2.3 Secondary Source Calibration Verification Stock Materials
A secondary source standard should be prepared to verify the calibration of the HPLC or
UHPLC on an ongoing basis, minimally immediately following each ICAL. The secondary
source stock standard should be purchased from a different supplier than the calibration stock
material or, only if unavailable from a different supplier, may be of a different lot from the same
supplier as the calibration material.
5.9.2.4 Holding Time and Storage Requirements
Unopened stock materials are appropriate for use until their expiration date provided they are
stored per manufacturer requirements. Once opened, stock materials may not be used past the
manufacturer recommended period or, if no time period is specified, not beyond six months from
the opened date. To use the standard materials past this time period, standards should have been
demonstrated to not be degraded or concentrated by comparison to freshly opened standards.
Unopened stock materials should be stored per manufacturer recommendations. All stock and
diluted working calibration standards should be stored at < 4°C in a refrigeration unit separate
from sample cartridges and sample extracts.
5.9.3 Cartridge Holding Time and Storage Requirements
All field-collected cartridges should be stored at < 4°C and extracted within 14 days of the end of
collection. These conditions similarly apply to laboratory-prepared QC samples, which are to be
stored at < 4°C and extracted within 14 days of preparation (note that freezing does not adversely
impact sample integrity). Extracts are to be analyzed within 30 days of extraction. For sample
results exceeding these holding times or exceeding the storage temperature, they should be
appropriately qualified when input to AQS ("HT" for failure to meet the holding time and "TT"
for storage temperature exceedance). Note that the "HT" QA qualifier relates to exceeding the
sample retrieval period and the "TT" QA qualifier relates to the exceedance of the sample
transport temperature. The qualifier related to exceeding the holding time or transport
temperature is "TS", which is a Null qualifier that does not permit the user to enter a
concentration value. At the time of publication of this TAD, additional QA qualifiers were being
added to AQS to include a QA qualifier for exceeding holding time. As additional AQS
qualifiers are available, these will be communicated to the PAMS workgroup.
5.9.4 Cartridge Extraction
5.9.4.1 Laboratory Extraction Batch Quality Control Samples
With each extraction batch of 20 or fewer field-collected cartridges, which may include the
various field QC samples such as those listed in Section 5.8.2, the following negative and
positive laboratory QC samples should be prepared (except LCS/LCSD which should be
prepared/analyzed minimally monthly - recommended with each batch). For batch sizes of more
than 20 field-collected cartridges, n such QC samples of each type should be added to the batch,
where n = batch size / 20, and where n is rounded to the next highest integer. Thus, for batch
sizes of 30, two of each of the following QC samples would be included in each batch. A best
practice would be to process field-collected cartridges in batches of no more than 20 at a time.
• Extraction Solvent Method Blank (ESMB): An ESMB is a negative control sample
prepared by transferring the extraction solvent into a flask just as an extracted sample.
109
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
The purpose of this negative control is to demonstrate that the extraction solvent is free of
interferences and contamination and that the labware washing procedure is effective.
Analysis should show target compound responses are less than the laboratory MDLsp.
• Method Blank (MB): The MB is a negative control that may also be referred to as the
cartridge blank. The MB is a blank unopened cartridge (that has not left the laboratory)
which is extracted identically to field samples. Target analytes should be less than MDL.
• Laboratory Control Sample (LCS): The LCS, also referred to as the laboratory fortified
blank, is a positive control prepared by spiking a known amount of underivatized or
derivatized DNPH-carbonyl target analyte onto a cartridge such that the expected extract
concentration is in the lower third of the calibration range. The spiked cartridge is
allowed to sit for minimally 30 minutes to allow the solvent to dry following addition of
the DNPH-carbonyl in solution. The LCS is then extracted with the same extraction
solvent and method employed for field samples to assess bias in matrix of the extraction
and analysis procedures. Recovery of the LCS should be within 80 to 120% of nominal
for formaldehyde and 70 to 130% of nominal for all other target carbonyls.
• Laboratory Control Sample Duplicate (LCSD): The LCSD is prepared and extracted
identically to the LCS. The LCSD assesses precision through extraction and analysis.
Recovery of the LCSD should be within 80 to 120% of nominal for formaldehyde and 70
to 130%) for all other target carbonyls. The LCS and LCSD results should show RPD of
< 20%.
All field-collected and laboratory QC samples in a given extraction batch should be analyzed in
the same analysis batch (an analysis batch is defined as all samples analyzed together within a
24-hour period).
Laboratories are to take corrective action to determine the root cause of laboratory QC
exceedances. Field-collected sample results associated with failing QC results (in the same
preparation batch or analysis batch) should be appropriately qualified as "QX" when input into
AQS. In order to simplify troubleshooting when experiencing QC failures, QC sample cartridge
media and extraction solvent lots should be the same, where possible.
5.9.4.2 Cartridge Extraction Procedures
Cartridges are extracted with carbonyl-free HPLC grade ACN. Field-collected and stored QC
sample cartridges should be removed from cold storage and allowed to equilibrate to room
temperature, approximately 30 minutes, prior to extraction. Cartridges are removed from the foil
pouch, the end caps are removed, and the cartridges are installed in a holding rack with the inlet
of the cartridge pointed down to facilitate elution. Field-collected samples and associated field
and laboratory QC samples discussed in Section 5.9.4.1 should be extracted in the same batch.
The ACN extraction solvent is added to the cartridge so that elution occurs in the direction
opposite of sample air flow (unless the laboratory can demonstrate that reverse elution is not
necessary). Luer syringe barrels or other commercially-available funnels are available for use as
solvent reservoirs for extraction, if needed. Elution may be performed by gravity or vacuum
methods. The cartridge eluent is collected in a clean volumetric flask or other appropriate
110
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
volumetrically certified vessel. Once the eluent is collected, the extract is brought to a known
final volume with ACN extraction solvent.
A minimum 2-mL extraction volume is necessary to ensure complete elution of the target
analytes from the sorbent bed. An extraction volume up to 5 mL may be employed; however,
larger volumes do not increase the extraction efficiency and may overly dilute the extract.
Once brought to volume, it is highly recommended that an aliquot of the extract is transferred to
an autosampler vial for analysis and the remaining extract stored in a sealed vial protected from
light at < 4°C. The stored extract affords reanalysis if there are problems during analysis (up to
40 days from extraction).
5.9.5 Analysis by HPLC
5.9.5.1 Instrumentation Specifications
For separation of the DNPH-carbonyls by HPLC or UHPLC, the analytical system should have
the following components:
• Separations module capable of precise pumping of ACN, methanol, and/or deionized
water at 1 to 2 mL/min
• Analytical column, C18 reversed phase, 4.6 x 50-mm, particle size 2.7-|im, pore size
90-A, or equivalent
• Guard column
• Absorbance detector set to 360 nm or mass selective detector capable of scanning m/z
range of 25 to 600
• Column heater capable of maintaining 25 to 35 ± 1°C
• Degassing unit
5.9.5.2 Initial Calibration
On each day that analysis is performed, the instrument will be calibrated (meaning an ICAL
should be performed) or the ICAL will be verified by analysis of a CCV according to the
following guidance.
ICAL of the HPLC or UHPLC is performed initially, when continuing calibration checks fail
criteria, and when there are major changes to the instrument that affect the response of the
instrument. Such changes include, but are not limited to: change of guard or analytical column (if
analyte retention times change), backflushing of the analytical column (if analyte retention times
change), replacement of pump mixing valves and/or seals (if analyte retention times change),
replacement of the detector and/or lamp, and cleaning of the mass spectrometer (MS) source (if
HPLC/MS).
Working calibration standards are prepared in ACN at concentrations covering the desired
working range of the detector, typically from 0.01 to 3.0 |ag/m L of the free carbonyl. In order to
avoid confusion or error in concentration calculation, it is recommended that all concentrations
111
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
be expressed as the free carbonyl and not the DNPH-carbonyl. The ICAL should consist of a
minimum of five calibration standard levels which cover the entire calibration range.
Prior to calibrating the HPLC, the instrument is warmed up and mobile phase should be pumped
for a time sufficient to establish a stable baseline. All solutions to be analyzed are removed from
cold storage and equilibrated to room temperature prior to analysis.
Once a stable baseline is established, minimally one solvent blank (SB; an aliquot of extraction
solvent dispensed directly into a vial suitable for the HPLC autosampler, or similar) should be
analyzed to demonstrate the instrument is sufficiently clean, after which analysis of calibration
standard solutions may commence. The SB should show target compound responses are less than
the laboratory MDLsp.
To establish the ICAL, each standard solution is injected minimally once and preferably in
triplicate. The instrument response (area units) is plotted on the y-axis against the nominal
concentration on the x-axis and the calibration curve generated by linear regression for each
target compound. The calibration curve correlation coefficient (r) will be > 0.999 for linear fit
and the curve should not be forced through the origin. The calculated concentration of each
calibration solution should be within 20% of its nominal concentration.
The absolute value of the concentration equivalent to the intercept of the calibration curve
(|intercept/slope|) converted to concentration units (by division by the slope) should not exceed
the laboratory MDLsp. When this specification is not met, the source of error (likely
contamination or suppression) should be corrected and the calibration curve re-established before
sample analysis may commence.
RT windows are calculated from the ICAL by determining the mean RT for each target
compound. For positive identification the RT of a derivatized carbonyl should be within three
standard deviations (3.s) or ± 2%, whichever is smaller, of its mean RT from the ICAL. Note that
heating the column to a constant temperature of approximately 25 to 30°C promotes consistent
RT response by minimization of column temperature fluctuations.
5.9.5.3 Secondary Source Calibration Verification Standard
Following each successful ICAL, a SSCV is analyzed to verify the accuracy of the ICAL. The
SSCV is prepared in ACN at approximately the mid-range of the calibration curve by dilution of
the secondary source stock standard. Alternatively, two or more concentrations of SSCV may be
prepared covering the calibration range. All SSCVs should recover within ± 15% of nominal.
5.9.5.4 Continuing Calibration Verification
Once the HPLC has met ICAL criteria and the ICAL verified by the SSCV, a CCV is to be
analyzed prior to the analysis of samples on days when an ICAL is not performed, and minimally
every 12 hours of analysis. The CCV is also recommended to be analyzed after every 10 sample
injections and at the end of the analytical sequence. On days when an ICAL is not performed, a
SB should be analyzed prior to the CCV to demonstrate the instrument is sufficiently clean to
commence analysis.
112
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
At a minimum, a CCV is to be prepared at a single concentration recommended to be at
approximately the mid-range or lower end of the calibration curve, should be diluted from the
primary stock or secondary source stock material, and CCV recovery should be 85 to 115% for
each target compound. As a best practice, two or more concentrations of CCV may be prepared
and analyzed so as to better cover instrument performance across the range of the calibration
curve.
Corrective action should be taken to address CCV failures, including, but not limited to,
preparing and analyzing a new CCV, changing the guard or analytical column, backflushing of
the analytical column, replacement of the detector and/or lamp (if HPLC/UV), and cleaning of
the MS source (if HPLC/MS).
5.9.5.5 Replicate Analysis
For each analytical sequence of 20 or fewer field-collected samples, at least one field-collected
sample extract should be selected for replicate analysis. For sequences containing more than 20
field-collected samples, n such replicates should be analyzed, where n = batch size / 20, and
where n is rounded to the next highest integer. Thus, for batch sizes of 30, two replicate analyses
would be performed. Replicate analysis should demonstrate precision of < 10% RPD for
concentrations > 0.5 |ig/cartridge.
5.9.5.6 Compound Identification
The following criteria are to be met in order to positively identify a target compound:
1. The S:N ratio of the target compound peak is > 3:1, preferably
>5:1. Refer to Section 4.2.4.2 for more information on S:N.
2. The RT of the compound is within the acceptable RT window determined from the
ICAL average (see Section 5.9.5.2).
3. **HPLC-MS only ** - The target and qualifier ion peaks are co-maximized (peak
apexes within one scan of each other) - as discussed in the following paragraphs and
shown in Figure 5-2.
4. **HPLC-MS only ** - The abundance ratio of the qualifier ion response to target ion
response for at least one qualifier ion is within ± 30% (on a relative basis) of the
average ratio from the ICAL. These are discussed in the following paragraphs and
shown in Figure 5-2.
Item 1 above does not need to be evaluated closely with each identified peak. Rather the
interpretation of the experienced analyst should weigh heavily on whether the peak meets the
minimal S:N. Item 2 above may be automated by the analysis software such that it is
automatically flagged. RT windows are updated with each new ICAL. Note that the CDS may
overlook peaks outside the designated RT window and that analysts should manually examine
chromatograms for RT shifts resulting in missed identifications.
Refer to Figure 5-2 for an example of the qualitative identification criteria listed above and the
following discussion. The RT is within the retention time window defined by the method (red
box A), and the abundance ratios of the qualifier ions are within 30% of the ICAL average ratio
113
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
(red box B). The S:N of the peak is shown to be greater than 5:1 (red oval C) and the target and
qualifier ion peaks are co-maximized (dashed purple line D).
If any of these criteria (as applicable) are not met, the compound may not be positively
identified. The only exception to this is when in the opinion of an experienced analyst the
compound is positively identified. The rationale for such an exception should be documented.
nfcur.cance Scan 10431? 660 ir.nv C3191404 Dlaafa '>1032; <•) ai*
84 niethvlen 40 oO SO TooTJo UoTIo TSC 20022C240260MO30C
A!>ur-oance
49
Aaundartc*
30000-
JOODO
"50
d4
0*rr»| mr] r-nyj.«?yyr
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
additional chromatographic peaks as a result of formation of side products from the consumption
of the DNPH. This verification can be estimated and should be prescribed within the SOP or
similar controlled document. Once sample identification is confirmed, field-collected samples
are to be qualified as estimated concentrations (qualified as "LL" signifying a potential low bias)
when entered into AQS since depletion of the DNPH to below 50% of typical levels indicates the
potential for negative bias in the measured concentrations.
The concentrations of target carbonyls in unknown samples are calculated by relating the area
response of the target carbonyl to the relationship derived in the calibration curve generated in
Section 5.9.5.2.
Sample extracts with concentration results exceeding the instrument calibration range should be
diluted and analyzed such that the peak responses are within the calibration range. The result
corrected for dilution is then reported and the associated MDL adjusted accordingly by the
dilution factor (the MDL multiplied by the dilution factor).
While TO-11A allows for blank subtraction, this is not an acceptable practice and results should
not be corrected for SB, MB, or FB levels. Concentrations exceeding acceptance criteria for
these blanks should prompt investigation as to the source of contamination and associated field
collected sample results may require qualification as estimates.
For sampling units which do not provide an integrated collection volume, the beginning and
ending flows are averaged to calculate the collected air volume. For computer-controlled
sampling units, the integrated collected volume is typically available from the data logging
system. Sampled air volumes are to be in standard conditions of temperature and pressure (STP),
25°C and 760 mm Hg. Sampling unit flows should be calibrated in flows at standard conditions
so conversion from local conditions to standard flows is not necessary.
The air concentration in |ig/m3 of each target carbonyl is determined by multiplying the
concentration in the extract by the final extract volume and dividing by the collected sample air
volume at standard conditions of 25°C and 760 mm Hg:
r _ CfVe
A ~ ~V
VA
where:
Ca = concentration of the target carbonyl in air (|ig/m3)
Ct = concentration of the target carbonyl in the extract (|ig/mL)
Ve = final volume of extract (mL)
Va = volume of collected air at STP (m3)
Carbonyl concentrations can also be calculated in ppbv using a conversion factor based on the
molecular weight of the target carbonyl at STP:
115
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
MW
CF =
0.082059-298.15
where:
CF = conversion factor (|ig m"3 ppbv"1)
MW = molecular weight of the target carbonyl (g/mol)
The air concentration of the target carbonyl in ppbv is then calculated as follows:
c
A,ppb £p
where:
CA,PPb = concentration of the target carbonyl in air (ppbv)
Ca = concentration of the target carbonyl in air (|ig/m3)
CF = conversion factor (|igm"3ppbv"1)
The air concentration of the target carbonyl in ppbC is then calculated by multiplying the
concentration in ppbv by the number of carbon atoms in the molecule:
QA,ppbC — 0A,ppb " Nc
where:
CA,PPb = concentration of the target carbonyl in air (ppbv)
Nc = number of carbon atoms in the molecule
5.10 Summary of Quality Control Parameters
A summary of QC parameters is shown in Table 5-5.
116
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
Table 5-5. Summary of Quality Control Parameters for Carbonyls Analysis
Piii'iiiiKMcr
Description ;iihI Dciiiils
Kc(|iiiml I-'iv(|ik*iio
Acit|>I;iiht ( rilcriii
Holding Times
Maximum duration from end of
sample collection for sample
extraction
Maximum duration from
sample extraction to analysis
All field-collected and
laboratory QC cartridges
14 days from end of sample
collection to extraction
30 days from sample
extraction to analysis
Solvent Blank
(SB)
Aliquot of ACN analyzed to
demonstrate instrument is
sufficiently clean to begin
analysis
Prior to ICAL and daily
beginning CCV
All target carbonyls < MDLsp
Initial Calibration
(ICAL)
Analysis of a minimum of five
calibration levels covering
approximately 0.01 to 3.0
lig/mL
Initially, following failed
CCV, or when changes to the
instrument affect calibration
response
Linear regression
r > 0.999, the concentration
of each target carbonyl at
each calibration level should
be within ± 20% of nominal;
intercept/slope < MDLsp
Second Source
Calibration
Verification
(SSCV)
Analysis of a second source
standard at the mid-range of the
calibration curve to verify
curve accuracy
Immediately following each
ICAL
Recovery of each target
carbonyl within
± 15% of nominal
Continuing
Calibration
Verification
(CCV)
Analysis of a known standard
at the mid-range of the
calibration curve to verify
ongoing instrument calibration
Prior to sample analysis on
days when an ICAL is not
performed, and minimally
every 12 hours of analysis.
Recommended following
every 10 sample injections,
and at the conclusion of each
analytical sequence
Recovery of each target
carbonyl within
± 15% of nominal
Extraction Solvent
Method Blank
(ESMB)
Aliquot of extraction solvent
analyzed to demonstrate
extraction solvent and labware
are free of interferences and
contamination
One with every extraction
batch of 20 or fewer samples,
at a frequency of no less than
5%
All target carbonyls
< MDLsp
Method Blank
(MB)
Unexposed DNPH cartridge
extracted as a sample
One with every extraction
batch of 20 or fewer samples,
at a frequency of no less than
5%
All target carbonyls < MDL
Laboratory
Control Sample
(LCS)
DNPH cartridge spiked with
known amount of target analyte
at approximately the lower
third of the calibration curve
Minimally quarterly.
Recommended: One with
every extraction batch of 20
or fewer samples, at a
frequency of no less than 5%
Formaldehyde recovery 80-
120% of nominal spike
All other target carbonyls
should recover 70-130% of
nominal spike
Laboratory
Control Sample
Duplicate (LCSD)
Duplicate LCS to evaluate
precision through extraction
and analysis
Minimally quarterly.
Recommended: One with
every extraction batch of 20
or fewer samples, at a
frequency of no less than 5%
Should meet LCS recovery
criteria
Precision < 20% RPD of LCS
117
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
Table 5-5 (continued). Summary of Quality Control Parameters for Carbonyls Analysis
Piii'iiiiHMcr
Description ;nnl Dcliiils
Kc(|iiiml I'it(|ucik\\
Accept mice Ci'ilci'iii
keplicale Anal\ bib
Replicate anal\ bib of a liekl-
collected sample
Once w illi e\ er\ anal} bib
sequence of 20 or fewer
samples, at a frequency of no
less than 5% (as required by
workplan)
Precibion _ 10%RPD for
concentrations
>0.5 |ig/cartridgc
Retention Time
(RT)
RT of each target compound in
each standard, QC sample, and
unknown sample
All qualitatively identified
compounds
Each target carbonyl within
± 35 or ± 2% of its mean
ICALRT
Lot Blank
Evaluation
Determination of the
background of the DNPH
cartridge media
Minimum of 3 cartridges or
1% (whichever is greater) for
each new lot of DNPH
cartridge media
All cartridges should meet
criteria in Table 5-2
Zero Certification
Challenge
Clean humidified gas sample
collected over 24 hours to
demonstrate the sampling unit
does not impart positive bias
Annually prior to PAMS
season
Each target carbonyl in the
zero certification < 0.25
|ig/m3 above reference
sample
Field Blank
Blank DNPH cartridge exposed
to field conditions for
minimally 5 minutes in the
primary sampling location
Twice per month
Should meet criteria in Table
5-4
Duplicate Sample
Field sample collected through
the same inlet probe as the
primary sample
10% of primary samples for
sites performing duplicate
sample collection (as detailed
in ANP and/or PAMS QAPP)
Precision < 20% RPD of
primary sample in-air
concentration for
concentrations
>0.5 |ig/cartridge
Collocated Sample
Field sample collected through
a separate inlet probe from the
primary sample
10% of primary samples for
sites performing duplicate
sample collection (as detailed
in ANP and/or PAMS QAPP)
Precision < 20% RPD of
primary sample in-air
concentration for
concentrations
>0.5 |ig/cartridgc
118
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
5.11 References
1. Carter, W.P. L., Development of the SAPRC-07 Chemical Mechanism. Atmospheric
Environment 2010, 44, (40), 5324-5335.
2. Determination of Formaldehyde in Ambient Air using Adsorbent Cartridges Followed by
High Performance Liquid Chromatography (HPLC) [Active Sampling Methodology]; EPA
Compendium Method TO-11A; U.S. Environmental Protection Agency: January 1999.
Available at (accessed March 2018): https://www3 .epa.gov/ttnarnti 1/files/ambient/airtox/to-
pdf
3. MacGregor, I. C., Hanft, E. A., and Shelow, D. M. (2015). Update on the Optimization of
U.S. EPA Method TO-11A for the Measurement of Carbonyls in Ambient Air. Paper
presented at the National Ambient Air Monitoring Conference, St. Louis, MO, August 10,
2016.
4. Definition and Procedure for the Determination of the Method Detection Limit, Revision.
EPA Office of Water, EPA 821-R-16-006, December 2016. Available at (accessed March
2018): https://www.epa.eov/sites/production/flles/2' ;uments/mdl-
procedure rev2 12-13 -2.016. pdf
5. MacGregor, I.C., Seay, B.A., Skomrock, N.D., McCauley, M.W., Turner, D.J., Mullins,
L.A., Tomcik, D.J., Saeger, C. "Optimization of US EPA Method TO-11A for the
Measurement of Carbonyls in Ambient Air." Presentation given at the National Ambient Air
Monitoring Conference, August 15, 2018. Available at
https://proiects.ere.com/conferences/ambientair/confl8/MacGTeeoi' Ian AirToxics 8-
i" ^00 Saloni FOST 508.pdf
6. Urone, P., & Ross, R. C. (1979). Pressure change effects on rotameter air flow rates.
Environmental Science & Technology, 13(6), 732-734. doi: 10.1021/es60154a003
7. Steven Sai Hang Ho, Ho Sai Simon Ip, Kin Fai Ho, Wen-Ting Dai, Junji Cao, Louisa Pan
Ting Ng. "Technical Note: Concerns on the Use of Ozone Scrubbers for Gaseous Carbonyl
Measurement by DNPH-Coated Silica Gel Cartridge." Aerosol and Air Quality Research, 13:
1151-1160, 2013.
8. Care and Safe Handling of Laboratory Glassware. Corning Incorporated. RG-CI-101-REV2.
2011. Available at (accessed March 2018):
http://csmedia2.cornine.com/LifeSciences/media/pdf/Care and Safe Handli ) Glassw-
areKG-
119
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
6.0 OXIDES OF NITROGEN
Each agency is to prescribe in an appropriate QS document, such as an SOP or equivalent, its
procedures for measurement of true NO2. EPA has developed a national SOP for the analysis of
true NO2 that monitoring agencies can adopt or use as a starting point for their approved SOP.
Various requirements and best practices for true NO2 measurement are given in this section.
Oxides of nitrogen are released from emission sources primarily as nitric oxide (NO) and to a
lesser extent as nitrogen dioxide (NO2). The two species are collectively termed NOx (NOx = NO
+ NO2). Through atmospheric processes, NOx is converted to many other inorganic and organic
nitrogen oxides, such as nitrous acid (HONO), nitric acid (HNO3), and peroxyacetyl nitrate
(H3CC(0)00N02, PAN), which when taken together are named NOz. The total of all reactive
nitrogen species in ambient air is called NOy, which is the sum of NOx and NOz (i.e., NOy = NOx
+ NOz). Determining NO2, NO, and NOy concentrations in ambient air is useful in understanding
nitrogen oxide emission patterns and temporal trends, and in assessing the photochemical age
and reactivity of air masses. NO2 plays a critical role in the photochemical production of O3, as
shown in the simplified series of reactions, below (Reactions A through E). Accurate
measurement of NO2 concentrations in ambient air is necessary to support air quality modeling
efforts aimed at evaluating and tracking the progress of control strategies for attaining the ozone
NAAQS.
NO + HO2 —> OH + NO2 Reaction A
OH + VOCs (+O2) —» —» RO2 Reaction B
RO2 + NO —> RO + NO2 Reaction C
NO2 + hv —» NO + O Reaction D
O + O2 —^ O3 Reaction E
It is important to note that NOx and HOx (HOx = OH + HO2) are not consumed in the process of
O3 production, and are available to continue to generate ozone in the presence of VOCs and
sunlight. However, other reactions that convert NOx to NOz, such as those represented by
Reaction F and Reaction G, below, can remove NOx from the photochemical O3 production
cycle.
NO + RO2 —> RONO2 Reaction F
NO2 + OH —> HNO3 Reaction G
Organic nitrates, such as PAN and other peroxyacyl nitrates, are formed from the reaction of
NO2 with VOCs, are important as carriers for NOy into rural regions, and cause ozone formation
in the global troposphere. This is shown as the thermal equilibrium between peroxyacetyl
radicals (PA) and PAN in Reaction H and Reaction I, below.
120
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
N02 + CH3C(0)00 (PA) -> CH3C(0)00N02 (PAN)
Reaction H
CH3C(0)00N02 (PAN) -> NO2 + CH3C(0)00 (PA)
Reaction I
All Required PAMS Sites will monitor for true NO2 in addition to NOy using continuous
monitoring equipment, as described in more detail in the following sections. Measurements will
be conducted continuously and hourly averages reported for every day during the sampling
period.
6.1 NO/NOy
Measurements of NO/NOy are required atNCore stations and the guidance and acceptance
criteria for their measurement are addressed within the NCore Precursor TAD 1 and the EPA QA
Handbook2 and are not addressed in this TAD.
6.2 True NO2
The term "true NO2" refers to a collection of measurement techniques that provide more
selective detection of NO2 compared to heated bed chemiluminescent analyzers that have
traditionally been used for NOx measurements. One such method is a photolytic
chemiluminescent analyzer - a U.S. EPA FEM. Like the conventional chemiluminescent
technique, the method can only directly measure NO. The conventional heated molybdenum
converter is replaced with a more specific photolytic converter, resulting in a more selective
measurement of NO2. Other commercially-available methods offer direct detection of NO2 and
include CAPS and CRDS technologies. These methods employ laser light at a specific
wavelength to probe NO2 absorption to determine the NO2 concentration. The CAPS instrument
method is also a designated FEM. The general sampling and analytical methods of the two FEM
techniques (i.e., photolytic chemiluminescence and CAPS) are described in more detail the
following sections. At the time of preparation of this document, FEM CRDS technologies were
not commercially available and are therefore not addressed in this document.
PAMS Required Sites are expected to utilize an FEM instrument for measuring true NO2.
6.2.1 Photolytic Conversion Chemiluminescent Detection NO2 Instruments
The photolytic conversion chemiluminescent detection NO2 method (refer to Figure 6-1)
measures NO and NO2 using the conventional chemiluminescence signal that is produced from
reaction of NO with added 03. This method employs the chemiluminescent detection of NO as in
traditional NOx analyzers, which indirectly measure NO2 and NOx species by conversion to NO.
However, replacing the heated molybdenum bed converter with a photolysis cell provides greater
selectivity for the NO2 reaction channel such that other nitrogen containing compounds such as
HN03 and PAN are not also converted to NO. Thus, the measured NO2 value is indicative of the
true concentration of NO2 in the sampled air stream and is not subject to the interferences caused
by the presence of NOz. The photolysis of NO2 to NO is shown in Reaction J:
NO2 + hv —» NO + O (k~ 400 nm) Reaction J
121
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
Use of a high-power light source maximizes conversion of NO2 to NO, though the conversion
efficiency is typically less than unity (< 100%) and the possibility of negative spikes in NO2
remains during fast-changing ambient NOx conditions. The instrument switches back and forth
from the NO2 channel to the NO channel and calculates NO2 by difference. The negative spikes
are due to quick transient changes in NO2 concentrations and the instrument operating with a
single NO measurement detector. Since the instrument cannot measure the NO2 and NO
channels simultaneously, quick drastic decreases in NO2 concentration can occur between
switches between channels, resulting in negative concentrations when the NO reference
concentration exceeds the converted NO2 concentration.
03 generator
photolvtic-chemiluminescence
NO channel
Figure 6-1. Schematic Diagram of Photolytic Chemiluminescence NO2/NO/NOX FEM 3
At the time of this document's publication, commercially-available photolytic conversion
chemiluminescence analyzers are available from Teledyne API, which carries the Model
T200UP Trace-level True N02/N0/N0x Analyzer (U.S. EPA FEM EQNA 0512-200). The
Model T200UP uses a light-emitting diode (LED) array to selectively photolyze NO2 to NO with
little interference from other gases, as reported by the vendor. Although not independently
verified, conversion efficiencies are reported by Teledyne API to be similar to that of a heated
molybdenum bed converter under typical ambient NO2 conditions. In an independent
intercomparison between the photolytic chemiluminescence FEM and heated molybdenum bed
chemiluminescence FRM methods, side-by-side ambient air measurements were conducted at
two locations (Visalia, California and Research Triangle Park [RTP], North Carolina) in winter
and summer seasons, respectively.4 Results of linear regression analysis of the photolytic NO2
instrument versus the heated bed FRM for the two locations are summarized in Table 6-1.
Table 6-1. Ambient Air Intercomparison Results for NO2 FEM (photolytic conversion)
versus FRM (molybdenum bed conversion) Method Reported by Beaver et al., 2013
Location
Season
Ambient NO2
Range (ppb)
Slope
Intercept
(PPb)
Coefficient of
determination
(r2)
Visalia, CA
Winter
0-60
0.89
3.07
0.78
RTP, NC
Summer
0-20
1.04
-0.79
0.99
122
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
Agreement between the two methods appears to depend on ambient conditions, with the FRM
overpredicting NO2 compared to the photolytic NO2 instrument at the California site, where
higher NO2 levels were observed. The two methods were also less correlated than expected at the
California site (r2 = 0.78) compared to the North Carolina site (r2 = 0.99) due to NOz
interferences in the molybdenum converter FRM.
6.2.2 Cavity Attenuated Phase Shift (CAPS) Instruments
CAPS is an optical absorption technique that is related to cavity ringdown spectroscopy. The
technique utilizes a modulated broadband incoherent light source (a 430 to 450 nm LED)
coupled to an optically resonant cavity, which consists of two highly reflective mirrors, and a
photodetector. NO2 in the cell causes a phase shift in the signal, that is proportional to the NO2
concentration, and this phase shift is measured by the photodetector. Figure 6-2 shows a
simulation of the squarewave modulated LED light before it enters the cavity and the attenuated
waveform detected after the cavity.
4
>
5-2
Squarewave Before Cavity
1
Time (Seconds)
x 10
2
-4
Ringdown Waveform After Cavity
1
Time (Seconds)
x 10
Figure 6-2. Simulated Squarewave LED Light before the Cavity and Attenuated Phase
Shifted Waveform after Passing through the Cavity5
The advantage of this technique is that it provides a direct NO2 measurement without need for a
converter or additional reagents and has a fast response. However, any compound that absorbs
light at the excitation wavelength (-430 to 450 nm) will cause interference.
There are currently four commercially-available CAPS NO2 monitors available:
• Teledyne API Model T500U CAPS NO2 Analyzer (FEM EQNA-0514-212)
• Environnement S.A. AS32M CAPS NO2 Analyzer (FEM EQNA-1013-210)
• Ecotech Serinus 60 CAPS NO2 Analyzer (FEM EQNA-0217-242)
• Aerodyne Research, Inc. CAPS NO2 Monitor
The Teledyne T500U CAPS analyzer was evaluated at the RTP, North Carolina site in the
intercomparison study described above.6 Linear regression analysis of the CAPS NO2 values
versus the FRM NO2 resulted in a slope of 0.97, intercept of -0.20 ppb, and r2 of 0.97, showing
that the FRM overpredicted NO2 values compared to the CAPS unit and that the two were highly
correlated. The Teledyne T500U was EPA FEM approved (EQNA-0514-212). Figure 6-3 shows
a schematic diagram of the Aerodyne Research CAPS NO2 monitor.
123
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
Flow
Drier
Vacuum
Photodiode
Pump
LED
Mixer
Pre Amp
Digital Electronics
LOq u t
Figure 6-3. Schematic Diagram of Aerodyne CAPS NO2 Monitor 7
6.2.3 Cavity Ring-down Spectroscopy (CRDS) Instruments
As cavity ring-down spectroscopy instruments have not been approved by EPA as an FEM, these
are not approved for measuring true NO2 for PAMS and will not be further discussed in this
TAD.
6.2.4 True NO2 FEM Instrument Response
The photolytic conversion and CAPS FEM instruments respond quickly to changing NO2
concentrations when compared to heated molybdenum bed conversion chemiluminescent FRM
instruments as evidenced in Figure 6-4 (courtesy of the Missouri Department of Natural
Resources). Here, the FEM instruments (Teledyne API T500U CAPS and Teledyne API T200UP
photolytic conversion) respond more quickly when compared to the FRM (Thermo 42i
molybdenum conversion) at both the span concentration of NO2 (approximately 375 ppb) and at
the level at which the instrument precision is checked (approximately 75 ppb). A qualitative
observation from Figure 6-4 is that the direct-reading CAPS monitor responds the most quickly
to transient concentrations, which is in accord with observations in the EPA Near Road
Monitoring Network which report very fast (-15 second) response times with CAPS
instruments.8
124
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
400
Thermo 42i
350
API T200-UP
(Photolytic)
300
API T500
(Direct)
250
200
150
PRECISION >~
Figure 6-4. NO2 FRM and FEM Response Time 8
6.2.5 Minimizing Bias in NO2 Measurements
Measurement bias can result from incorrect calibration technique, uncalibrated or poorly
calibrated gas metering devices, background due to contaminants in standard gases or diluent
gases, poor instrument hygiene, incompatible materials in standard or sampling flow paths, or
instrument drift. It is important to minimize the influence of these sources of bias to the extent
possible.
Standard gases should be sourced from reputable suppliers, and the sourced standards should
indicate acceptably low levels of contaminants or interferences and should comply with the EPA
Traceability Protocol for Assay and Certification of Gaseous Calibration Standards.9 Several gas
suppliers offer such EPA Protocol Gases for which the certified concentrations are traceable to a
reference material, such as those prepared and certified by the NIST or the Van Swinden
Laboratorium (VSL). COAs for such standards should indicate the traceability to a calibrated
instrument and the associated certified standard employed to calibrate the instrument. In general,
stock standard gases of NO or NO2 require specially-treated cylinders to ensure the standard gas
concentration is stable for the indicated expiration period. For monitoring agencies performing
gas phase titration (GPT) of ozone and NO to generate NO2 calibration gas, the levels of residual
NO2 in the NO cylinder should not exceed 1% of the NO concentration. Such will be listed on
the cylinder COA; however, proper cylinder purging and handling procedures should be
practiced to ensure ambient air entrained in the regulator is purged and is not permitted to
backflow into the cylinder which can result in the formation of NO2 in the cylinder when oxygen
reacts with the NO standard gas. Furthermore, regulators should be purged properly before each
125
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
use, as practical, to remove contaminants such as water. Refer to the guidance beginning on page
5 of Issue 15 of The QA Eye Newsletter from December 2013, available at the following link:
https://www3.epa.eov/ttnamtil/files/ambient/qa/qanew ;
Cylinder regulators, tubing, connecting components, valves, and other portions of the wetted
pathway for transferring stock and diluted gas standards and sampled atmosphere should be of
compatible materials, namely PTFE or chromatographic-grade stainless steel. Incompatible
materials include, but are not limited to: plastic, rubber, brass, and copper. As with the proper
purging of cylinder regulators, gas lines should be properly purged to ensure the pathways have
been properly passivated prior to utilizing measurements in generating calibration responses.
This is typically accomplished by observing instrument response until a stable measurement is
achieved.
MFCs in dynamic dilution calibrators (DDC) used for gas-phase dilution and GPT should be
calibrated at their range of use within the previous 12 months and have the flow calibration
verified quarterly. Monitoring agencies should perform maintenance on the zero air generator(s)
as indicated by the manufacturer recommendations and prescribed in the appropriate SOP.
Instrument maintenance (mirror cleaning, UV lamp replacement, and particulate filter changes)
should be conducted per the manufacturer recommendations and per the appropriate SOP.
Lastly, the instrument zero drift should be monitored and adjusted per the guidance listed in
validation template Appendix D of the EPA QA Handbook, Volume II, January 2017. As of the
publication of this TAD, the allowable zero drift was < ± 3.1 ppb NO2 over 24 hours and < ± 5.1
ppb over 14 days.
6.2.6 Generation of NO2 Standards
NO2 standards can be prepared using either GPT or dilution of a known concentration of a
certified NO2 stock standard gas from a high-pressure cylinder. While both methods are
discussed below, monitoring agencies have reported slow instrument responses and a negative
bias (refer to Section 6.2.6.2) with dilution of NO2 from a high-pressure cylinder.
6.2.6.1 Gas Phase Titration
GPT has been widely employed to generate standard concentrations of NO2 and is described in
detail in the EPA Quality Assurance Handbook, Volume II, Part II, Section 2.3.1 (2002). Briefly,
generation of NO2 concentrations by GPT is performed by providing a known amount of ozone
and providing excess NO. The stoichiometry of the reaction of NO and ozone is such that for
every mole of ozone, one mole of NO2 is produced per the following reaction:
NO + O3 —»NO2 + O2
The NO2 is generated in the mixing area of the DDC and assumes that all ozone is consumed to
generate NO2. Advantages of this method of NO2 standards generation are:
• The NO reagent is stable in high pressure cylinders
126
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
• NO2 contamination in the NO standards is typically a small, negligible percentage (e.g.,
< 1%) and is unlikely to increase with appropriate cylinder hygiene
• Passivation times of gas supply lines are minimized resulting in fast response of
downstream analyzers
The disadvantages to GPT include:
• Inability to generate single-digit ppb or lower concentrations as ozone generators are not
typically stable at such low concentrations
• NO gas cylinders can be easily contaminated
• Requires frequent calibration of the DDC ozone generator
6.2.6.2 Dilution of Standard NO2 Gas
Stock standard NO2 gas may be sourced from reputable gas providers and the gas diluted with
zero air with a DDC to desired concentrations.
A small number of monitoring agencies have reported success with directly diluting calibration-
level concentrations of NO2 with a DDC from a high pressure cylinder. However, monitoring
agencies have generally indicated problems with calibration with such methods. Reports indicate
that the calibration appears to be biased low by several percent (approximately 7 to 13%) when
compared to a calibration performed with GPT and that stabilization time at each concentration
level is extensive, in some instances up to an hour or longer, with the measured concentration
gradually increasing to a plateau as illustrated in Figure 6-5 (courtesy Pinellas County Air
Quality Division). This method provides the ability to generate low concentration (sub-ppb)
standards such as those needed for determining MDLs; however, monitoring agencies should
consider the potential bias and extensive stabilization times for routine calibration use.
HO
Figure 6-5. Calibration of CAPS NO2 Analyzer using NO2 Dilution Method 10
127
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
6.2.7 Calibration of True NO2 Instruments
The calibration of the true NO2 instruments is established by introducing zero blanks and span
levels of an NO2 standard to the instrument at approximately 80% of the full scale of the selected
instrument reporting range. The determined instrument responses are used to set the slope and
offset of the calibration. The zero calibration is performed first and the span calibration point
follows. The span calibration point should be greater than approximately 80% of the expected
measured concentrations. For example, if the full-scale (expected) measurement range is 200
ppb, the span concentration would be approximately 160 ppb.
Once the calibration is established, the instrument operator immediately verifies the calibration
with a multi-point verification (MPV) consisting of analyzing five concentration levels covering
the full-scale range and including a zero (for example, a full-scale range of 0 to 200 ppb, the
points could be: 0, 25, 75, 125, 175 ppb). The resulting measured concentrations are fit to a
linear regression which will show:
• Correlation coefficient, r2 > 0.995
• x-intercept within ±0.2 ppb of origin
• each concentration level is within 10% of the nominal concentration
Corrective action should be taken for failures, which may include performing instrument
maintenance (cleaning mirrors, replacing particulate filters, etc.) followed by recalibrating the
instrument.
6.2.8 True NO2 Sampling
True NO2 measurements shall be conducted continuously every day during the PAMS sampling
period; data shall be reported as hourly averages in ppb. Sampling hours for which less than 45
minutes of measurement are available are considered to be incomplete, are not valid, and are to
be reported to AQS with a null qualifier. For NO2 measurements, the ambient air inlet should be
positioned 2 to 15 m above the ground and at least 1 m in horizontal and vertical distance from
supporting structures. The distance between the inlet and any surrounding trees shall be at least
10 m, and is recommended to be at least 20 m.
6.2.9 Method Detection Limits for Continuous Gaseous Criteria Pollutant Methods
Determination of the MDLs for continuous gaseous monitors, including instruments for true
NO2, ozone, and NO/NOy, is performed according to a similar convention. The MDL is
determined according to the MUR as described in Section 3.3.5.1, where a series of low
concentration standards and blanks is analyzed to establish the instrument variability at low
concentration and the average blank background. These aspects are input into the MDL
procedure to establish the lowest concentration that is distinguishable from background with
99% confidence. Experience has shown that the manufacturer published MDLs detailed in the
FRM/FEM designation for the instruments typically represent an ideal instrument operation and
are unrealistically low in practical terms. The monitoring agency should establish an MDL for
the instruments initially, preferably prior to PAMS season. EPA has convened an MDL
workgroup as part of the PAMS Required Site workgroup to develop guidance for the equipment
128
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
specifications, standard materials, and procedures for determining MDLs for the various true
NO2 FEMs. Note that at the time of this document's publication, the workgroup was in process
and had not published such guidance.
A brief general discussion for determining continuous gaseous monitor MDLs follows.
Prior to determining the MDL, the instrument is shown to be free of interferences and
contamination and an initial calibration is established. Instrument maintenance, such as lamp
replacement, mirror cleaning, and particulate filter replacement should be completed prior to
beginning the MDL process.
The instrument operator conducts the MDL determination by measuring zero air blanks and a
low concentration standard. In order to capture an aspect of temporal variability, the
measurements for the zero blanks and standards should occur over the course of three different
dates, preferably non-consecutive. Once calibrated, the instrument zero drift should not be
adjusted throughout the course of the MDL determination. This allows any variability related to
the instrument drift to be characterized in the MDL determination.
Measurements, activities performed, and equipment used should be documented so that MDL
determinations can be reconstructed.
An example scenario for determining an MDL with calculations is shown in Section 6.2.9.5.
6.2.9.1 Determining the MDLb
Once the instrument is calibrated, the instrument operator introduces zero air to the instrument
and measures the zero air matrix as is done for routine zero checks. The instrument operator
should review the short-term data (e.g., 5-minute or 1-minute data) to ensure the instrument
background is stable and not continuing to decrease. Experienced instrument operators should
use their best judgement to determine that the blank reading is stable. The instrument
measurements of zero air blank data are recorded for minimally seven discrete 20- to 30-minute
periods over the course of minimally three different dates, preferably non-consecutive. The
average is computed for each of these seven measurement periods to generate seven
concentration values. More than seven blanks can be included in the calculation, and, when
included, will typically provide a better approximation of the background contribution. Zero
blanks with technical problems (e.g., power surge detector spikes, missing minute data, or
ambient air leaks) may be excluded from the subsequent calculations; however, a valid technical
reason to exclude data should be documented and justified.
The MDLb is calculated as follows:
1. Calculate the average concentration of the zero blank measurement concentrations,
Xb. If Xb< 0, let Xb = 0.
2. Calculate the standard deviation of the zero blank measurement concentrations, 5b.
3. Multiply sb by the one-sided 99th percentile Student's t value corresponding to the
number of blanks analyzed (refer to Table 3-3). Values of t for additional samples (n
> 34) may be found in standard statistical tables.
129
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
4. Calculate MDLb as the sum of Xb and the product of sb and the associated Student's t
value:
MDLb = Xb+ 5b-t
6.2.9.2 Determining the MDLsp
The most difficult aspect of determining the MDLsp portion of the MDL is generating suitable
low concentration standards. Modern "trace" level analyzers are capable of detecting single-ppb
concentrations of gaseous criteria pollutants, and the ability to generate concentrations in this
range is limited by the starting stock gas concentration and the effective DF of the gas calibrator
used to generate the standards. In general, generating true NO2 concentrations in the range
suitable for determining MDLs (< 1 ppb) is not possible with GPT due to the inability of onboard
ozone generators to reliably produce accurate and stable ozone concentrations < 1 ppb. In such
cases, it may be more practical to source NO2 from a pressurized cylinder to dilute into the
proper range for delivering the standards. As discussed in Section 6.2.6.2, preparation of
standard concentrations by dilution of NO2 from a high pressure cylinder can be difficult as it
may require extensive times for passivating delivery lines and ensuring a stable instrument
response; however, once stabilized, this method of standard gas preparation provides better
control of the challenged concentration and the ability to generate low (sub-ppb) concentrations
reliably.
The first step in determining the MDLsp is selecting a concentration at which to perform the
MDLsp procedure. To select a concentration, instrument operators should consider the following,
in decreasing order of importance:
1. The concentration at which the instrument response is approximately three- to five-fold
the baseline response.
2. Analysis of a suite of zero blanks, such as the measurements recorded determining MDLb
- calculate the standard deviation of the measured concentration and multiply by 3.
3. Previously acceptable MDL studies and related experience.
As practical, this selected concentration would be generated to determine the MDLsp. If the
standards dilution equipment cannot generate a concentration sufficiently low to achieve that
selected, the instrument operator should generate as low as concentration as possible given the
system limitations. This may require use of the highest diluent flow possible combined with the
lowest flow available on the DDC. Recall that the absolute accuracy of the challenged
concentration is not evaluated in the determination and is not as important as the ability to
generate a stable concentration. Therefore, operation of the DDC channels outside the typical
operating range of 10 to 90% full scale will not impact the MDL determination.
Similarly to the MDLb, instrument measurements of the low level concentration standard are
recorded for minimally seven discrete 20- to 30-minute measurement periods over the course of
minimally three different dates, preferably non-consecutive. The instrument operator should
review the short-term data (e.g., 5-minute or 1-minute data) to ensure the instrument reading is
stable and not demonstrating an increasing or decreasing trend. Experienced instrument operators
should use their best judgement to determine that the reading is stable. The average is computed
for each of these seven measurement periods to generate seven concentration values. All
130
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
appropriate data collected should be included unless a valid technical reason exists to exclude the
data. Standard measurement periods with technical problems (e.g., power surge detector spikes,
missing minute data, or ambient air leaks) may be excluded; however, a valid technical reason to
exclude data should be documented and justified.
Calculate the MDLsp as follows:
a. Calculate the standard deviation of the calculated concentrations for the standard
measurement periods (%>).
b. Calculate the MDL for the standard measurement periods (MDLsp) by multiplying .ssp by
the one-sided 99thth percentile Student's t value at 99% confidence corresponding to the
number of measurement periods analyzed according to Table 3-3. Other values of t for
additional samples (n > 34) may be found in standard statistical tables.
MDLsp = Ssp-t
Compare the resulting calculated MDLsp value to the nominal standard level. The nominal spiked
level should be greater than MDLsp and less than 10-fold MDLsp. If this is not the case, the
MDLsp process should be repeated with an adjusted spiking concentration, if possible. For
MDLsp values greater than the nominal spike level, the MDL spiking level should be adjusted
higher by approximately two or three-fold. For nominal spike levels which are greater than the
10-fold the MDLsp, the MDL spiking level should be adjusted lower by approximately two or
three-fold.
6.2.9.3 Calculating and Verifying the Instrument MDL
Compare MDLsp and MDLb. The higher of the two values is reported as the MDL for the given
analyte.
1. If the MDL is determined as the MDLsp, the determined MDL should be verified by:
a. Analyzing one or more standard levels at one- to five-fold of the determined
MDL to ensure the determined MDL is reasonable. Recall that at the MDLsp
concentration there is a 50% chance that the analyte will not be detected;
however, the analyte should be detected at two- to five-fold the determined MDL.
b. Comparing the measured values to reasonable acceptance criteria for the MDL
verification. For example, an MDL verification that recovers 2% of the nominal
amount is not realistic, nor is one that recovers 300%. Appropriate acceptance
limits are to double the acceptance window prescribed by the method for the
given analyte. For example, for ozone measurements, the bias specification is ±7,
therefore 14% may be achievable. Note that agencies may develop alternate
acceptance criteria through control charts or other similar tools. For methods with
a significant background contamination, blank subtraction may be necessary to
evaluate the recovery of the MDL verification standard.
c. Examining the MDL procedure for reasonableness if the verification sample is
outside of the laboratory-defined acceptance criteria. Such an examination might
include investigating the S:N ratio of the analyte response in the standard data and
relying on instrument operator experience and expertise to evaluate the MDL
131
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
procedure and select a different spiking level. The MDL study should then be
repeated with a different spiking level, if possible.
6.2.9.4 Ongoing Determination of the Instrument MDL
Once the MDL has been initially established, the MDL should be re-determined when changes to
the instrument would reasonably expect to affect the sensitivity (such may include replacing a
detector lamp, detector, and/or flow cell mirrors, for example). MDL standards could be
measured periodically to prepare a dataset for calculating MDLsp (as in Section 6.2.9.2) and the
ongoing collection of routine zero blank data would provide a population of blanks to calculate
MDLb (as in Section 6.2.9.1). If changes have not been made to the instrument that affect
sensitivity, the MDL should be updated annually by including these ongoing collected zero blank
and low level standard hourly data. Calculate the MDLb and MDLsp as described in Sections
6.2.9.2 and 6.2.9.3 and update the method MDL for the monitor.
6.2.9.5 Example MDL Calculation for Continuous Gaseous Criteria Pollutant Monitors
A site is determining the MDL for a brand new true NO2 instrument. The instrument is powered
on and conditioned per the manufacturer instructions and then is calibrated per the monitoring
agency SOP. Over the course of a week, the instrument is challenged with zero air for several
20- to 30-minute periods every other day and the values combined to generate the MDLb. The
monitoring agency also uses blank data to determine an approximate spiking level based on the
variability of the zero blanks as in Section 6.2.9.2. The site determines the approximate spiking
concentration by calculating the standard deviation of the suite of zero blanks and multiplying
this by three. This value is approximately 0.33 ppb and is rounded up to 0.5 ppb to achieve a
better instrument signal based on recording some measurements in the zero blanks at
approximately 0.4 ppb. The following week, the instrument is challenged with an NO2 standard
every other day by diluting a cylinder of NO2 to 0.5 ppb with zero air with a DDC. Once the true
NO2 instrument response shows a stable response each day, the averages for each measurement
period are computed. The average (x) and standard deviation (s) of measured average true NO2
concentrations are determined for both the zero blanks and known standard analyses of 0.5 ppb.
Refer to the collected data in Table 6-2.
To calculate the MDLb, the standard deviation of the zero blank measurements (5b) is multiplied
by the associated student's T for the 27 aliquots (degrees of freedom = 26) and this product is
added to the average blank value, Xb:
MDLb = 0.110 ppb • 2.479 + 0.248 ppb
= 0.521 ppb
132
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
Table 6-2. Example True NO2 MDL Determination
Zero 1 Jlank
sample dale and start time
Zero I Jlank
measurement period
average (ppb)
Standard Analysis
sample date and start time
Standard Analysis
measurement period
average (ppb)
3/8/18 9:00
0.247
3/15/18 10:00
0.552
3/8/18 10:00
0.323
3/15/18 11:00
0.504
3/8/18 11:00
0.178
3/15/18 12:00
0.612
3/8/18 12:00
0.371
3/15/18 13:00
0.688
3/8/18 13:00
0.098
3/15/18 14:00
0.512
3/8/18 14:00
0.212
3/15/18 15:00
0.663
3/8/18 15:00
0.090
3/15/18 16:00
0.443
3/8/18 16:00
0.112
3/15/18 17:00
0.529
3/8/18 17:00
0.265
3/15/18 18:00
0.422
3/10/18 9:00
0.400
3/17/18 10:00
0.602
3/10/18 10:00
0.146
3/17/18 11:00
0.555
3/10/18 11:00
0.220
3/17/18 12:00
0.514
3/10/18 12:00
0.441
3/17/18 13:00
0.633
3/10/18 13:00
0.356
3/17/18 14:00
0.642
3/10/18 14:00
0.214
3/17/18 15:00
0.611
3/10/18 15:00
0.141
3/17/18 16:00
0.472
3/10/18 16:00
0.443
3/17/18 17:00
0.522
3/10/18 17:00
0.312
3/17/18 18:00
0.599
3/12/18 9:00
0.167
3/19/18 10:00
0.623
3/12/18 10:00
0.224
3/19/18 11:00
0.492
3/12/18 11:00
0.163
3/19/18 12:00
0.466
3/12/18 12:00
0.089
3/19/18 13:00
0.616
3/12/18 13:00
0.311
3/19/18 14:00
0.546
3/12/18 14:00
0.254
3/19/18 15:00
0.612
3/12/18 15:00
0.443
3/19/18 16:00
0.678
3/12/18 16:00
0.267
3/19/18 17:00
0.445
3/12/18 17:00
0.222
3/19/18 18:00
0.687
total zero blanks (n)
27
total standard analyses (n)
27
degrees of freedom (n-1)
26
degrees of freedom (n-1)
26
student's T for n-1 =26
2.479
student's T for n-1 =26
2.479
Xb(ppb)
0.248
(PPb)
0.564
Sb
0.110
•Ssp (ppb)
0.080
student's T
2.479
student's T
2.479
MDLb (ppb)
0.521
MDLsp (ppb)
0.197
3- 5b
0.330
To calculate the MDLsp, the standard deviation of the standard analyses (.ssp) is multiplied by the
associated student's T for the 27 aliquots (degrees of freedom = 26).
MDLSp = 0.080 ppb • 2.479
= 0.197 ppb
The MDLsp is subsequently verified to be less than the nominal standard level, and the nominal
standard level is confirmed to be less than 10-fold the MDLsp:
MDLsp < nominal standard level < 10-fold MDLsp
0.197 ppb <0.5 ppb < 1.97 ppb
133
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
The MDLsp and MDLb are compared to determine which is greater, and the greater of the two
values is reported as the monitor MDL.
0.521 ppb> 0.197 ppb
The monitor reported MDL is 0.521 ppb.
6.2.10 True NO2 Quality Control
40 CFR Part 58, Appendix A provides details about the number of QC samples that will be
implemented for FRMs and FEMs. The QC parameters and acceptance criteria are shown in
Table 6-3. For NO2, QC samples should include the following:
• Zero point checks - Bi-weekly zero point checks are conducted analyzing gas provided
by zero air generators. The NO2 analyzer will operate in its normal sampling mode during
the QC check and the test atmosphere will pass through all filters, scrubbers, conditioners
and other components used during normal ambient sampling and as much of the ambient
air inlet system as is practicable. The QC check is to be conducted before any calibration
or adjustment is made to the monitor. The zero will be < 0.2 ppb or MDL, whichever is
lower.
• QC checks - A one-point QC check for NO2 will be performed at least once every 2
weeks on each automated monitor used to measure NO2; however, more frequent
checking is strongly encouraged. The QC check is made by challenging the monitor with
a standard gas of known concentration selected to represent the approximate mean or
median concentrations at the site. If the mean or median concentrations are below the
MDL of the instrument, the monitoring agency can select the lowest concentration in the
prescribed range that can be practically measured. If the mean or median concentrations
are above the prescribed range the agency can select the highest concentration in the
prescribed range. An additional QC check point is encouraged for those organizations
that may have occasional high values or would like to confirm the monitor linearity at the
higher end of the operational range or around NAAQS concentrations. The NO2 analyzer
will operate in its normal sampling mode during the QC check and the test atmosphere is
to pass through all filters, scrubbers, conditioners and other components used during
normal ambient sampling and as much of the ambient air inlet system as is practicable.
The QC span check is conducted before any calibration or adjustment is made to the
monitor. These one-point QC span checks should be reported to AQS. The percent
differences between these concentrations are used to assess the precision and bias of the
monitoring data. Measured values will be within ± 10% of the nominal concentration.
• Span Point - A bi-weekly span point is performed at 80 to 90% of the analyzer full scale.
The span check concentration should be above 99% of the routine data over a 3-year
period. Measured values will be within ± 10% of the nominal concentration.
• Precision Point - A bi-weekly standard point in the lower third of the full-scale range.
Measured values will be within ± 10% of the nominal concentration.
134
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
Table 6-3. Quality Control Parameters and Acceptance Criteria for True NO2
QC Parameter
Description
Required
Frequency
Acceptance Criteria
Suggested Corrective
Action
Initial calibration
(ICAL)
Set ZERO and SPAN
levels on the true NO2
analyzer using zero air
and an NO2 standard at
80% of the full-scale
measurement (e.g., 0
and 160 ppb NO2 for a
full-scale range of 0 to
200 ppb)
Initially before
the beginning of
PAMS season,
following
maintenance to
the instrument
expected to alter
the instrument
response,
following failing
SPAN checks,
and at the end of
PAMS season
None. Verified by
MPV.
Repeat calibration. It may
be necessary to investigate
for system contamination
or interferences resulting
in suppression or
enhancement.
Multipoint
Verification (MPV)
Verification performed
by analyzing five
concentration points
including a zero and
covering the
calibration range, (e.g.,
0, 25, 75, 125, and 175
ppb)
Immediately
following
establishing a
new calibration
For linear regression,
will show r2 of >
0.995 and have an x-
intercept within ± 0.2
ppb NO2 of the origin.
Each standard level
evaluated against the
calibration curve will
be within 10% of the
nominal
concentration.
Repeat verification. It may
be necessary to investigate
for system contamination
or interferences resulting
in suppression or
enhancement of analytes.
Recalibration may be
necessary.
Zero/Span
Verification
Analysis of zero air
and mid-level NO2
standard to monitor for
drift in zero and span
levels
Optional,
recommended
daily during
nighttime hours
Zero level will be less
than 0.2 ppb or
analyzer MDL,
whichever is lower.
Span level will be
within 10% of the
nominal
concentration.
Repeat zero and span
checks. Investigate system
for contamination. Qualify
data since the last passing
zero/span check. May be
necessary to repeat ICAL
and MPV.
Zero/Span/Precision
Verification
Verification performed
by analyzing three
points including a zero
and two standard
concentration levels,
(e.g., 0, 170, 50 ppb)
Biweekly - The
SPAN check is
reported to AQS
Zero level will be less
than 0.2 ppb or
analyzer MDL,
whichever is lower.
Span level will be
within 10% of the
nominal
concentration.
Repeat
Zero/Span/Precision
Verification. Investigate
system for contamination.
Qualify data since the last
passing zero/span check.
May be necessary to
repeat ICAL and MPV.
6.3 NOy
A standard reference method for total reactive gaseous nitrogen (NOy) has not been designated.
However, an instrument design modification of the NOx heated bed chemiluminescence
approach by moving the converter to the sample inlet avoids line loss of adsorbent NOy species,
such as HNO3, is in wide use (Figure 6-6). Measurement of NOy is described in detail in the EPA
Precursor Gas TAD, Section 4.1 NOy measurements conducted for NCore are acceptable for the
135
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
PAMS network. Following is a summary of commercially-available NOy measurement
technologies.
Operation Principle
\
TeledyneT200U NOy
Heated-bed chemiluminescence
• Measures NO, NOy and NOy-NO by thermal conversion to
NO, then detection by chemiluminescence
• External molybdenum converter at ~ 10 m
• Converter temperature set point 315±7 °C
* 51
Heated-bed chemiluminescence
• Measures NO, NOy and NOy-NO by thermal conversion to
NO, then detection by chemiluminescence
• External molybdenum converter at ~ 10 m
• Converter temperature set point 325 °C
Thermo 42i-Y
Ecotech EC9843
Heated-bed chemiluminescence
• Measures NO, NOy and NOy-NO by thermal conversion to
NO, then detection by chemiluminescence
• External molybdenum converter at ~ 10 m
• Converter temperature set point 375 °C
6.4
Figure 6-6. Summary of Commercially-Available NOy Analyzers 11
References
1. Technical Assistance Document (TAD) For Precursor Gas Measurements in the NCore
Mirfti-Pollutant"MonitoringNetwork, Version 4, EPA-454/R-05-003, September 2005.
Available at (accessible March 2018):
https://www3.epa.gov/ttnamtil/ncore/guidance/tadversion4.pdf
2. Quality Assurance Handbook for Air Pollution Measurement Systems. Volume II - Ambient
Air Quality Monitoring Program, EPA-454/B-08-002, U. S. Environmental Protection
Agency, January 2017.
3. Beaver, Long, and Kronmiller "Characterization and Development of Measurement Methods
for Ambient Nitrogen Dioxide (NO2)," May 16, 2012
4. Beaver et al„ Direct and Indirect Methods for the Measurement of Ambient Nitrogen
Dioxide. Extended Abstract 57, Conference, 2013.
5. T. K. Boyson, T. G. Spence, M. E. Calzada, and C. C. Harb, "Frequency domain analysis for
laser-locked cavity ringdown spectroscopy," Opt. Express 19, 8092-8101 (2011) [FIGURE
C]
136
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
6. Beaver, M., K. Kronmiller, R. Duvall, S. Kaushik, T. Morphy, P. King, and R. Long. Direct
and Indirect Methods for the Measurement of Ambient Nitrogen Dioxide. Presented at
AWMA Measurement Technologies meeting, Sacramento, CA, November 19-21, 2013.
7. Aerodyne Research, Inc. CAPS N02 Monitor brochure, accessed February 2018.
http://www.aerodyne.coTO/gites/deraiilt/ftles/CAPS%20N02.%2021
8. Kuebler, Dustin, and Gaddie, Satchel. N02 Measurement: A Comparison of Direct vs.
Traditional Methods at a Near-Roadway Site. Missouri Dept of Natural Resources. Available
at (accessed March 2018): https://dnr.rno.gov/env/apcp/docs/N02%20Measurernent%20~-
%20A%20Comparison%20of%20Direct%20vs.%20Traditional%20Methods%20at%20a%2
0Near-roadwav%20 Site .pdf
9. U.S. EPA EPA Traceability Protocol for Assay and Certification of Gaseous Calibration
Standards. EPA/600/R-12/531. May 2012. Available at:
https://cfpub.epa.gov/si/si public file download. cfm?p download id=522029&Lab=NRMR
L Accessed February 2019.
10. Taylor, Vicki. API N02 CAPS Performance Review. Presented at the 2017 EPA Region 4
Ambient Air Monitoring Conference.
11. Russell Long. Overview of ORD N02, NOx and NOy Measurement Research. Presented at
NAAMC August 8-11, 2016.
137
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
7.0 OZONE
EPA revised the primary and secondary O3 standard NAAQS to 0.070 ppm in 2015. Ambient air
ozone monitoring has been a component of the State and Local Ambient Monitoring Stations
since the 1970s. Ozone monitoring will take place at PAMS sites using automated FRMs or
FEMs, as established in 40 CFR Appendix C to Part 58, Ambient Air Quality Monitoring
Methodology. Ozone measurements will be conducted continuously on a daily basis during the
PAMS monitoring period and measurements reported as hourly averages.
Production of tropospheric ozone, the concentration of which is regulated under the CAA, occurs
through complex reaction chemistry involving NOx and VOCs, as introduced previously in the
discussion of NOx, above. The atmospheric processes involved in ozone photochemistry are
complex; a simplified representation is shown in Figure 7-1. Generally, high ozone
concentrations are most likely to reach unhealthy levels and exceed EPA's ozone standards in
urban areas on warm, sunny summer days. However, the lifetime of ozone in the troposphere is
sufficient to cause elevated ozone concentrations in downwind rural areas. Ozone reduction
programs generally target emission control strategies for VOCs and/or NOx - as ozone
precursors. Since ozone production rates are not linear with respect to VOC and NOx
concentrations, atmospheric modeling of ozone photochemistry and meteorology is needed to
design effective control strategies. Concentration measurements from the PAMS program assists
modelers in the evaluation of the accuracy of these models.
Measurement of ozone is required atNCore sites, therefore the policies, procedures, and
requirements for ozone measurement will be those as described within the 40 CFR Part 58
Appendix C, NCore Precursor Gas Measurements TAD,1 and in the EPA QA Handbook Volume
II.2
138
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
Compounds
••-+OH, 031 NOj
} VOCs
(aromatics, terpenes)
ROOH
combustion
CH
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
8.0 METEOROLOGY
Information and guidance for meteorological measurements are provided in detail in the Quality
Assurance Handbook for Air Pollution Measurement Systems, Volume IV - Meteorological
Measurements. EPA-454/B-08-002.1 The following summarizes the guidance presented in the
Handbook. However, it is recommended that monitoring personnel also read the Handbook for a
more complete discussion of these measurements. MLH measurements using ceilometers are not
covered in the current version of the EPA QA Handbook, and are consequently discussed in
more detail in Section 8.8.
The data quality indicators and associated measurement quality objectives for each of the
meteorological measurements are presented in Sections 3.2 and 3.3 and in Table 3-1. A
summary of meteorology quality control checks is shown in Table 8-1.
8.1 Wind Speed and Wind Direction
Wind speed and direction measurements are essential to the evaluation of transport and
dispersion processes of all atmospheric pollutants. Wind speed is typically measured with a cup
or propeller anemometer; wind direction is typically measured with a wind vane. The use of
sonic anemometers has also become more prevalent in recent years. The standard height for
surface layer wind measurements is 10 m AGL.1'3'4
The location of the site for the wind measurements should ensure that the horizontal distance to
obstructions (e.g., buildings, trees) is at least 10 times the height of the obstruction.1'4 An
obstruction may be man-made (e.g., a building) or natural (a tree). A wind instrument should be
securely mounted on a mast that will not twist, rotate, or sway. Roof mounting is not
recommended and should only be resorted to when absolutely necessary. If a wind instrument
must be mounted on the roof of a building, it should be mounted high enough to be out of the
wake of an obstruction. Sensor height and its height above the obstructions, as well as the
character of nearby obstructions, is to be documented in site planning documentation.
An open lattice tower is the recommended structure for monitoring of meteorological
measurements at the 10-m level. In the case of wind measurements, certain precautions are
necessary to ensure that the measurements are not significantly affected by turbulence in the
immediate wake of the meteorological tower. To avoid such tower effects, the wind sensor
should be mounted on a mast a distance at least one tower width above the top of the tower, or if
the tower is higher than 10 m, on a boom projecting horizontally from the tower. In the latter
case, the boom should extend a distance at least twice the diameter/diagonal of the tower from
the nearest point on the tower. The boom should project into the direction which provides the
least distortion for the most important wind direction (i.e., into the prevailing wind).
There are several types of open lattice towers: fixed, tilt-over, and telescopic. A fixed tower is
usually assembled as a one-piece structure from several smaller sections. This type of tower must
be sturdy enough so that it can be climbed safely to install and service the instruments. Tilt-over
towers are also one-piece structures, but are hinged at ground level. This type of tower has the
advantage of allowing the instruments to be serviced at the ground. Telescopic 10-m towers are
140
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
usually composed of three sections, each approximately 4 m in length. The top section is the
smallest in diameter and fits inside the middle section which, in turn, fits inside the base section.
The tower can be extended to a height of 10 m by use of a hand crank located at the lowest
section. The top of the tower can be lowered to a height of about 4 m providing easy access to
the wind sensors. Telescopic and tilt-over towers are not generally recommended for heights
above 10 m. Regardless of which type of tower is used, the structure should be sufficiently rigid
and properly guyed to ensure that the instruments maintain a fixed orientation at all times.
Instrumentation for monitoring wind speed and direction should never be mounted on or near
solid structures such as buildings, stacks, water storage tanks, cooling towers, etc., because such
structures create significant distortions in the flow field.
A sensor with a high accuracy at low wind speeds and a low starting threshold is recommended
for PAMS applications. Lightweight materials (e.g., molded plastic or polystyrene foam) should
be employed for cups and propeller blades to achieve a starting threshold (lowest speed at which
a rotating anemometer starts and continues to turn and produce a measurable signal when
mounted in its normal position) of < 0.5 m/s. Wind vanes or tail fins should also be constructed
from lightweight materials. The starting threshold (lowest speed at which a vane will turn to
within 5° of the true wind direction from an initial displacement of 10°) should be < 0.5 m/s.
Overshoot must be < 25% and the damping ratio should lie between 0.4 and 0.7.
Wind speed measurements should be accurate to within ±0.2 m/s or ±5%, whichever is greater.
Wind direction measurements should be accurate to within ±5° including the combined error in
the system from the alignment with true north and the error inherent in the instrument.
Alignment with true north should be < 1° and instrumental error should be < 3° to ensure the
±5° tolerance threshold is met.
8.2 Temperature
Temperature affects photochemical reaction rates and, consequently, is an essential measurement
for PAMS applications. Sensors used for monitoring ambient temperature include wire bobbins,
thermocouples, and thermistors. Platinum resistance temperature detectors are among the more
popular sensors used in ambient monitoring; these sensors provide accurate measurements and
maintain a stable calibration over a wide temperature range.
The standard height for surface layer ambient temperature measurements is 2 m AGL. Higher
mounting is permitted; if a tower is used, the temperature sensor should be mounted on a boom
which extends at least one tower width/diameter from the tower. The measurement should be
made over a uniform plot of open, level ground at least 9 m in diameter centered on the sensor.
The surface should be covered with non-irrigated or un-watered short grass or, in areas which
lack a vegetation cover, natural earth. Concrete, asphalt, and oil-soaked surfaces and other
similar surfaces should be avoided to the extent possible. The temperature sensor should be at
least 30 m horizontally from any such paved area. If these siting criteria (open ground and
distance from paved surfaces) cannot be achieved, it should be identified in site characterization
documentation. Other areas to avoid include extraneous energy sources (subway entrances,
rooftops, electrical transmission equipment), large industrial heat sources, roof tops, steep slopes,
hollows, high vegetation, swamps, snow drifts, standing water, tunnels, drainage culverts, and air
141
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
exhausts. The distance to obstructions for accurate temperature measurements should be at least
four times the obstruction height.7
Temperature measurements should be accurate to ±0.5°C over a range of -30 to +50°C with a
resolution of 0.1°C. The thermal time constant (the time it takes the temperature sensor to reach
63.2% of the total difference between its initial and final temperature) should be 60 seconds.
Solar heating is usually the greatest source of error and, consequently, adequate shielding is
needed to provide a representative ambient air temperature measurement. Ideally, the radiation
shield should block the sensor from view of the sun, sky, ground, and surrounding objects. The
shield should reflect all incident radiation and not reradiate any of that energy towards the
sensor. The best type of shield is one which provides forced aspiration at a rate of at least 3 m/s
over a radiation range of-100 to +1100 W/m2. Errors in temperature should not exceed ±0.25 °C
when a sensor is placed inside a forced aspiration radiation shield. The sensor must also be
protected from precipitation and condensation, otherwise evaporative effects and other forms of
radiational heating or cooling will lead to a depressed temperature measurement (i.e., wet bulb
temperature). Temperatures may be reported to AQS in °F or °C; °C is the standard default
temperature unit for AQS.
8.3 Relative Humidity
Measurements of atmospheric humidity are essential to understanding chemical reactions
involving ozone precursors and water vapor. Measures of atmospheric humidity include vapor
pressure, dew point temperature, specific humidity, absolute humidity, and RH. For the PAMS
program, RH will be reported. The methods described here to measure the water vapor content of
the atmosphere require measuring the ambient temperature and require similarly protecting the
measuring probe from temperature influences as ambient temperature monitoring equipment. To
protect from such influences, probes are typically installed within naturally or mechanically
aspirated shields.
Electrical hygrometers are commonly available for measuring RH and are excellent alternatives
to chilled-mirror, wet-bulb thermometer, and wire-wound salt-coated bobbin sensors used
historically. Modern electrical hygrometers operate by measuring the changes in voltage output
of thin hygroscopic films that react to the presence of moisture by changing resistance and
capacitance. The moisture changes are correlated with corresponding temperature measurements
to determine the RH.
The standard height for humidity measurement installation is 2 m AGL. The humidity sensor
should be installed using the same siting criteria as used for temperature, noting that nearby
standing water should be avoided. If possible, the humidity sensor should be housed in the same
aspirated radiation shield as the temperature sensor. The humidity sensor should be protected
from contaminants such as salt, hydrocarbons, and particulates. The best protection is the use of
a porous membrane filter which allows the passage of ambient air and water vapor while keeping
out particulate matter. Measurements should be accurate to ±5% as expressed as RH over a range
of 10 to 100% RH.
142
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
8.4 Solar Radiation
Solar radiation refers to the electromagnetic energy in the solar spectrum (0.10 to 4.0 |im
wavelength). The solar spectrum is commonly subdivided as ultraviolet (0.10 to 0.40 |im),
visible light (0.40 to 0.73 |im), and near-infrared (0.73 to 4.0 |im) radiation. About 97% of the
solar radiation reaching the earth's outer atmosphere lies between 0.29 and 3.0 |im.3 A portion of
this energy penetrates through the atmosphere and is either absorbed or reflected at the earth's
surface. The remaining radiation is scattered and/or absorbed in the atmosphere before reaching
the surface. Solar radiation measurements are used in heat flux calculations, for estimating
atmospheric stability, and in modeling photochemical reactions.
Energy fluxes in the solar radiation spectrum are measured using a pyranometer. These
instruments are configured to measure what is referred to as global solar radiation (i.e., direct
plus diffuse [scattered] solar radiation). The sensing element of a typical pyranometer is
protected by a clear glass dome to prevent entry of energy (wavelengths) outside the solar
spectrum (i.e., long-wave radiation). These glass domes are transparent to wavelengths in the
range of 0.28 to 2.8 |im.
Solar radiation measurements should be taken in a location with an unrestricted view of the sky
in all directions. In general, locations should be avoided where there are obstructions that could
cast a shadow or reflect light on the sensor; light colored walls or artificial sources of radiation
should be avoided. The horizon as viewed from the pyranometer should not exceed 5 degrees.
Sensor height is not critical for pyranometers; consequently, tall platforms or roof tops are
typical locations. Regardless of where the pyranometer is sited, it is important to ensure that the
instrument is maintained level and that the glass dome is cleaned as necessary. To facilitate
leveling, pyranometers should be equipped with an attached circular spirit level.
Instrument manufacturer's specifications should match the requirements of the World
Meteorological Organization3 for either a secondary standard or first class pyranometer, in order
to meet the performance specifications in Table 3-1.
8.5 Ultraviolet Radiation
UV radiation can be divided into three sub-ranges: UV-A (0.315 to 0.400 |im), UV-B (0.280 to
0.315 |im), and UV-C (0.100 to 0.280 |im). Due to absorption by stratospheric ozone, the UV
radiation reaching the surface of the earth consists primarily of wavelengths longer than 0.28 |im
(UV-A and UV-B ranges). The most important photochemically active chemical species at these
wavelengths are ozone, NO2, and formaldehyde; the latter two chemical species are important in
the formation of ozone. Pyranometers with a spectral response covering both the UV-A and UV-
B (0.280 to 0.400 m) ranges, commonly referred to as "total ultraviolet" or "broadband"
radiometers, are recommended for PAMS applications. These instruments provide a relatively
constant response covering the UV-A and UV-B ranges and are more suitable for measuring total
UV radiation than employing multiple sensors each covering only a portion of the UV spectrum.
Users are discouraged from monitoring UV radiation with two separate instruments each
uniquely measuring one UV spectral range (e.g., one UV-A instrument and one UV-B
143
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
instrument), as the typical overlap in response is small and exhibits a depression in the spectral
response resulting in under-reporting the total UV.
Guidelines for instrument siting of UV radiation sensors are identical to those for solar radiation
measurements, above.
8.6 Barometric Pressure
Barometric pressure (station pressure) is used in the calculation of fundamental thermodynamic
quantities (e.g., air density). The type of sensor used to measure pressure is called a pressure
transducer for which there are numerous commercially-available instruments that meet the
specifications in Table 3-1. Ideally, the pressure sensor should be located in a ventilated shelter
about 2 m AGL. If the pressure sensor is placed indoors, accommodations should be made to
vent the pressure port to the outside environment. One end of a tube should be attached to the
sensor's pressure port and the other end vented to the outside of the shelter so that pressurization
due to the air conditioning or heating system is avoided. The wind can often cause dynamical
changes of pressure in a room where a sensor is placed. These fluctuations may be on the order
of 2 to 3 hPa when strong or gusty winds prevail. The height of the station above mean sea level
and the height of the pressure sensor AGL should be documented.
8.7 Precipitation
Precipitation should be measured with a recording precipitation gauge such as a tipping bucket or
weighing bucket. Precipitation gauges that operate using acoustic methods typically do not
comply with the performance requirements in Table 3-1 and are difficult or impossible to
independently audit in the field.
The precipitation gauge should be located on level ground in an open area. Obstructions should
not be closer than two to four times their height from the instrument. The area around the
precipitation gauge should be covered with natural vegetation. The mouth of the gauge should be
level and should be as low as possible while still precluding in-splashing from the ground (30 cm
AGL is the recommended minimum height). For ease in user access, the gauge should not be
mounted higher than 2 m AGL. A wind shield/wind screen (such as an Alter-type wind shield
consisting of a ring with approximately 32 free-swinging separate metal leaves) should be
employed to minimize the effects of high wind speeds.
8.8 Mixing Layer Height
8.8.1 Definition and Measurement of Mixing Layer Height
The planetary boundary layer (PBL), or atmospheric boundary layer, is the lowest part of the
Earth's atmosphere. It is directly influenced by its contact with the Earth's surface. The PBL
responds to heat transfer, pollutant emission, and other surface forcings in a timescale of an hour
or less.
144
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
The depth of the PBL depends on the location, season, time of day, and weather. Typically, this
boundary layer extends 50 to 3000 m from the Earth's surface. Fog, haze, mist, and air pollution
are typical phenomena in the PBL.
The PBL contains several layer types:
• Convective boundary layer: Layer of air in which particles mix well due to mechanical
and thermal forces.
• Nocturnal boundary layer: Stable layer of air that forms around sunset. Its top is often
marked by a temperature inversion. The layer usually dissolves by convection in the
morning hours, but it can also stay during daytime when solar heating is not sufficient to
disperse the nocturnal boundary layer.
• Residual layer: Layer of air containing the particles left from the previous convective
boundary layer after sunset and before the onset of convection the following morning or
from long-range transport by winds.
• Surface layer: Layer of air that is situated closest to the ground. Its thickness is typically
50 to 100 m, about 10% of the boundary layer height. This is the layer with the greatest
wind shear.
There are a variety of different definitions that describe the mixing layer of the atmosphere.
The goal for the PAMS measurements is to collect a consistent data set of hourly values
describing the mixing height in the lowest atmospheric boundary layer. For the purposes of the
intended PAMS measurements, the mixing height is the layer of air adjacent to the ground in
which any pollutant or particle released into it will be mixed vertically. If there are multiple
layers identified, then the mixing height is the lowest of those layers.
The MLH is the height of the layer adjacent to the ground over which pollutants or any
constituents emitted within this layer or entrained into it become vertically dispersed by
convection or mechanical turbulence.
During the day, diurnal variation has the following effects on the planetary boundary layer (see
Figure 8-1):
• The turbulence in the air is driven by solar radiation and radiative cooling, both of which
occur simultaneously. At night, the radiative cooling of the surface controls the boundary
layer, creating the nocturnal layer. The nocturnal layer blocks the interference between
the surface layer and the residual layer. Before sunrise, the nocturnal boundary layer
height is the mixing height as depicted in Figure 8-1.
• After sunrise the solar radiation warming the ground destabilizes the surface layer,
leading to thermals of warm air that rise upwards initiating the convective mixing
process. The thermals continue to rise until their temperature has dropped to the same
temperature as the surrounding air. At the same time thermals of cool air sink down from
the top of the clouds. The resulting vertical turbulence transport mixes temperature,
moisture, and particles uniformly within this convective boundary layer. The convective
145
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
boundary layer reaches its maximum mixing height in the late afternoon. During daylight
hours the convective boundary layer height is the mixing height.
• When the convective boundary layer reaches the level of the residual boundary layer,
both layers merge together. This is an important process for air pollution transportation in
time and space, with this horizontal dispersion and transport of pollutants and particles
having a strong influence on the air quality.
• When the sun sets, radiative cooling of the ground results in the collapse of the
convective boundary layer. Driven by the radiative cooling process, a new nocturnal
boundary layer is formed which is again replaced with a new convective boundary layer
during the next day.
• The entrainment zone is the interface or boundary between the convective boundary layer
and free atmosphere.
Nocturnal boundary layer Convective boundary layer Nocturnal boundary layer
Figure 8-1. Diurnal Variation of the Planetary Boundary Layer Structure 8
8.8.2 Ceilometer Theory of Operation
Ground-based remote sensors such as a ceilometer are effective tools for acquiring upper-air
information and have played an increasingly important role in atmospheric boundary layer
studies. EPA has demonstrated successful accession of mixing layer height data with the
Vaisala Ceilometer CL51. A ceilometer (Figure 8-2) employs pulsed diode laser Light
Detection and Ranging technology, where short, powerful laser pulses are sent out in a vertical
or near-vertical direction. The reflection of light, backscatter, caused by haze, fog, mist, virga,
precipitation, and clouds, is measured as the laser pulses traverse the sky. The resulting
Free atmosphere
Entrainment zone
Radiative boundary layer
Residual layer
Surface layer
146
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
backscatter intensity profile, that is, the signal strength versus the height, is stored and
processed, and the mixing height is measured using the characteristics of the backscattered
profile. The time delay between the launch of the laser pulse and the detection of the
backscatter signal provides the measure of the layer heights. The operating principle of a
ceilometer is based on the measurement of the time needed for a short pulse of light to traverse
the atmosphere from the transmitter emitted from the ceilometer to the top of the
backscattering layer and back to the receiver of the ceilometer. The general expression
connecting time delay (t) and backscattering height (h) is:
h = ct/2
where c is the speed of light (c = 2.99 « 10s m/s)
A reflection from 25,000 feet can be seen by the receiver after t = 50.9 ps
Figure 8-2. Vaisala CL51 Ceilometer
Using this relationship between time and distance, a vertical backscatter profile is created (see
Figure 8-3). The backscatter signal is typically stronger in the planetary boundary layer where
particle concentration is higher, but weaker in the free atmosphere where the atmosphere
typically has fewer particles.
147
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
Backscatter from a convective boundary layer
3000
2500 -
Entrainment zone
2000
£
c
£ 1500 -
'I
I
1000
500
100
200
300 400 500 600 700
Backscatter in 10"9 m"1 sr"1
800
900 1000
Figure 8-3. Example Vertical Backscatter Profile
8.8.3 Ceilometer Siting and Installation
The ceilometer measurements are intended for more macro-scale application than are the surface
meteorological measurements. Consequently, the location of the ceilometer site need not be
associated with any particular PAMS surface site. Factors that should be considered in selecting a
site for the upper-air monitoring include whether the upper-air measurements for the proposed
location are likely to provide the necessary data to characterize the meteorological conditions
associated with high ozone concentrations, and the extent to which data for the proposed location
may augment an existing upper-air network.
The ceilometer should be securely installed on a stable level surface such as a concrete pad or
wooden platform suitably located to provide an unobstructed view of the sky. A wide-open
location is recommended where there are no tall trees, overhead lines, or antennas nearby.
Proximity to powerful radars should also be avoided to the extent possible. Any object in the
cone projecting upward created by an angle of 25° from vertical will impede the ability of the
ceilometer to properly measure atmospheric backscatter. Common interfering objects would
include powerlines, tree branches, tower support guidewires, flagpoles, or similar features which
may be permanently or transiently present above the ceilometer. Ceilometers are commonly
installed at airports, therefore there is no siting restriction with respect to air traffic.
A personal computer is necessary to communicate with the ceilometer unit, collect and store the
ceilometer data, and automatically estimate the mixing heights. Figure 8-4 shows a typical setup
of the Ceilometer CL51 using ethernet communications and cabling.
148
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
Laptop PC and
BL-VIEW242218
Ceilometer CL31 orCLSI
with Ethernet
communication option
Ethernet switch
EU, UK, and US powering
242770
— UK and US
~
W
mains cables
231353 (UK) Ethernet
231352 (US) cable 2 m
shielded
212326
EU, UK, and US
— mains cables
Ethernet cable 25 m
shielded
CBL210412-3M
CBL210413-3M
CBL210414-3M
Figure 8-4. Ceilometer Configuration 9
8.8.4 Ceilometer Operations
EPA has developed an SOP specific to operation of the Ceilometer CL51 for the PAMS network
to detail instrument and data handling operations. Briefly, little preventive maintenance is
required beyond verifying that the aperture window is clean and that the system does not
indicate warning messages or lights. The ceilometer operating system includes a number of
diagnostics to evaluate the operational status of the instrument. This includes an automatic
check of window contamination, resulting in a warning status if contamination is detected. The
system also has an autocalibration feature that simulates a delayed return of a laser firing,
testing the ranging operation critical to measuring the MLH.
8.8.5 Ceilometer Mixing Height Calculations
Regardless of the instrument used for the measurement, the system will need to employ
software for automatically calculating the hourly average mixing height. For the Ceilometer
CL51, Vaisala Boundary Layer View (BL-View) software is available to automatically make
these estimates, and is discussed below, providing an example of how these estimates are
made.10 Users are encouraged to operate the instrument with the latest software version to
ensure the most refined mixing layer height algorithm is employed for measurements.
Mixing heights are estimated by detecting the backscatter gradient or changes between the
planetary boundary layer and free atmosphere (the mixing height), as well as other atmospheric
structures, such as residual boundary layers and elevated smoke or aerosol plumes that may
149
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
produce strong backscatter gradients. In addition to looking at the backscatter gradient, the
mixing height algorithm also performs a "profile-fit" against common mixing height profiles
for the date and time of day. The date and time of day, combined with the latitude and
longitude of the site, are taken into consideration for establishing whether the measurements
are occurring during nocturnal or daytime conditions.
This merging of gradient and profile fit methods uses the following rules to select the mixing
height from the gradient and profile-fit retrievals:
• The gradient method's lowest retrieval for cloud profiles and shallow, high quality
boundary layers.
• Profile-fit retrieval in all other cases. The gradient method's lowest retrieval is not
displayed unless it differs by more than 1000 m (3281 ft).
• Second and third gradient retrievals are displayed in all cases, as they may indicate
residual layers or other aloft aerosol layers.
The algorithm identifies the various boundary layers, such as the nocturnal, convective, marine,
and residual layers, and differentiates the mixed layer from other aerosol layers detected by
BL-View. It also provides an outlier removal method, cloud filter, and other changes to
improve performance for evening boundary layer transitions.
The algorithm determines the mixing height by fitting an idealized backscatter profile to
observed range-corrected ceilometer backscatter profiles. Clouds and precipitation produce
backscattering profiles that deviate substantially from an idealized profile, which results in
poor mixing height estimates. The algorithm can produce valid retrievals even if the
backscatter profile deviates significantly from the idealized profile. When a ceilometer detects
multiple aerosol layers, the algorithm attributes one aerosol layer as the mixing height. When
there are multiple aerosol layers present, the lowest layer is a reasonable first guess for
attributing one of these layers as the mixing height. In general, backscatter data collected
during obvious precipitation events are not applied to mixing height measurements.
Figure 8-5 presents an example of mixing height data as displayed by BL-View for a typical
diurnal cycle. The mixing height as defined above (using a 1-hour time period, identified by the
thin black step-like lines in the figure) during the daytime is represented by the convective
boundary layer, which is consistent with estimates using either the parcel method to estimate
the layer mixed by thermal turbulence, or by aerosol or pollutant gradients that are mixed by
the thermal and/or mechanical mixing. During these periods of thermal instability, the mixing
height may grow rapidly during the morning hours reaching altitudes of several kilometers or
more depending on other atmospheric phenomena such as subsidence inversions. Again,
mixing to these altitudes is occurring over the defined one-hour time period. At the end of the
daytime unstable period, a transition through an evening neutral period typically occurs, with
the surface layer reforming that cuts off the pollutants (acts as a ceiling) at the surface from
those pollutants at the higher altitudes (the residual layer). It is important to distinguish this
residual layer from the mixing height as they are not the same. Furthermore, as the nighttime
stable layer grows, pollutants will accumulate at a slower rate through the residual layer, much
slower rate than hourly. For modeling purposes, there can be a large difference in the nighttime
150
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
mixing height (10s of meters over 1-hour) and the stable layer including the residual layer (the
pollutant accumulation depth, accumulating up to hundreds of meters over multiple hours). The
automated mixing height algorithms for the ceilometers address this nighttime period of
mixing.
VitSAU"""™
6 Jun / 13:05:52 use*
Figure 8-5. Example Graphical Display of Mixing Height using BL-View
Using BL-View as an example, the system should collect the hackscatter and mixing height
data automatically, as soon as it is connected to the ceilometer. Data are still collected even
when the software is not actively used to view the data, automatically storing the data to
netCDF files. Network Common Data Form (netCDF) is a set of software libraries and self-
describing, machine-independent data formats that support the creation, access, and sharing of
array-oriented scientific data. For BL-View, the following three file types are collected:
• The netCDF LI data file contains level 1 (LI) raw data from the ceilometer.
• The netCDF L2 data file contains level 2 (L2) data that have gone through the
precalculation service and averaging.
• The netCDF L3 data file contains level 3 (L3) data that have gone through the
calculation service and includes all the data from the algorithms, including mixing layer
height values and quality index data.
Of the three file types, the calculated mixing height is contained in the L3 files. The
backscatter profile is information rich, with the mixing height value comprising a small amount
of the overall data output in each discrete backscatter measurement with an associated data
quality indicator factor. PAMS monitoring agencies will report the hourly mixing height;
however, monitoring agencies are strongly encouraged to maintain the full wealth of the
151
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
collected backscatter data which are of interest to the greater scientific community. At
publication of this document, EPA, in concert with other atmospheric science organizations,
was in the early stages of developing a database repository to house the full suite of additional
backscatter data collected by the ceilometers. In the event the database comes to fruition, EPA
intends to request monitoring agencies submit or provide access to their ceilometer backscatter
data. While still aspirational, EPA may facilitate data verification and validation and coding of
mixing layer data for reporting the hourly MLH to AQS. Such would eliminate the need for
monitoring agencies to commit substantial resources to data handling and validation activities
for MLH data. It is expected that EPA will provide updates on the database status during
periodic workgroup meetings for PAMS Required Site stakeholders.
In the interim period until a centralized ceilometer backscatter database is available for
monitoring agency use, monitoring agencies will be responsible for collecting and storing
ceilometer backscatter data. The ceilometer software provides an hourly MLH value; however,
monitoring agencies will still need to visually review the backscatter data to eliminate
documentable issues that affect the mixing height measurements, including:
• Precipitation
• Fog
• Aperture window transmittance issues
• Local sources (wildfires, fireworks, etc.)
Situations that may impact the ability of the MLH algorithm to properly identify the hourly
MLH include instances of rapid change in the MLH (as occurs with rapid daytime temperature
increases), instances when levels of aloft aerosols are particularly low resulting in a low
backscatter signal, and high winds that disperse aerosol layers, among other situations. In such
cases, the algorithm may report an MLH value for the hour, but the quality indicator may
reflect low confidence in the value.
8.9 Quality Assurance/Quality Control for Meteorological Measurements
Very little additional QC is required for the PAMS meteorological measurements beyond that
specified in Section 2 and routine review of the data for signs of instrument failure. Quality
control, calibration and audit methodologies are presented in the Quality Assurance Handbook
for Air Pollution Measurement Systems, Volume IV - Meteorological Measurements,1 with
exception of those for the ceilometer. An SOP for the Ceilometer CL51 has been developed for
the PAMS network to address specific QC operations in detail.
As in the audits of any of the measurement methods, the mixing height values retrieved from
ceilometers will be verified for accuracy against an appropriate "standard." As part of an audit,
the accuracy of the altitude reporting of the ceilometers will be verified by aiming the ceilometer
at a hard target a known distance away. This "hard target" audit should be performed by pointing
the ceilometer at an object that reflects the light source a known distance at least 300 m from the
ceilometer. Such a hard target could be a wall at ground level, a vehicle, or other large profile
object of known distance (the ceilometer would be angled down, the beam aimed roughly
parallel to the ground).
152
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
While the above audit verifies the accuracy of the ceilometer components, it does not provide
any assurance that the algorithm is accurately estimating the mixing height. This can only be
achieved through the comparison of measurements against an established MLH methodology.
Traditionally, radiosondes have been used for measuring the temperature profile to determine the
thermal stability and estimate the layer of thermal turbulence that mixes the lower atmosphere.
Briefly, radiosondes are instruments (altitude, temperature, relative humidity, wind speed, wind
direction, cosmic ray, and global positioning system) which are attached to helium weather
balloons and are released at the surface. They report measurements at short intervals with respect
to altitude (up to approximately 20,000 m) and the data are analyzed to investigate gradients in
the measurements which correspond to atmospheric layers. Since the Holzworth method was
first implemented, this "Parcel" method has been used in one form or another to estimate the
mixing height. The method lifts the surface parcel dry adiabatically until it intersects the actual
sounding profile. This marks the height of the thermal turbulent layer. While some refinements
to the method will estimate slightly different altitudes to account for other factors, this height is
an accepted method for daytime mixing heights. The software program RAOB (The Universal
RAwinsonde OBservation program) implements this method from temperature soundings and
further refines the estimate that if no intersection with the profile is found, the height is located at
the top of the surface inversion, or else at the bottom of the first elevated inversion. While there
are further refinements that can be made, this "Parcel" method in RAOB can be used with the
radiosonde launches as the objective method for comparison with any of the instruments tested.
For each of the radiosonde soundings performed, the data should be quality controlled to assure
that artifacts are removed from the profiles prior to the mixing height determination. The profiles
should also be reviewed for meteorological reasonableness.
While currently outside the scope of the PAMS QA requirements, comparison of the ceilometer
measurements to radiosonde measurements can provide agencies with verification that the
ceilometer is providing representative data for their locality and environment. Such a comparison
would include radiosonde measurements at three diverse conditions in order to ensure the
ceilometer algorithm is properly configured to determine mixing height throughout a range of
atmospheric conditions. These three conditions should ideally be during, but toward the end of
the nocturnal period, during a transition period as in the early morning just following sunrise,
and during the afternoon. The goal is to capture the three rather distinct periods where mixing
heights would be different yet characteristic of conditions the ceilometer may encounter during
routine monitoring.
153
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
Table 8-1. Quality Control Parameters for Meteorology Measurements
Meteorology
C'iilihi'iilion Check
I'requencv
Accepl;i nee
Recommended
I'iinimeter
Sliindiircl
Crilerisi
Corrects e Action
Ambient
Verification in a water
<± 0.5°C at each of
Temperature
bath against a NIST-
traceable thermistor or
thermometer at three
points bracketing the
temperature range of use
the three
temperatures
checked
Relative
Compared to a NIST-
< ± 5% RH of the
Humidity
traceable psychrometer
or standard solutions
hourly average from
the certified
standard over the
duration of
comparison
Inspect instrument
for damage or worn
components.
Correct data where
possible (e.g. wind
direction).
Recalibrate
instrument. Qualify
all collected data
since the most
recent calibration
or acceptable
calibration check as
"QX" in AQS, as
applicable.
Potentially
invalidate data
since last most
recent calibration
or acceptable
calibration check.
Barometric
Pressure
Compared to a NIST-
traceably certified
barometer or pressure
transducer over the
course of several
consecutive hours
Semi-
< ± 3 hPa
Wind Speed
Compared to a NIST-
traceable synchronous
motor or CTSa method
annually
< ± 0.2 m/s or ± 5%,
whichever is greater
Wind Direction
Compared to solar noon,
GPS, magnetic compass,
or CTSa method
< ± 5 degrees
Solar Radiation
Compared to a NIST-
traceable pyranometer
< ± 5% b
UV Radiation
Compared to a NIST-
traceable radiometer
< ± 5% b
Precipitation
Add water at a constant
rate such that the gauge
tips every 15 seconds
and measure output with
a graduated cylinder
<± 10% of input
volume
Mixing Height
Altitude determination
verified against a hard
target of known distance
< ± 5 m or ± 1%,
whichever is greater
a CTS = collocated transfer standard
b Comparison should be made during sunny conditions.
154
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
8.10 References
1. Quality Assurance Handbook for Air Pollution Measurement Systems. Volume IV -
Meteorological Measurement, EPA-454/B-08-002, U. S. Environmental Protection Agency,
January 2008.
2. On-site Meteorological Program Guidance for Regulatory Modeling Applications, EPA-
454/R-99-005. Research Triangle Park, NC: U. S. Environmental Protection Agency,
February 2000.
3. Guide to Meteorological Instruments and Methods of Observation, WMO No. 8 Geneva,
Switzerland: World Meteorological Organization, 2014.
4. Standard Practice for Characterizing Surface Wind Using a Wind Vane and Rotating Cup
Anemometer, ASTM Designation: D 5741-96, Annual Book of ASTM Standards, Section
11, 1996.
5. Holzworth, G. C. Estimates of Mean Maximum Mixing Depths in the Contiguous United
States. Monthly Weather Review, 92, 235-242, 1964.
6. Holzworth, G. C. Mixing Heights, Wind Speeds, and Potential for Urban Air Pollution
Throughout the Contiguous United States, Publication No. AP-101. Research Triangle Park,
NC: U. S. Environmental Protection Agency, 1972.
7. Quality Assurance Handbook for Air Pollution Measurement Systems. Volume IV -
Meteorological Measurement, EPA-454/B-08-002, U. S. Environmental Protection Agency,
January 2008. Available at (accessed March 2018):
https://www3.epa.gov/ttnamtil/files/ambient/met/Voliii Meteorological Measurement
s.pdf
8. Irwin, J. S., and J. O. Paumier. Characterizing the Dispersive State of Convective Boundary
Layers for Applied Dispersion Modeling. Boundary-Layer Meteorology, 53, 267-296, 1990.
9. Vaisala Ceilometer CL51 User's Guide, M210801EN-A. Vaisala Oyi, 2010.
10. Vaisala Boundary Layer View (BL-View) User Guide, M211185EN-B. Vaisala Oyi, 2017.
155
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
9.0 DATA HANDLING
9.1 Data Collection
All records are to be documented in detail sufficient to reconstruct the activities and
transformations to generate reported concentration data. If such records are not available, validity
of the data cannot be determined. Such records minimally include observations, field and
laboratory measurements, and photographs as well as instrument calibration records and COAs.
Records related to transformation or adjustment of data such as through data reduction
spreadsheets, peak integrations, hand calculations, or calculations handled by a laboratory
information management system (LIMS) or data acquisition system (DAS) are to be maintained
and be transparent so the actions may be independently verified.
9.1.1 Validation of Data Reduction and Transformation Systems and Software
Data reduction algorithms and software, such as electronic spreadsheets or LIMS, simplify and
automate data collection and transformation actions. Prior to their implementation, data
calculations and transformations should be validated to ensure their function is accurate. If
updated or revised, such validation should be repeated to ensure proper function prior to use.
Errors in spreadsheets can occur during spreadsheet or program development and in the
continued use of the program. Implementing a verification check of the calculations in these
spreadsheets and programs ensure errors do not propagate in the data generated. Verification
checks are accomplished by using the spreadsheet or program to calculate results for a known
data set and compared to hand calculated results or to a previously validated result. A more
complete test of the user developed program would include testing at the minimum and
maximum expected input values and may include testing at values below the anticipated
minimum values and exceeding the maximum anticipated values where appropriate. If the
expected result is not obtained the program may be in error and the appropriate corrective action
should be taken.
9.2 Data Backup
Electronic data acquired from laboratory instruments, field instruments, databases, and data
manipulation software in support of PAMS Required Site program work should be maintained
for a minimum of five years following acquisition. In order to maintain electronic records for this
duration, it is necessary to prevent data loss and corruption by ensuring data redundancy. Each
PAMS Required Site agency should prescribe data redundancy policies and procedures, which
may be included in the program QAPP, SOP, or similar controlled document.
For DAS systems such as CDSs, auto-GC control and operation software, and environmental
control tracking software systems that are connected via computer network, a best practice is to
enable automated nightly backups of data to a separate physical hard drive (such as is done with
a redundant array of independent disks 1 [RAID1]) or server, preferably one at a different
physical location. Backing up of data to a separate partition on the same physical hard drive
provides little additional security if the hard drive fails. For software systems that are not
networked to a server, a best practice is to manually back up the data after completion of each
156
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
day's activities to removable media (thumb drive, external hard drive, etc.) for transfer to a
networked computer or server.
These daily backups should be protected from inadvertent alteration and compiled on a regular
frequency, recommended weekly but not to exceed monthly, to an archival system such as a tape
drive, DVD, additional external server, cloud storage, etc. This archival should be access-limited
by password and/or other security means to a select few individuals as deemed responsible by
cognizant management.
Archived electronic data should remain accessible such that retired computer and software
systems needed to read or view data should be maintained to access data, or archived data should
be converted such that it remains accessible and legible, until the archival period has lapsed.
Once archived, archived data should be reviewed or tested to ensure complete records are
maintained and data have not been corrupted. Such a review is recommended every six months,
but should not exceed annually.
9.3 Recording of Data
Data generated are to be recorded so that it is clear who performed the activity, when the activity
was performed, and, if applicable, who documented performance of the activity.
9.3.1 Paper Records
Data entries created on paper records such as field collection forms, COC forms, or laboratory
notebooks, are to be recorded in legibly in indelible ink and identify the individual creating the
entry. Measurements should clearly indicate appropriate units. Individuals creating paper data
records will be identified by way of signature or initials unique to the individual and in such a
manner that unambiguous identification is possible. One method by which such may be
accomplished is to create a cross-reference for each staff person that shows each staff person's
printed name, signature, and initials.
9.3.2 Electronic Data Capture
Electronic data recording systems such as electronic logbooks, CDSs, LIMS, DAS, and
instrumental data acquisition software generally require a user to log in with a username and
password to utilize the system. Each action (entry, manipulation, instrument operation) recorded
by such software systems should be attributable to an individual and the corresponding date and
time recorded. If so equipped, audit trail functionality should be enabled on software systems in
order to record changes made to electronic records. A best practice is to scan associated paper
records for conversion into electronic files of these records to be stored with electronic data.
9.3.3 Error Correction
Changes to recorded data or data transformation may be required due to calculation errors,
incorrectly recorded measurements, or errors noted during data verification and validation. When
157
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
records are amended, whether paper or electronic, the original record is to remain legible or
otherwise intact, and the following information should be recorded: the identity of the individual
responsible for making the change, the date the change was made and the rationale for the
change. For example, hand-written data records may be corrected by a single line through the
entry with the correction, the initials of the responsible individual, the date of correction, and the
rationale for change documented in close proximity to the correction or identifiable by annotated
footnote. For common corrections such as those for incorrect date, illegible entry, calculation
errors, etc., a list of abbreviations may be developed to document change rationale. Any such
abbreviations should be defined in a quality systems document such as an SOP, or in the front of
a logbook, etc.
9.3.3.1 Manual Integration of Chromatographic Peaks
Automated functions for the integration of chromatographic peaks are included in the CDS that
control the auto-GC and HPLC instruments. These integration functions should be configured
such that little intervention or correction is needed by the analyst, so as to best ensure that peak
integration is as reproducible and introduces as little human error as possible. While these
functions ensure consistent integration practices, subtle differences in peak shape, coeluting
peaks, and baseline noise may result in inconsistent or incorrect peak integration.
Analysts are to be properly trained to review and adjust peak integration performed by CDS
automated functions, and specific procedures for integration should be codified into each
agency's quality system. Manual changes to automated peak integration are to be treated as error
corrections. Typical corrections to peak integration may include: adjustment of the baseline,
addition or removal of a vertical drop line, or peak deletion if the requisite compound
identification criteria are not met. The identification criteria for the chromatography methods are
listed as follows:
VOCs: Section 4.2.4
Carbonyls: Section 5.9.5.6
Manual peak deletion, that is, effectively reporting that the compound was not detected, is not
permitted in instances in which the peak specified identification criteria are met.
For each adjustment to chromatographic peak integration (manual integration), the record of the
original automated integration should be maintained and it is strongly recommended that the
adjustment be justified with the documented rationale (S:N too low, incorrect retention time,
incorrectly drawn baseline, etc.), analyst initials, and date.
9.4 Numerical Calculations
Numerous calculations and transformations are necessary to determine the target analyte
concentration of a given field-collected sample or QC sample or to determine evaluate whether
data generated during calibration verifications meet acceptance criteria.
158
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
9.4.1 Rounding
Rounding of values should be avoided until the final step of a calculation. Rounding during
intermediate steps risks the loss of fidelity of the calculation which may lead to significant
calculation error.
EPA Region IV Science and Ecosystem Support Division has developed guidance for rounding
which is adopted into the revision of the Volume II of EPA's QA Handbook. This guidance is
included in Appendix A of this TAD.
9.4.2 Calculations Using Significant Digits
Final reported results should be rounded to the correct number of significant digits per the rules
below. To the extent feasible, carry the maximum number of digits available through all
intermediate calculations and do not round until the final calculated result. Non-significant digits
that are carried through calculations may be represented using subscripted numerals. (For
example, 2.32i has three significant figures, with the final 1 being non-significant and carried
through to avoid unnecessarily introducing additional error into the final result.)
9.4.2.1 Addition and Subtraction
The number of significant digits in the final result is determined by the value with the fewest
number of digits after the decimal place. For example:
A 5.6
B 63.71
C +9.238
78.5
The final result is limited to one decimal place due to the uncertainty introduced in the tenths
place by measurement A.
9.4.2.2 Multiplication and Division
The number of significant digits in the final result is determined by the value with the fewest
number of significant digits. For example, benzene was measured by the GC at a concentration
of 2.721 ppb from a canister that was diluted with zero air resulting in a dilution factor of 1.41.
The dilution factor is applied to the measured result to calculate the air concentration:
2.721 ppb • 1.41 =3.83? ppb
3.84 ppb
The final result is limited to three significant digits due to the dilution factor containing three
significant digits.
159
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
9.4.2.3 Standard Deviation
Standard deviation in a final result should not display digits in a place that the sample average
does not have a significant digit. Take, for example, the following average and standard
deviation of the form x ± 5:
107.2 ±2.31 is reported as 107.2 ±2.3
The standard deviation is rounded to the appropriate significant digit of the sample average.
9.4.2.4 Logarithms
For converting a value to its logarithm, retain as many places in the mantissa of the logarithm (to
the right of the decimal point in the logarithm) as there are significant figures in the number
itself, for example (mantissa underlined):
logio 24.5 = 1.389
For converting antilogarithms to values, retain as many places in the value as there are digits in
the mantissa of the logarithm, for example (mantissa underlined):
antilog (1.131) = 13.5
9.5 In-house Control Limits
The analysis methods detailed in Section 4, 5 and 6 specify acceptance criteria for routine QC
samples. These acceptance criteria are the maximum allowable ranges permitted, however,
monitoring agencies and ASLs may find that they rarely or never exceed the acceptance criteria.
As each laboratory/site and the associated instrument operator, instruments, and processes are
unique, development of in-house control limits is recommended to evaluate trends and identify
problem situations before exceedances to method acceptance criteria occur.
In-house control limits may be generated to evaluate the bias of quality control samples such as
the LCS, CCV, SSCV, and to evaluate precision of LCSD, matrix spike duplicate, etc. Warning
limits and control limits are established following acquisition of sufficient data points, generally
more than seven, per the guidance in the subsequent sections. Under no circumstances should
data be accepted which exceed method specified acceptance criteria even if in-house warning or
control limits have not been exceeded.
9.5.1 Warning Limits
Warning limits are established as a window of two standard deviations surrounding the mean
(x ± 2s). Exceedance of the warning limit should prompt monitoring of the parameter for values
which remain outside the warning limits. For repeated values exceeding the warning limits,
corrective action should be taken to address the trend.
160
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
9.5.2 Control Limits
Control limits are established as a window of three standard deviations surrounding the mean
(x ± 3.s), Corrective action should be taken when control limits are exceeded.
9.6 Negative Values
In general, negative values of small magnitude may be expected from certain analytical
platforms, specifically those which do not apply calibration regressions which are forced through
the origin. However, depending on the situation, negative numbers can be problematic and
indicative of bias due to faulty sensors, contamination in reagents and labware, improper
calibration, or calculation errors.
Negative values should be evaluated to ensure that their magnitude does not significantly impact
the resulting measurements.
Negative values for all qualitatively identified analytes are to be reported to AQS as-is without
censoring or replacing (substituting) with zero.
9.6.1 Negative Concentrations
For analysis measurements, a negative concentration result generated by a positive instrument
response (i.e., positive millivolt response or area count response) should be investigated to
ensure that the negative concentration is of small magnitude such that the absolute value of the
concentration is less than the MDLsp. Where negative concentrations fail this criterion, corrective
action should be taken to determine and remediate the source of the bias.
9.6.2 Negative Physical Measurements
For physical measurements such as temperature, mass, absolute pressure, and flow, negative
values generated by an instrument should be evaluated to ensure they do not adversely impact
future measurements.
For example, a pressure gauge reads -0.4 psia upon connection to a canister at hard vacuum. The
acceptable evacuated canister pressure threshold is 0.5 psia. Since negative absolute pressures
are impossible, the -0.4 psia reading is significant, especially when compared to an acceptance
criterion of 0.5 psia. Due to the -0.4 psia bias, the pressure in another canister at 0.8 psia would
be read 0.4 psia and would incorrectly meet the acceptance criterion for sample collection due to
the incorrect calibration of the pressure transducer.
161
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
10.0 PAMS DATA VERIFICATION AND VALIDATION
Verification and validation of PAMS data are critical steps in ensuring that the data produced are
of the type and quality needed to support environmental programs and decisions. While the
purpose of QC techniques is to minimize the amount of poor quality or unusable data being
collected, the data verification and validation process seeks to prevent poor quality data that may
have been collected from becoming incorporated into the data storage system (e.g., AQS) and
ultimately into the dataset utilized to satisfy the DQO. 1 When included in data analysis and
modeling efforts, poor quality or unusable data with serious errors can cause errors in
downstream data analysis and adversely impact policy decisions. It is highly recommended that
PAMS monitoring agencies develop an SOP or combination of SOPs for PAMS data verification
and validation. This section describes the purpose, workflow, methods, techniques, and tools to
verify and validate PAMS data; however, each monitoring agency will have a unique data
handling system, software tools, and set of circumstances for accomplishing the data verification
and validation activities. The SOP or group of SOPs should include a detailed set of instructions
for monitoring agencies to cover aspects described in this section.
This section describes data verification and validation methods, tools, and techniques used to
accept, reject (invalidate), or qualify (flag) PAMS network data in an objective and consistent
manner. Data verification is the process for confirming that established method, procedural, or
contractual specifications have been fulfilled. Data review is a component of data verification
conducted during the development of the initial data set, and includes reviews performed by the
data collectors and technical reviewers of collected data and QC data to ensure that records are
complete, accurate, and are representative of the conditions at the time measurements were
conducted. Such activities may include manual inspection of the collected data, confirming the
requisite number of samples were collected, ensuring that QC activities have been conducted and
meet criteria, and verifying that collection and analysis procedures comply with the program
QAPP and SOP to meet the program needs. Data may be corrected as practical during the data
verification process, or may be identified as problematic and qualified, or in certain instances,
invalidated.
Once the data verification process has been completed and the data are considered "clean", the
data undergo validation. Data validation is a process that investigates the individual data points
within the context of other co-collected or historical data to determine the analytical quality and
acceptability of the data relative to the intended end use. Data validation activities include, but
are not limited to, examining QA reports, examining chromatographic data, calculating summary
statistics, and developing plots/graphs to identify data that do not correspond to expectations and
warrant further investigation. This process increases the confidence in the data collected.
Additional validation activities include examining data for patterns or relationships that are
expected for routine ambient air data or anomalies in such patterns or relationships that may be
indicative of sampling, analytical, or data transformation issues.
When used as a general term describing the complete process of data assessment, data validation
can be subdivided into four levels:2
162
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
• Level 0: Comprises data verification activities, the major components of which are
routine review of data for completeness, correctness, and compliance with the associated
QAPP and SOP as well as technical review (peer review) by a person familiar with the
data generation process. The end goal of data verification is a dataset that is "clean"
where data have been verified to meet criteria and are flagged or invalidated when their
quality or integrity is compromised according to established criteria.
o Routine (self) review - The individual responsible for data generation (e.g., site
operator or instrument operator) performs routine checks during the initial
generation and processing of data. Such checks are basic verifications that data
collection records are complete and that recorded data are reasonable and
accurate. Timely routine reviews permit records to be corrected or corrective
action to be taken in a timely fashion to limit the impact on subsequent
measurements. Data of known substandard quality can be flagged or invalidated
at this point per the monitoring agency and/or program policies.
o Technical (peer) review - An individual familiar with, but not directly involved in
the data generation process, such as a supervisor or other site operator or analyst,
performs a higher level of review of the collected data for completeness and
correctness. In such reviews, the technical reviewer assesses compliance with the
governing SOPs such as verifying that all appropriate records have been
generated (electronic data files, sample collection forms, checklists, logbook
entries, etc.), QC activities are performed at the required frequency and have met
acceptance criteria, unusual circumstances or events are properly documented and
impact to the data explained, and that data transformations and calculations were
performed properly.
Activities performed during routine review and technical review should be documented.
Such documentation is particularly important for activities in which data are modified or
manipulated during the review process. Such records are necessary for subsequent data
validation activities once the data verification is complete. PAMS data should go through
data verification including routine (self) review and technical review prior to release for
validation.
• Level 1: Identifies data that are atypical within the dataset under examination
• Level 2: Comparison of the dataset with historical data to verify consistency over time
• Level 3: Comparison of the dataset with a different dataset collected from the same
population to investigate a systematic bias
The Level 0 activities comprise the data verification steps. Activities conducted under Levels 1,
2, and 3 comprise data validation activities, each employing different comparisons and tools but
with the common goal of identifying data that do not conform to expectations and warrant
further investigation.
In general, data verification and validation activities will be performed at a frequency specified
in the PAMS monitoring agency QAPP and supporting SOPs. The procedures, required
personnel, and frequency of the assessments should be included in the QAPP and/or supporting
SOPs. Data assessment activities (verification and validation) need to be completed prior to
163
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
submitting data to AQS. Once data are submitted to AQS, the monitoring agency should query
the data and ensure that the upload was error-free. A schematic of the data flow process from
generation through verification of AQS upload is shown in Figure 10-1.
Data Generation
Jl
Routine Self Review
Technical Review
J L
Level i Data Validation
Tl
Data
Verification
level 2 Data Validation _
"
level 3 Data Validation
Data
Validation
| Data Upload to AQS
| Verify AQS Data
Data
Reporting
Figure 10-1. Schematic of PAMS Data Generation, Verification, Validation, and Reporting
Data Verification and Validation Processes and Policies: Monitoring agencies should develop
data flow diagrams and procedure documents to indicate the steps taken to process the data
following collection, including data formatting, transmission, and processing or transformation
for AQS. Performance checks of the automated data processing systems and supplemental
procedures developed to handle the data, including telemetry, should be implemented. These
performance checks should be carried out prior to the beginning of PAMS season and
periodically during ongoing data collection to ensure data are not corrupted during collection,
transmission, or reduction. Such performance checks include reviewing data for inconsistencies,
missing data files, or nonsensical information such as would occur if database mapping or
programming was incorrectly performed. Computerized programs utilized to transform data
should be validated at a minimum using methods for checking errors such as developing a
standard set of test output parameters, processing a test data set, and comparing the results to the
reference. Further information on validation of software is discussed in Section 9.1.1.
10.1 Data Verification
In the data verification process, PAMS measurement data are evaluated for completeness,
correctness, and conformance/compliance according to the program requirements. The goal of
data verification is to ensure and document that the reported results reflect the activities
performed and measurements acquired. Any deficiencies in the data should be documented and,
where possible, resolved by corrective action. PAMS data verification applies to activities in the
164
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
field as well as in the ASL performing carbonyl cartridge extraction and analysis. As discussed
above, data verification includes routine (self) review of collected data by the instrument
operator and subsequent technical (peer) review.
Staff responsible for data verification activities will be familiar with the project requirements
defined in the monitoring agency PAMS QAPP, SOPs, and ANPs. Individuals conducting data
verification activities include site operators, laboratory analysts, and staff independent of data
collection. Staff conducting data verification activities should be familiar with the software
systems employed to generate, process, and transform data, the location(s) of stored data whether
paper records or electronic (raw, processed, and final), measurement system data outputs; QC of
the measurement systems, and typical variations in measurement values.
At the completion of the data verification process, the outputs include the verified data and
documentation, or data verification records, indicating which data have been verified and any
technical non-compliance issues or shortcomings of the data, corrections or changes made to the
data, and corrective actions that were taken to address the issues. These corrective actions are
important to ensure that issues or problems with the data do not recur. Note that data which are
non-compliant with technical or acceptance criteria may still be valid and appropriate for
reporting, but should be coded or labeled (qualified) to indicate the nature of the issue(s) with the
data.
10.1.1 Routine (Self) Review
Routine (self) review involves a number of activities that include the site instrument operator,
sample collector, and/or laboratory analyst reviewing the procedures they are performing and the
associated documentation of those activities as they occur or shortly after they occur. At their
most basic, these activities involve establishing calibration curves, verifying proper instrument
operation, and ensuring that DASs are recording necessary information to generate data. Once
the data collection process begins and measurements and observations are recorded, these data
can be reviewed. It is preferable that data review be performed as soon as possible after data
collection so questionable data can be checked by recalling information on unusual events and on
meteorological conditions that can provide context for anomalous data. Also, timely corrective
actions should be taken when indicated to minimize further generation of questionable data.
Recorded data (measurements, observations, etc.) should be reviewed at a frequency that
minimizes the loss of data should an error or condition be found that risks data loss should the
condition or error go uncorrected. For example, if the site instrument operator has configured the
true NO2 instrument to automatically analyze a calibration check standard every week but does
not take the time to review the weekly check for several weeks, such a delay in reviewing the
collected data risks losing a week or more of sampling data in the event the instrument lamp fails
and the calibration check standard does not meet acceptance criteria. Ideally depending on the
measurement system, the individual will conduct a cursory review daily when data are generated,
preferably in the morning, to provide a status of the data and instrument performance at the
monitoring site.
165
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
The site operator or analyst routine (self) review should include reviewing recorded data to
ensure the records are complete and comply with the acceptance criteria in the monitoring
agency SOPs. It is typically most efficient for this individual to make corrections to collected
data or to situations such that the impact of any subsequent problem is minimized immediately.
Such reviews typically cover 100% of the collected data to ensure completeness and that QC
criteria have been satisfied and are within acceptable limits.
This routine (self) review is typically limited in scope to a particular phase of the data collection
activities and is a first step in the overall data verification process, which covers the generation
of data from the "cradle to the grave." The instrument operator should perform routine review of
the collected data as soon as possible after generation to verify (where applicable):
• Measurements did not exceed the alarm limits set in the DAS
• The rate of change observed for the parameter is consistent with ambient data trends
(specific to high frequency measurements - e.g., minute data)
• Measurement data that exceed the instrument calibration range
• Measurement data are complete (sample collection and COC forms are not missing
information, expected electronic files are recorded, and logbook entries are complete)
• Samples/data were collected in accordance with the sample design and approved SOP
• Sample collection and handling procedures were followed correctly
• Data files are properly identified
• Computer file entries match data on hand-entered data sheets
• Analytical procedures used to generate data were implemented as specified
• Instruments were calibrated properly (i.e., before sampling began, at the specified
frequency, included the proper number of points at levels that bracketed the range of
reported results)
• Routine QC checks met acceptance criteria
• Chromatography is acceptable (stable baseline, adequate peak separation, etc.) and that
analyte identification is appropriate based on the established RT windows
• Carbonyls sample holding times and storage conditions were met and the ASL reviewed
and validated carbonyl analysis data
• Deviations from stated procedures or acceptance criteria are documented and impacted
data are flagged or invalidated per monitoring agency policy
• Measurements that are known to be invalid because of instrument malfunctions are
invalidated as per monitoring agency policy
• Data are substituted from a backup in the event of failure of the primary data acquisition
system
• Changes to the data records are documented
166
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
Routine (self) reviews are described in more detail within each of the individual methods
sections for speciated VOCs (Section 10.7), carbonyls (Section 10.8), true NO2 and ozone
(Section 10.9), and meteorology (Section 10.10). Ideally, the instrument operator performs the
routine review as soon as practical after the data are generated. For continuous measurements,
the instrument operator is highly encouraged to review the instrument status and recently
collected data on a daily basis.
10.1.2 Technical Review
Once the data have undergone routine review by the instrument operator, the data are to be
comprehensively technically reviewed by an individual (a peer) not involved with the data
generation. The technical review serves to verify that the routine review was completed properly
and expands the routine review activities. The technical reviewer performs many of the same
activities performed during routine review, but does not verify instrument operation or status in
real time. The technical reviewer verifies correctness of the data generation process by ensuring
that documentation is clear and traceable from the measurement back through to the certified
standards and verifies the data comply with governing SOPs and QAPP. The technical reviewer
will verify (where applicable):
• Measurements below the MDL are flagged appropriately
• Concentration measurements exceeding the instrument calibration range were calculated
correctly and flagged appropriately
• Measurement data are complete (sample collection and COC forms are not missing
information, expected electronic files are recorded, and logbook entries are complete)
• Samples/data were collected in accordance with the sample design and approved SOP
• Sample collection and handling procedures were followed correctly
• Data files are properly identified
• Computer file entries match those on hand-entered data sheets
• Analytical procedures used to generate data were implemented as specified
• Instruments were calibrated properly (i.e., before measurements began, at the specified
frequency, included the proper number of points at levels that bracketed the range of
reported results)
• Calibration standards were within expiration
• Calibration standards and check standards preparation calculations are correct and that
the nominal (known or theoretical) value is input into the instrument, as appropriate
• Supporting equipment to make critical measurements (mass flow controllers, adjustable
pipettes, pressure transducers, etc.) are within calibration and have passed calibration
checks
• Routine QC checks met acceptance criteria
167
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
• Chromatography is acceptable (stable baseline, adequate peak separation, etc.) and that
analyte identification is appropriate based on the established RT windows
• Chromatographic integration is performed correctly and consistently
• Carbonyls sample holding times were met and the ASL reviewed and validated carbonyl
analysis data
• Deviations from stated procedures or acceptance criteria are documented and impacted
data are flagged or invalidated per monitoring agency policy
• Measurements that are known to be invalid because of instrument malfunctions are
invalidated as per monitoring agency policy
• Data have been substituted from a data backup (such as the instrument) in the event of
failure of the primary DAS
• Changes to the data records have been documented and are attributable to the person
making the change
10.2 Data Validation
Data validation is a process that investigates the individual data points within the context of other
co-collected data, historical data, or data collected at a similar location in proximity to the site to
determine the quality of the data relative to the end use. Only after a given dataset has been
verified and validated can it be fully assessed and/or used to address the specific scientific and
regulatory questions embodied in the DQO.
Data validation activities build on the data verification processes described in Section 10.1 and
should not be conducted on data which have not gone through data verification. Data validation
processes may identify data which require further investigation which may include repeating
some steps of the data verification process such as reviewing QC data, calculations, or raw data.
Data validation examines the data set for internal, historical, and spatial consistency:
• Level 1 Data Validation - Evaluates internal consistency of the dataset to identify
values that appear atypical when compared to the values of the entire dataset. Tests
for internal consistency are conducted to identify measurements that do not conform
to expectations - outliers and extreme differences within the dataset that warrant
further investigation. After tracing the path of the measurement, if nothing unusual is
found, the value can be assumed to be a valid result of an environmental cause.
Unusual values are identified during the data interpretation process as extreme values
or outliers. Outliers and extreme differences can be identified and confirmed by the
use of statistical tests, or may be identified by graphical and visual presentation of the
data. Visualization tools (plots, graphs, charts, etc.) are powerful as they allow the
user to quickly identify values that are atypically higher or lower or that do not
conform to a typical or expected pattern, unlike reviewing data in tabular format.
Visualization tools include scatter plots, timeseries plots, or fingerprint plots, among
others, such as those listed in Section 10.4.
168
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
• Level 2 Data Validation - Data that have undergone Level 1 validation for internal
consistency are then compared with historical data to evaluate temporal consistency
of the dataset with previous datasets. The historical data may be recent (e.g., one
week or one month prior) or may cover a longer period (e.g., the previous year or
years). Simple statistical analysis and visualization tools are useful here, as they
enable identification of values that do not conform to expectations.
• Level 3 Data Validation - Data that have undergone Level 2 validation for temporal
consistency may then be evaluated for spatial consistency against data collected at
nearby sites, i.e., those in the same airshed, regional network, or monitoring agency,
to identify systematic bias.
Note: While Level 1 data validation should be conducted on all PAMS data, Level 2 data
validation may not be possible for some of the PAMS Required Sites for certain parameters
during the first PAMS season following implementation in 2019, particularly speciated VOCs
and carbonyls. Similarly, for Level 3 validation, some PAMS Required Sites may not share an
airshed with other monitoring sites or have collocated monitors for some parameters.
Monitoring agencies should perform Level 2 and Level 3 data validation where possible. EPA
Regional contacts may be of assistance to identify nearby sites that can provide data for Level 3
comparisons.
10.2.1 Level 1 Data Validation
Level 1 data validation requires that the dataset has undergone data verification, at which point
the data are presumed to be complete and correct. Data which are believed to be suspect will
have been appropriately qualified or invalidated per the monitoring agency policies and
procedures and the rationale for each instance is documented. Data validation requires
documentation is available for routine reviews, technical reviews, and data manipulations or
changes to data. Data to be validated should be in a common format that permits the data to be
combined and analyzed for graphing and statistical testing. This will typically be a database of
some type or similar information structure that includes the descriptive aspects of the data
including collection times and dates, standard units, qualifier codes, and identifiers of QC data
(duplicates, collocated, field blanks, etc.). Audit reports from internal technical systems audits
(TSAs), instrument performance audits (IPAs), audits of data quality (ADQs) and external TSAs
and IP As should be available to data validators as well as corrective action reports.
1. Data validators should begin by performing simple statistical tests on the datasets by
calculating the central tendency (mean, median, mode), variability (standard deviation),
maximum, and minimum for each parameter.
The central tendency may be calculated as the arithmetic mean, geometric mean, median,
or mode:
a. Arithmetic mean: The sum of the measured concentration values divided by the
total number of samples in the dataset.
b. Geometric mean: The nth root of the product of n concentration values.
169
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
c. Median: The concentration value represented by the midpoint of the dataset when
the concentration values are placed in numerical order. Fifty percent of the
resulting concentration values will be above this value and 50% will be below.
d. Mode: The concentration value with the highest frequency
2. These statistical data should be examined for unrealistic values (extremely high maxima,
large standard deviations, etc.). Extreme values may be identified statistically, such as
values that exceed several (e.g., three or four) standard deviations from the mean or by
employing standard statistical tests designed to identify extreme values. Such extreme
values may be reasonably expected or may be indicative of an underlying issue. Refer to
Section 10.2.1.1 for more information on identifying outliers. Unrealistic data values
should be investigated.
3. Review audit reports for nonconformances or issues that may impact data. Such findings
may vary, but typically require corrective actions to address. Of particular importance are
findings related to calibration acceptance criteria failures, QC check failures, or other
systemic issues that may directly impact the integrity and/or acceptability of data. Audit
reports should reference corrective action reports and indicate when findings have been
resolved.
4. Review the corrective action reports for problems or issues that impact data. This may
include corrective actions that have been addressed with demonstrated return to
conformance or may be corrective actions that remain open and may be actively
impacting data at the time of collection.
5. Utilize data visualization tools to examine data for expected patterns and unexpected
variability. Visualization tools can highlight anomalous data which may not have been
identified by examination of basic statistics or by review of data in tabular or list format.
Visualization tools are discussed further in Section 10.4.1.
10.2.1.1 Identification of Outliers
Outliers are measurements that are extremely large or small relative to the rest of the data and,
therefore, are suspected of misrepresenting the population from which they were collected.3
Outliers may result from transcription errors, data-coding errors, or measurement system
problems such as instrument breakdown. However, outliers may also represent true extreme
values of a distribution (for instance, hot spots) and indicate more variability in the population
than was expected. Not removing true outliers and removing false outliers both lead to a
distortion of estimates of population parameters. The overall premise of data validation is that
data are presumed to be valid unless there is evidence to invalidate the data.
Potential outliers may be identified by assessing data exceeding several standard deviations from
the mean, conducting statistical outlier tests, graphically representing the data, or by reviewing
data summarized with simple statistics (i.e., highest, lowest and average values). Such tests
should be used only to identify data points that warrant further investigation. The decision
whether a data point should be corrected or discarded (invalidated) should be based on expert or
scientific grounds as part of the data validation process. Potential outliers should be documented
in validation notes by identifying the statistical tests performed and the potential scientific
170
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
explanations that were investigated.
It is important to reiterate that data should be invalidated only if there is sufficient evidence to
show that the value(s) are not real/correct. If there is insufficient rationale to invalidate a data
point, but in the opinion of the validator the value is suspect, the data may be appropriately
qualified according to the monitoring agency policy. Monitoring agencies are encouraged to
discuss such situations with their PAMS Regional contact.
10.2.2 Level 2 Data Validation
Utilize simple statistical and data visualization tools to perform temporal comparisons of the
data, where possible. Plotting data from the current dataset against data from a historical dataset
from the same site may identify values that did not stand out as extreme values or outliers in
Level 1 assessment or may identify step changes or other anomalies indicative of measurement
system changes, drift, or performance degradation.
10.2.3 Level 3 Data Validation
Utilize statistical tests and data visualization overlays from two different sites within the same
airshed or collocated instruments to investigate significant differences between the sites for
similar parameters. Examine differences for systematic bias or expected differences. Similarly
impacted sites should indicate reasonably coordinated behavior. Collocated instruments should
indicate similar concentrations and concentration changes within an expected tolerance. Such
collocations could be two different methods that evaluate the same parameter, e.g., benzene
concentrations from a 24-hour canister collection compared to the same collection period with an
auto-GC.
10.3 Reporting of Validated Data to AQS
After the data validation has been completed minimally through Level 1, the data may be
uploaded to AQS (refer to Section 11 for data upload to AQS). Prior to upload, the data validator
should verify flagged data have been qualified appropriately, which may involve performing
parity checks on the data translated into AQS format and performing spot checks on the data.
Monitoring agencies are encouraged to have an independent reviewer verify data have been
appropriately coded for AQS submission. Such verification checks should be documented.
Once reported to AQS, the monitoring agency should query AQS to verify the data were
uploaded properly and perform parity checks to verify there are no discrepancies. These
verifications should be documented.
Following upload of data to AQS, users may generate a report from AQS indicating the rank of
reported data for specific parameters in relation to historic values at the site. Such reports may be
maintained and referenced in Level 2 data validation activities.
10.3.1 Reporting Values below Method Detection Limits
Instrument sensitivity for the PAMS Required Sites is characterized by determining the MDL as
described in Section 3.3.5.1. The MDL for each parameter represents the lowest concentration
171
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
that can be detected above background with a 99% false positive rate. Given that the false
negative rate for MDLs is 50%, concentrations measured at less than the MDL, so long as the
qualitative identification criteria have been met (analyte is positively identified), are valid and
are necessary for properly performing trends analysis. Substituting values (such as one-half
MDL) or censoring (reporting as 0) concentrations measured below the MDL is not permitted.
EPA recognizes that many monitoring organizations are not comfortable reporting
concentrations measured less than the MDL as these concentrations are outside of the calibrated
range of the instrument and are associated with an unknown and potentially large uncertainty.
However, values as actually measured (even when below the MDL) are more valuable from a
data analyst's standpoint and far superior than censored or substituted values. Addition of
qualifiers as described in Section 11.5.1 and in Table 11-3 indicates proximity of reported
concentrations to the MDL and communicate the level of associated uncertainty to the data user.
10.4 Data Validation Tools and Methods
The following sections describe general tools for conducting data validation for PAMS Required
Site data. These tools are useful in identifying anomalous data and increasing confidence in
datasets; however, validators should use a combination of such tools to validate data, and not
rely on one specific tool to confirm or nullify data validity. As mentioned previously, each
monitoring agency should describe the PAMS Required Site data validation process and tools in
an SOP or similar controlled document.
10.4.1 Data Validation Visualization Methods
Graphical techniques permit comparison of concentrations of each PAMS parameter to the
expected concentrations and relative concentrations of other datasets to inspect for values which
stand out. These graphical techniques can combine and contrast different parameters temporally
and spatially to help accentuate data which may stand out from the dataset and warrant further
investigation. Some of the simplest of these graphical tools are available in the Data Analysis
and Reporting Tool (DART) and include time series plots, scatter plots, fingerprint plots, and
stacked bar charts.
• Time Series Plots: Concentrations are plotted on the y-axis against collection date
(time) on the x-axis over extended time increments (e.g., four to 12 weeks). Extreme
or anomalous values are immediately identifiable in individual plots, and may be
more powerful when multiple related parameters are plotted together. Pollutants that
are typically emitted from the same type of source (i.e., benzene and toluene from
mobile sources) and from different sources (i.e., formaldehyde and NO2) can provide
insight on whether concentration anomalies are realistic to the collected sample or
may be an artifact of the collection or analysis of the sample. Related parameters can
also be plotted together, for example, to allow the data validator to examine whether
changing meteorological conditions (e.g., wind speed/direction, rain) could explain
unusual behavior in a given PAMS parameter. Time series plots are also useful for
locating unusually high changes in the data from one value to the next (peaks), long
periods of constant or no change (i.e., "sticking"), and general trends. For example, a
172
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
slow increase or decrease in concentration over time in a single pollutant may
indicate an instrument problem. The spike in ethane concentrations on August 3, 2015
shown in Figure 10-2 exhibits an obvious difference in the concentration of ethane for
that day relative to the rest of the three-month period.
Time series plots are also useful for investigating baseline changes. Longer time-
periods of data (e.g., one year) are plotted on smaller y-axis scales (e.g., 0 to 10 ppb
for ozone, 0 to 15 ppb for NO2), allowing the data validator to look for step functions
(abrupt shifts downward or upward - such may indicate a reset of the zero offset for
an ozone analyzer) or gradual drift in the baseline resulting from improper
maintenance, postprocessing of the data, etc.
Data visualized in a time series plot may also accentuate missing data points within
the dataset. Validators should closely examine periods of missing data to verify the
omission is intentional, such as would be the case, for example, for instrument or
hardware malfunction or invalidation due to QC failures.
VAX
1
hnrrr
«L A .A 1 i J K
u JV. A. i* Ll
T LjW-Vf Jr\, f
/1/15 6/8/15
6/15/15 6/22/15 6/29/15 7/6/15
7/13/15 7/20/15 7/27/15 8/3/
15 8/10/15 8/17/15 8/24/15 8/31/1
Figure 10-2. Time Series Plot of Ethane
• Scatter Plots: Scatter plot matrices provide a convenient means of identifying
relationships among variables. Concentrations of pairs of parameters are plotted such
that each species (e.g., benzene and toluene) is dedicated to the y-axis or x-axis such
that the coordinates of each plotted point are set by the chosen species pair
concentrations measured during a given sampling event. For parameter pairs that are
correlated, the resulting plots generally show points that are clumped together,
showing a well-defined relationship. In Figure 10-3, propane and TNMOC are
graphed together and a regression line shows the points clumped around the line,
indicating a general trend that propane and TNMOC typically increase in
concentration together. Points that lie outside of the well-defined area are then
generally identifiable and can be further investigated.
173
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
Total MMOC (Non-Methane Organic Compound)
Figure 10-3. Scatter Plot of Propane and TNMOC
• Fingerprint Plots: Concentrations of all pollutants within a given class (e.g., VOCs,
carbonyls, etc.) are plotted on the y-axis against the molecular weight, alphabetical
order, or some other consistent order on the x-axis that enables discerning patterns or
identifying anomalies. Typically, bar charts are used to produce fingerprint plots.
Fingerprint plots prepared for each sampling event will typically be very similar
among events. The fingerprint plots in Figure 10-4 show PAMS target VOCs
organized by alphabetical order and demonstrate VOCs concentration patterns. Notice
for these two samples measured at 19:00 one month apart that the general pattern and
relative concentrations of measured VOCs is similar with ethane being the most
abundant VOC. Plots that show markedly different patterns may indicate anomalous
results. For instance, examination of the VOCs fingerprint plot during a specific
sampling event may reveal that one or more VOC was observed at a concentration
much higher or much lower than expected given the typical pattern; such a result
would warrant further investigation of the individual chromatogram or entire
sampling event. Fingerprint plots may be useful in confirming diurnal patterns such
as the increase in isoprene concentrations during warm daytime periods with
decreases in isoprene concentrations overnight.
174
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
VAX
+
la
i. , i, i,i
'iHUffif!f|lf{| i ii
I!
07/01/2015 19:00
VAX
+
¦! I
¦'"lljfniii
in
¦ ¦ ¦ ¦ ¦ ¦
i!ljff!iii!ii!{i!,l!,ii,li
08/01/2015 19:00
Figure 10-4. Fingerprint Plots of PAMS Target VOC Analytes
• Stacked Bar Charts: Stacked bar charts allow combini ng concentrations of target
analytes. Typically, discrete samples or combinations of samples are shown on the x-axis
chronologically and concentration is shown on the y-axis. The chosen target analytes are
stacked on top of one another in a set configuration where the total bar height is the sum
concentration of the individual analytes that are distinguished as different colors within
the bar. Figure 10-5 illustrates a stacked bar chart for a two-day period for ethane,
propane, n-butane, and n-pentane. This chart indicates a potential data issue for ethane on
July 31, 2016; ethane is at fairly high concentrations relative to the other three
compounds around sunrise for July 29, 2016 and July 30, 2016, but is missing from the
06:00 and 07:00 hours on July 31, 2016. It can be seen from the chart that the total
concentrations of these target compounds exhibit a diurnal pattern with the highest
concentrations occurring between the 04:00 and 08:00 hours.
175
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
STACKED BAR CHART
+
,11
[i!uil ¦¦!?---¦
555555535555555555555555555535559999^999999^9!^^!^^^^^^^^^
BKBBKBBBBSBBBBB8B8BBBBSBlsSSlggllsggSggggggg|gSg=2222222222S2==22222a=22
12E22221121122 2111 222222 221112 21111111222 211122 2ESSSSSESSESSSSSSESSE ESSS
cnoiai(Jioi
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
of the various statistics provides dense information graphically. It is immediately
evident that concentrations measured in Chicago, Illinois during this time period are
on average lower and are more tightly clumped than those at the other sites.
Formaldehyde (2013-2015)
15
10
0.1 ug/m3
Atlanta, GA Boston, MA Chicago, IL Houston, TX Sacramento, CA San Diego, CA Washington, DC
(Northbrook)
Site
Figure 10-6. Example Box Plots of Formaldehyde Concentrations at Seven Sites 4
• Diurnal Profiles: Many of the PAMS pollutant species have well-defined diurnal
cycles that are related to source activity (e.g., traffic) and meteorological patterns
(e.g., mixed layer height, solar radiation). Diurnal profiles can be used to compare the
mean daily cycle of several different variables between different monitoring sites,
different years, or weekend vs. weekday. Diurnal profiles are prepared by plotting the
parameter concentration or other parameter magnitude against the measurement time
of day.
• Pollution Rose: Pollution rose plots show concentration by wind direction and can be
useful in establishing trends in pollutant emissions/sources, and can also be prepared
on subsets of data (e.g., daytime versus night or by season). In addition, major
changes in wind rose plots (wind speed and frequency as a function of wind direction)
can help explain unusual pollutant patterns.
Confidence is increased for measurement data that fit the general trend or expected pattern and
do not appear anomalous when plotted using graphical tools such as those described above. Data
that appear to be anomalous should be flagged for follow up investigation.
177
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
10.4.2 Data Validation Tools
Tools for conducting data validation will include features included in instrument-specific
software, such as the CDS Chromatotec VistaCHROM, Agilent OpenLab, and PerkinElmer
TotalChrom software packages. Data acquisition systems (e.g., DRDAS and AirVision) may
include features for calculating summary statistics, creating timeseries plots, scatter plots, etc. In
addition, standard spreadsheet software such as Microsoft® Excel may be used for calculating
summary statistics and creating basic plots, whether manually or through creating macros.
Commercially-available software systems that perform data validation functions are presumed to
have been validated by the vendor or manufacturer; however, custom-built validation algorithms
or spreadsheet programs should be verified for accuracy prior to use and locked to prevent
inadvertent corruption as described in Section 9.1.1. Such custom validation software or
algorithms should be revalidated when changes are made to ensure proper function.
The free DART software was developed with EPA funding and is integral for validation of
measurement data for the EPA's PM2.5 CSN. DART incorporates many of the preparation of the
graphical displays mentioned above and is available at airnowtech.org at the following URL
(users must have an account with username and password):
http: //aim owtech. org/dart/dartwel com e. cfm
Users can upload datasets to DART or may query AQS through DART to analyze and screen
many types of air quality data, including criteria pollutants, VOCs, etc. DART provides tools for:
• Uploading data files
• Making data requests to AQS
• Performing unit conversions
• Aggregating data
• Creating time series graphs and editing data
• Creating scatter plots
• Creating bar charts (fingerprint plots)
• Performing screening checks
• Exporting data and summary statistics
• Adding qualifiers to data
• Preparing data for submission to AQS (generating AQS ready transaction strings)
In addition to including specific screening checks for PAMS VOCs data (discussed in Section
10.7), five general screening options are available in DART:
• Species Threshold - identify data values that exceed a user-defined threshold
concentration
• Species Variability - identify data within a specified variability (e.g., to identify data
greater than twice the standard deviation, enter a 2 for the criterion)
178
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
• Species Comparison - compare data values between parameters according to user-
defined criteria
• Species Fraction - identify data values that are within a user-defined specified fraction
of another data parameter value
• Multi-Condition - create data screening queries that meet more than one condition.
Step-by-step instructions for using DART for PAMS data validation are available at:
https://vocdat.airnowtech.org/documentation/Content/UserGuide Jan! DFforDARTUG.pdf
As of this document's publication, DART did not include functions for uploading or validating
meteorological data; however, such a capability may be added in the future.
10.5 Data Verification and Validation Records
Observations and activities conducted during initial instrument operator routine self reviews,
technical reviews, and data validation should be comprehensively recorded and maintained such
that these activities can be reconstructed. Such records include site operator logs, data
reviewer/validator checklists and notes, email communication between reviewing/validating staff
and operators, outputs from data validation tools (such as charts, plots, and regressions), and data
approvals, among others. These records should be stored for ready access.
10.6 Data Flagging
Instrument operators, technical reviewers, and data validators can mark data as suspect or
compromised at many points during the generation, review, and validation processes. Instrument
software systems can be configured to automatically add flags to data that do not meet certain
default or user-defined criteria. Instrument operators, data reviewers, and validators should
review these automatically flagged data and may add additional flags as appropriate as the data
move through the validation process to final reporting.
As part of the completion of data validation following data verification and validation activities,
the data validator should ensure that compromised or invalid data are appropriately flagged when
uploaded to AQS. Data validators should reference site operator logs and notes from review, data
reviewer notes, and data validation notes to verify the flags are appropriate per the defined
guidelines.
10.7 Data Verification and Validation of Speciated VOCs
Measuring VOCs in the atmosphere on a daily and hourly basis with auto-GC systems produces
extremely large and complex datasets. Managing, processing, and validating the data requires
technical expertise and intensive effort to obtain reliable and consistent data for timely input of
the data into the AQS database. It is of primary importance that monitoring agencies plan and
practice the data collection, manipulation, and storage processes for auto-GC data prior to
beginning data collection for reporting data to AQS. During this planning stage, monitoring
179
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
agencies can make adjustments to the systems and procedures and identify problems that occur
or are likely to occur. Instrument configurations which permit remote access to the instrument
computer and DAS are strongly recommended as they permit instrument operators or designees
the ability to review system data and status as well as control instrument function without the
need to be on site. Additionally, monitoring agencies are encouraged to quarantine raw and
processed data files such that they cannot be overwritten when it is necessary to reprocess the
collected data.
10.7.1 Speciated VOCs Data Sources
10.7.1.1 Calibration Data
Prior to examining ambient data, the auto-GC calibration will be verified to have met criteria as
part of the data verification process. Records needed to verify proper calibration include the
certificate of analysis for the primary calibration stock as well as documentation detailing
subsequent measurements for dilutions of the primary calibration stock and the associated
calibrations of instruments involved in the dilution, e.g., pressure gauges for static dilution or
MFCs for dynamic dilution. Subsequent calculations to generate the calibration curves, whether
hand-calculated or manipulated in an electronic spreadsheet, should be available for review.
Nominal concentrations entered for generating the calibration curves, as well as the calibration
linear regression, intercept, and comparison to nominal concentration; response factors, and the
instrument quantitation method should be available to ensure they are correct and traceable.
Chromatograms and CDS reports described in the following sections should be available for
review.
10.7.1.2 Auto-GC Reports and Datafiles
Auto-GC CDSs typically include the ability to generate the following data reports, and may
provide for custom reports based on user-defined criteria:
• Chromatogram - Graph of the instrument response per unit time during the GC run.
There will be one chromatogram file per FID per sample hour. For example, for the 10:00
a.m. sample collection, there will be one chromatogram for the light HC channel FID
(PLOT column) and one chromatogram for the heavy HC channel FID (PDMS column).
• Result file - List of the target analytes to be analyzed included in the analytical method.
Depending on the GC system, there may be two results files, one for each FID, or the two
channels may be combined into a single report. Such reports typically show the following
information:
o Header
o Filename
o Sample collection start time
o GC acquisition start time
o Table of results by each compound listed in retention order
o Compound name
o Retention time
o Area response
o Calculated concentration
180
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
o Data flags - may indicate missing compounds, manual integration, or
concentrations greater or less than a specific user-defined threshold
• Summary report - Such reports may include reports detailing the data files collected
during a defined period (e.g., 24 hours), QC reports showing daily CCV, daily system
blank, and weekly precision checks with associated acceptance criteria.
10.7.1.3 Chromatographic Data File Identification
Chromatographic data files will be clearly and correctly identified to indicate the correct
acquisition time and date, sampling location (e.g., monitor address, site name, or AQS site
identification), sample name or type (e.g., ambient, CCV, calibration standard, blank, precision
check, etc.), processing and calibration methods, and will conform to the established file naming
convention. As discussed in Section 4.2.5, monitoring organizations should plan carefully to
establish a file naming and organization convention and structure that ensures file names are
unique, are indicative of the type and timing of the sample data, and indicate whether a datafile is
the original file or has been reprocessed.
Chromatographic files can be misidentified due to: incorrect sampling locations, especially if
instrument method files are copied from one site location to another; incorrect date and time
stamp due to daylight savings time change; or sample identified as an ambient sample which is a
blank check or other QC sample. Site operators should perform a cursory review of file
identification data during the routine checks; however, technical reviewers and data validators
should also be reviewing file identification information.
10.7.1.4 Auto-GC Chromatograms
Due to the large volume of data generated from PAMS monitoring, it is not practical to closely
scrutinize all chromatograms and result reports for the ambient sample data. Auto-GCs making
hourly measurements for 59 compounds will generate approximately 1300 concentration data
points each day. While such detailed review of each measurement and associated chromatogram
is not practical, review of the collected data can be performed efficiently to identify problems
and provide the necessary level of confidence in the collected data.
All chromatograms should go through a cursory review by the auto-GC operator to determine if
the quality of the chromatography is acceptable. This includes examining the appearance of the
chromatogram, peak shape, peak resolution, peak integration, retention times, and baseline.
Chromatogram review will verify the GC(s) is performing properly and can be accomplished
quickly by an experienced chromatographer. Preparation of overlays of chromatograms, such as
overlaying ambient sample data with a nightly CCV/RTS can simplify the chromatogram review
process.
The cursory review of chromatograms should include verifying the following:
• The signal from the FID or baseline is normal and the signal output is positive (onscale);
• Chromatographic peaks are present as expected, integrated correctly, and the peak-shape
is sharp and reasonably symmetric;
181
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
• The peak resolution or separation is acceptable based on historical instrument
performance;
• The CDS has not missed known target analyte peaks - possibly due to area threshold
settings or RT shifts
• All components have eluted from the analytical column as indicated by a flat or normal
baseline at the end of a run; and
• No chromatographic abnormalities exist, such as large contamination or non-target
coeluting compound peaks, electronic spikes, or atypical baseline behavior.
Once the chromatograms have been reviewed and found to be acceptable, further review of the
peak identification, peak integration, and other data checks may be performed. Based on the
review of the chromatograms and associated data reports for appropriate peak identification and
integration, identify any necessary corrective actions, such as altering one or more RT windows
in the method or updating an integration parameter. Perform manual integration as needed,
noting that the need to perform numerous manual integration steps should prompt the instrument
operator to adjust the automated integration parameters to minimize the need for manual
integration. As needed, update the acquisition method and reprocess impacted sample
chromatograms.
Examination of Low Concentration Data: Many target VOCs will typically behave well
chromatographically, providing good peak shape, few interferences, and sufficient peak area to
be readily and reliably integrated by the CDS auto-integration parameters. For such compounds,
review typically includes verifying the correct peak was identified (RT is within the assigned
window), that there were no interfering coelutions included in the integrated peak area, and that
peak integration was performed correctly. Conversely, for compounds which are typically at
much lower concentrations (e.g., approaching the MDL) and eluting at similar times to
interfering peaks, the CDS may have difficulty correctly identifying the analyte peak from
baseline noise or from an interfering peak with a similar RT that falls within the assigned RT
window. If instrument area reject thresholds are set too high, target peaks may not be identified
and the CDS will indicate the compound was not detected. Even when identified correctly,
automated integration parameters may not be optimized to properly integrate such peaks.
Each monitoring agency should strive to balance the amount of time spent on low concentration
data (i.e. those less than 0.5 ppbC) with the amount of resources available for technical data
review and validation. CDS integration parameters should be configured to optimize the proper
identification and integration of low concentration peaks, as possible. Even with optimization,
integration parameters will require adjustment as the chromatography systems change with time.
As a guideline, monitoring agencies should evaluate the proper identification and integration for
target analyte peaks greater than 0.5 ppbC or three times the determined MDL, whichever is
lower. Monitoring agencies are encouraged to review the identification and integration of target
analyte peaks below this threshold, as resources permit. The approach, procedures, and details of
how low concentration data are addressed should be prescribed in the agency quality system.
182
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
10.7.1.5 Instrument Maintenance and Site Logbooks
Each monitoring site should operate an on-site logbook or combination of logbooks (whether
hard copy or electronic log system) to record information relative to instrument repair and
maintenance (e.g., replacing a preconcentrator trap, trimming or replacing a separation column)
and unusual events (landscaping activities, power outages, or construction or repairs) at the site.
Auto-GC operators and technical reviewers should examine the site and instrument logbooks for
notes on events that may impact sample data which may explain variations or excursions in the
data and are critical to addressing missing data files, high sample concentrations, or other
measurement anomalies during subsequent data review and validation.
10.7.2 Speciated VOCs Data Verification Procedures
Data verification consists of routine auto-GC operator checks and technical review by an
individual intimately familiar with, but not responsible for, instrument operation and data
collection. Procedures specific to the auto-GC operator and technical reviewer are detailed in the
following sections. Aspects of data verification common to both the auto-GC operator and
technical reviewer include verifying that expected data files are present, that RT windows are
appropriate, that QC samples meet criteria, and that target analyte peak identification and
integration are correct and appropriate.
Auto-GC operators and technical reviewers should examine the collected CDS data files for each
day to verify all expected files are present. Missing data files may be indicative of an instrument
failure and should be investigated to discern the impact of the missing sample data on samples
collected following missing sampling hour(s). For example, instrument failures commonly result
in the failure to purge the preconcentrator trap and delay the instrument to be ready for the next
hourly sample. Instrument failures may also be due to the inability of the instrument to complete
the GC program. In both cases the auto-GC system may measure analytes collected during a
previous hour's sample and these suspected hours of data should be invalidated.
Once the auto-GC operator or technical reviewer have verified all intended data files are present,
they should review the data results file (the summary sample report for each hourly sample
generated by the CDS - refer to the instrument SOP for further information on generating such a
report) in conjunction with the associated chromatogram to ensure that key reference compounds
are correctly identified and that the target analyte peak RTs have not shifted outside of the
assigned windows. Summary reports listing the sample collection data with associated filenames
and timestamps can serve as a starting point to verify the analytical sequence was correctly
programmed and carried out. Summary reports for the daily CCV and system blank as well as
the weekly precision check allow the auto-GC operator, technical reviewer, and data validator to
quickly ascertain that these QC samples met acceptance criteria. The CDS may permit
assignment of acceptance criteria to automatically flag results which do not meet criteria.
Summary reports covering an entire day's analysis should include the samples collected since the
most recent QC samples and include those ending QC samples bracketing the sequence for the
day.
Auto-GC operators and technical reviewers should review the data results file to ensure that peak
assignments or identifications are correct and that the resulting concentrations are as expected.
The information is also reviewed to determine whether established RT windows require
183
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
updating. Some CDS permit the assignment of reference compounds which serve as anchors to
automatically adjust the RT windows for target analytes associated with the specific reference
compound as described in Section 4.2.3.6. In some cases, changes in humidity in the ambient air
due to rain events or other weather patterns can cause RT shifting, particularly for the light HC
channel for instruments with Nafion™ drying systems. These RT shifts can be transient and may
only last a few hours before returning to the previous typical conditions. Reference peaks can
help to maintain proper peak identification during such events. Additionally, calculating the RSD
of the RTs for each compound over the course of day or several days may identify such events
that are not immediately apparent. Elevated RT RSDs indicate increased variation in RTs for the
affected target analytes, and in such cases the reference peak identifications should be reviewed
in the chromatograms to verify they and the associated target analytes have been correctly
identified. When RT shifts are transient (meaning that they return to previous conditions after an
event), the RT windows should not be altered, but the identification should be manually assigned
with the analyst's judgment, where possible, in compliance with the procedures prescribed in the
agency quality system. Conversely, when RTs shift and the shift is not transient, RT windows
may need to be updated within the data processing method, as indicated by peak identifications
missed or incorrect peak identifications. The need for updating peak identification information in
the acquisition method is indicated by the frequency of missed or inaccurate peak identifications
automatically made by the GC system per the discretion of the analyst.
10.7.2.1 Correcting Chromatography Data
As possible, the CDS integration parameters should be optimized to properly identify and
correctly integrate target analyte peaks. However, given the nature of the auto-GC systems, the
number of target analytes, and the numerous combinations of interferences, corrections to
chromatography data will be required. Site operators will primarily be responsible for adjusting
chromatography parameters including RT windows, integration parameters (such as bunching
factors, smoothing factors, slope sensitivity, tangent skimming, etc.), and reprocessing data
according to an updated quantitation method.
Adjusted automatic integration parameter methods may address most of the peaks in the
chromatogram, but there will likely be some target analyte peaks that may not integrate properly
and will require manual integration. In such cases, the integration should comply with a detailed
SOP prescribing integration actions for specific situations. Most of the manual integration
needed should be addressed by the site operator (analyst), although technical reviewers and data
validators should be given authority to make appropriate changes when warranted. For target
analyte peaks that demonstrate coelutions and are difficult to integrate properly, the target
analyte concentration should be qualified as "LJ" to indicate the analyte was properly identified,
but the concentration is estimated. If the bias of the estimate is known, the concentration should
be qualified as "LL" or "LK," respectively for concentrations with a low or high bias. For
example, a low bias is expected when two peaks coelute and a vertical line is needed to separate
the peaks. The expected peak area on the peak front or tail is eliminated in such cases and would
be expected to underestimate the concentration.
10.7.2.2 Routine Auto-GC Operator Checks
Auto-GC operators are strongly encouraged to check on the auto-GC operation status and review
the most recent QC data each morning to verify proper instrument operation and that data were
184
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
acquired and recorded successfully. Such frequent checks on the instrument and data outputs will
catch issues and allow timely corrective actions, limit the amount of data affected, and reduce the
level of effort required to verify and validate speciated VOCs data. Though not a routine daily
check, the auto-GC operator should be reviewing the instrument initial calibration immediately
after it is established to ensure the calibration meets the acceptance criteria and that the
calibration is verified with the SSCV.
As part of the routine checks, the auto-GC operator should verify the instrument preconcentrator
and GC statuses are correct (e.g., sampling, analyzing, or waiting for the next sample to begin)
and that the clock is accurate. If the instrument status is not correct as expected, the operator
should immediately investigate the root cause of the problem, which may include reviewing the
most recently collected data. Causes of instrument failure are too numerous to list here, and may
be of a simple nature where failures can be corrected remotely through the CDS or may require a
site visit to further investigate. Please refer to the instrument SOP for further information on
diagnosing and correcting auto-GC instrument problems.
Once the instrument status is verified to be correct, or as part of troubleshooting a problem, the
operator should examine data collected since the last routine check. Some CDS incorporate
customizable summary reports to provide a snapshot for examining data files. Specific checks
may include examining datafiles for abundant species - target analytes which should always be
detected (such as propane, butane, benzene, etc.) - in ambient air samples, retention time
precision, instrument error messages or flags, TNMOC concentrations for reasonableness, and
QC data for CCV recovery and blank criteria. Monitoring agencies are strongly encouraged to
develop and utilize such summary reports within the CDS to streamline routine checks. During
these routine checks, operators may also choose to address issues within the data such as
adjustment of retention time windows or integration parameters, as needed. All such routine
checks and actions taken during these routine checks should be thoroughly documented and
attributable to the individual making the change so that later validation activities can reconstruct
the activities if data require further examination. Many DAS software systems and CDS
incorporate electronic logbooks or audit trail capabilities that can be utilized to capture data
processing and data review actions and observations.
10.7.2.3 Technical Review of Speciated VOCs Data
Due to the large volume and complexity of the hourly speciated VOCs datasets, monitoring
agencies are strongly encouraged to perform timely review of the data on a frequent regular basis
(e.g., daily or every other day) to ensure that data review is manageable and identifies issues
before they have a large impact on data completeness. Monitoring agencies are strongly
encouraged to develop an SOP and a checklist for conducting technical data review to ensure
review is comprehensive and to document the review process. Technical reviewers should be
notating issues, anomalies, or errors in the data to verify that appropriate corrections are made
when warranted and that compromised data are appropriately qualified or invalidated when
reported to AQS.
Technical data review should incorporate many of the aspects performed by the auto-GC
operator during routine checks such as: ensuring the expected number of datafiles are present and
that QC data meet criteria, ensuring instrument flags or alerts have been addressed, checking that
185
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
common target analytes are detected, reviewing retention times, and verifying sample collections
have started at the proper time. In addition to these checks, technical reviewers should perform a
more in-depth review of the collected data and associated records starting with completely
verifying the instrument calibration and closely reviewing all associated calibration records as
well as site and instrument logs for completeness and unusual events, target analyte
identification, chromatograms (for automatic integration, unusual chromatography, etc.), data file
naming, and reprocessed data (manual integration, retention time window adjustments,
traceability documentation for site operator reviews and actions, and audit trails).
Technical reviewers should perform screening checks which verify all expected data are present,
sample collection times are within specification, data have been properly handled for unusual
events, as well as verify presence/absence of specific compounds, perform deterministic
comparisons, and examine general trends in specific compound behavior. Deterministic
relationships examine relative amounts of target analytes and parameters and compare those to
the expected ratio or relationship. Appropriate screening checks and the associated follow-up or
data treatment actions are described in Table 10-1.
Table 10-1. Speciated VOCs Data Screening Checks
Screen inii Piiriimclcr
Dcliiils
Recommended l-'ollow-iip or Diilii
Trciilmcnl Action
Abundant Species
Verify compounds typically measured
at the site (such as: ethane, benzene,
propane, n-butane, iso-butane, iso-
pentane, n-pentane, isoprene, n-hexane,
toluenes, xylenes, and ethylbenzene)
are present and measured (not 0 ppbC)
Investigate chromatogram for incorrect
peak identification (retention time shifts),
improper integration. Correct as
appropriate. As applicable, if two or more
such compounds are missing and cannot be
corrected, invalidate all compounds for that
FID channel for the hourly sample.
TNMOC
Verify TNMOC concentration is
greater than 0 ppbC
Investigate chromatogram for incorrect
peak identification (retention time shifts),
improper integration. Correct as appropriate
and invalidate ambient sample hours for
which no TNMOC is measured.
TNMOC
Verify unidentified compounds total
concentration is < 15% of TNMOC
Examine chromatogram for missed peak
identifications and unidentified peaks.
Verify integration of unidentified peaks is
appropriate and that instrument noise or
baseline anomalies or rise is not integrated.
Correct as appropriate.
TNMOC
Verify TNMOC exceeds sum of
PAMSHC (total of all identified
compounds)
Instances of PAMSHC exceeding TNMOC
should be exceedingly rare. Verify TNMOC
calculation is not corrupted. If PAMSHC is
not < TNMOC, qualify TNMOC as biased
low ("LL").
Individual Compound
Variability
Compound concentration in a given
sample hour exceeds four standard
deviations of the historical mean
This check identifies potentially high
concentrations that should be investigated
further. Examine chromatogram of such
samples and those preceding and following
for chromatographic artifacts, interferences,
or misidentification.
186
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
Table 10-1 (continued). Speciated VOCs Data Screening Checks
Screening Piiriimclcr
Deliiils
Recommended l-'ollow-iip or l);ii;i
Trciilmcnl Action
Sticking
Species has same non-zero value for
three or more consecutive sample hours
(this is unlikely to occur typically)
Review affected sample data. This
condition may indicate consistent
contamination within the instrument such as
occurs with aging Nafion™ dryers or
preconcentrator trap contamination. If
contamination is confirmed, invalidate
affected target analytes as "SC."
Benzene : Toluene
If benzene concentration exceeds both
3x MDL and toluene concentration
(toluene should have a higher
concentration)
Ensure correct identification. Flag benzene
and toluene as "LJ"
Benzene : Ethane
If benzene concentration exceeds both
3x MDL and ethane concentration
(ethane should have a higher
concentration)
Ensure correct identification. Flag benzene
and ethane as "LJ"
Ethylene : Ethane
If ethylene concentration exceeds both
3x MDL and ethane concentration
(ethane should have a higher
concentration)
Ensure correct identification. Flag ethylene
and ethane as "LJ"
Propylene : Propane
If propylene concentration exceeds
both 3x MDL and propane
concentration (propane should have a
higher concentration)
Ensure correct identification. Flag
propylene and propane as "LJ"
o-Xylene : m/p-Xylene
If o-xylene concentration exceeds both
3x MDL and m/p-xylenes
concentration (m/p-xylenes should
have a higher concentration)
Ensure correct identification. Flag all
xylenes as "LJ"
2-Methylhexane : 2,3-
Dimethylpentane
2-methylhexane concentration should
exceed 2,3-dimethylpentane
Review chromatography for interference,
proper identification, and integration (no
qualification needed)
Methylcyclopentane :
2,4-Dimethylpentane
Methylcyclopentane concentration
should exceed 2,4-diemthylpentane
Review chromatography for interference,
proper identification, and integration (no
qualification needed)
Pentanes
Pentanes should show in the following
order of decreasing concentration:
isopentane, n-pentane, cyclopentane
Review chromatography for interference,
proper identification, and integration (no
qualification needed)
n-Butane : iso-Butane
n-Butane concentration should exceed
iso-butane concentration
Review chromatography for interference,
proper identification, and integration (no
qualification needed)
187
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
Table 10-1 (continued). Speciated VOCs Data Screening Checks
Screening Piiriimclcr
Del nils
Recommended l-'ollow-iip or l);ii;i
Trciilmcnl Action
Methylpentanes
If 3-methylpentane concentration
exceeds both 3x MDL and exceeds 0.6
times the 2-methylpentane
concentration
Note: 2-Methylpentane and
3-methylpentane elute closely with
2,3-dimethylbutane, as the "terrible
trio. " In general, 2-methylpentane
should have the highest concentration
in ambient air followed by
3-methylpentane then
2,3-dimethylbutane.
Flag 2-methylpentane and 3-methylpentane
as "LJ"
Trimethy lbenzene s
1.2.3-trimethylbenzene will typically
have the highest concentration of the
three.
1,3,5-trimethylbenzene and/or
1.2.4-trimethylbenzene should not
exceed 1,2,3-trimethylbenzene in
ambient air
Review chromatography for interference,
proper identification, and integration (no
qualification needed)
n-Undecane : n-Decane
If n-undecane concentration exceeds
both 3x MDL and n-decane (n-decane
concentration should be higher)
Flag n-undecane and n-decane as "LJ"
Olefins : Paraffins
If sum of olefins concentrations
exceeds sum of paraffins
concentrations (paraffins concentration
sum should be higher)
Flag all olefins and paraffins as "LJ"
Nighttime isoprene
Isoprene should not increase in
concentration between 8 pm and 3 am
local time
Isoprene should show a diurnal pattern.
Decreases for hourly daytime samples
can indicate integration error.
Verify instrument and/or DAS clock,
datafile timestamps, and chromatography
for misidentifications. If clocks and
timestamps are correct and isoprene is
correctly identified and integrated, flag
isoprene as "LJ" in samples between 20:00
and 03:00.
Decane and undecane
If decane and/or undecane are present
after nightly QC checks, look for
carryover effects indicated by
decreasing concentrations in
subsequent ambient hourly samples.
If evidence of carryover is confirmed for
these compounds, invalidate as "SC."
188
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
Table 10-1 (continued). Speciated VOCs Data Screening Checks
Screening Piiriimclcr
Deliiils
Recommended l-'ollow-iip or l);ii;i
Trciilmcnl Action
Missing sample data
The instrument did not capture an
hourly ambient sample for one or both
FID channels
Invalidate compounds from the affected
channel missing concentration data as
"AF."
Review data from collected sample hour
immediately following sample with missing
data. If problem affects more than one
sample hour, review sample hours through
next CCV. Invalidate analytes for sample
hours which show problems with "AN" or
other appropriate null code.
Sample collection time
Sample collection start time began
more than 10 minutes before or 30
minutes after the hour, possibly due to
wandering instrument clock, DAS
clock, computer clock, or problem
resulting in delayed instrument "ready"
condition.
Invalidate all target analyte concentration
data for affected hours as "AG."
Unusual events
documented in site log
Sample collection affected by known
source, interference, or event which
impacts the representativeness of the
sample, such as: vehicle idling near
site, landscaping activities, construction
equipment in close proximity, leaks in
inlet where instrument samples air
inside shelter, food truck parked
nearby, etc.
Invalidate all target analyte concentration
data for affected hours as "SC".
Confidence is increased in data that pass screening checks, and data that do not meet the defined
criteria should be qualified or invalidated as detailed in Table 10-1. For failures of checks for
which data qualification or invalidation is not prescribed, these data should minimally be
investigated for peak misidentifications, improper integration, or chromatographic artifacts
interfering with target analyte peaks, among other problems with the data. Sample hours or
lengthy periods of ambient sampling hours failing numerous screening checks may indicate
malfunction of the instrument, contamination of the sample, or an error in CDS configuration.
Technical Review should include the following steps:
• Examine the collected data for missing data files. Missing files likely point to an
instrument failure. For instances of instrument failure, the sample immediately following
a missing hour's sample data is to be invalidated. If possible, data files that are missing
due to computer drive mapping problems or inadvertent movement to another location
should be recovered so they may be included in the technical review.
189
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
• Review the calibration data:
o Ensure all records are available: COAs, support instrument calibrations,
preparation records including dilution measurements or settings, and data files
corresponding to analysis of the calibration standards
o Trace the concentrations values entered to generate the calibration curves back to
the primary stock COA(s) and associated calculations
o Verify the nominal concentrations entered to generate the calibration curves are
accurate to those calculated and verify the curves meet criteria for the linear
regression: correlation coefficient, y-intercept, backcalculated values, and RF
RSD
o Verify proper assignment of the calibration curve or average RF in the CDS
o Verify the recoveries of the target compounds in the SSCV meet acceptance
criteria
Note: The calibration data need only be technically reviewed with the first batch of ambient
data measurements. Subsequent ongoing analysis of QC samples will demonstrate the
calibration remains valid. Technical reviewers should continue to verify that the correct
calibration curve or average RF is programmed into the CDS and that options to update the RF
with each CCV are disabled.
• Examine summary reports for the ambient sample and QC sample target analyte
concentrations and RTs in conjunction with reviewing chromatograms.
o Prepare overlays of the ambient sample data to the daily CCV/RTS to investigate
suspected RT shifts
o Examine chromatograms for reference compounds to ensure proper identification
o Verify abundant species are present in ambient sample chromatograms (e.g.,
ethane, benzene, propane, n-butane, isoprene, n-hexane, and ethylbenzene)
o Examine the results reports for missing components and high concentrations that
exceed the calibration curve or detector range
• Review results reports and chromatography for the following problems:
o Sample type mismatch between columns (e.g., sample type shows blank on the
PLOT FID and ambient sample on the PDMS FID)
o Insufficient collection time or incorrect sample start time (samples collected for
fewer than 40 minutes or samples starting earlier than 10 minutes before the hour
or later than 30 minutes after the hour)
o GC start times deviate from expectations (should be approximately 40 minutes
after sample collection start)
o Sample collection volumes and flows outside of specification.
190
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
o Instrument malfunction as may be indicated by wild baseline behavior,
uncharacteristically broadened peaks, atypical compound elution patterns, or
absence of expected compounds
o Mismatch between sample chromatography and sample type (e.g., an ambient
sample appears to be a blank or a CCV appears to be an ambient sample)
o Sample data collected as part of instrument conditioning following maintenance
or troubleshooting, audit samples, etc.
o Peak responses that exceed the detector range
• For samples with responses exceeding the detector range, review chromatograms from
samples collected several hours before and after the sample.
o Response could be the result of electronic spiking at the detector or elsewhere in
the measurement system. If electronic spiking is suspected, the target analyte
concentration should be invalidated. Results may be compared to other nearby
sites or other measurement methods (e.g., TO-15) conducted at the site to verify
the high concentration, when possible.
o Sample data following high concentration samples can exhibit contamination
from carryover and may exhibit RT shifts.
o Daily QC sample data should be reviewed to ensure the system has returned to
proper calibration and lack of carryover.
o Compounds with responses exceeding the calibration or detector range will
require qualification when reported to AQS (qualify as "EH").
• Review QC sample reports to ensure that daily CCVs and blanks as well as weekly
precision checks meet criteria. For QC samples failing criteria, associated ambient data
minimally require qualification or may require invalidation for affected compounds.
Refer to Table 4-4 of this document.
• Perform screening checks as listed in Table 10-1.
• Follow up on data that appear to be suspect based on incorrect calculations, missing
information, unusual events, acceptance criteria failures, and screening checks. Qualify or
invalidate such data as appropriate.
After examining these data reports, organize review notes, pertinent communication with the
instrument operator (as applicable), printouts (or saved pdfs, screen shots, or similar) of
chromatograms, and any other reviewed information such that they may easily be examined by
the data validator. A summary of the technical review including the scope of the data reviewed,
high-level observations, and changes or adjustments made to data can streamline operations for
subsequent data validation activities.
10.7.3 Speciated VOCs Data Validation Procedures
Once data have gone through the data verification processes of routine auto-GC operator review
and technical review, the dataset is presumed to be complete and technically correct. Data that
191
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
have been verified will have been flagged or invalidated based on their suitability and
compliance with the governing SOPs. These data can then undergo validation to investigate the
internal consistency of the dataset, consistency of the dataset with historical data, and
consistency of the dataset with datasets from nearby monitors.
10.7.3.1 Level 1 Data Validation
As described previously, tests for internal consistency identify values in the data set that appear
atypical when compared to values of the whole data set. Manual data review for internal
consistency is impractical for the volume of continuous GC data generated by PAMS. As
mentioned previously, DART was designed to review large and complex data sets for
consistency.
The validator should generate a database of the measured ambient concentrations and QC
concentrations over appropriate time periods (e.g., a week, a month, several days, etc.). This
generated dataset can be graphed or can be examined for descriptive statistics such as maximum,
minimum, median, mean, standard deviation, etc. and evaluate these statistics against expected
values. The data validator should also visualize the data with the graphical methods described in
Section 10.4. These tools are further described in the following sections as they relate to
speciated VOCs data validation.
In addition to these generating these statistical values and data visualizations, the data validator
should be minimally reviewing the following:
• Minimum of three ambient samples for each day
• The first and last ambient sample from the time period selected (week, month, etc.)
• Ambient samples with the highest TNMOC per channel per day
• 10% of the QC data for the chosen time period - this includes the daily CCV and system
blank as well as the weekly precision check
During review of the data listed above, the validator should review both chromatograms from
each channel for:
• misidentified samples,
• electronic spikes disrupting the chromatogram or affecting target peaks,
• samples with obvious baseline abnormalities,
• incorrect or inconsistent peak integration,
• peak misidentification and missed peak identifications,
• co-elutions with target peaks, and
• samples with Dean Switch timing errors, where applicable
Validators should prepare overlays of the ambient sample chromatograms with the most recent
CCV/RTS to verify that the elution pattern is consistent and that RTs are consistent and stable,
192
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
and that the auto-GC operator and/or technical reviewer have corrected the data where necessary.
Data validators should additionally review site logs, instrument logs, correspondence records,
and technical data reviewer notes for additional information related to anomalous sample data,
instrument or site issues, or other notable events impacting data. Data validators may widen the
scope of their review and select other samples to examine based on the outcome of the review of
chromatograms and additional associated information.
Data validators may correct data, add qualifiers, invalidate data, or may request that auto-GC
operators or technical reviewers perform these actions, based on the privileges defined in the
monitoring agency data validation SOP.
10.7.3.2 Level 2 Data Validation - Historical Data Comparisons
Note: In the first year following implementation in 2019 many PAMS Required Sites will not
have historical data from previous seasons.
If possible upon completion of Level 1 validation of the dataset, the data should undergo Level 2
data validation. If monitoring agencies cannot perform Level 2 validation, the dataset should be
prepared for uploading to AQS.
Testing or comparing data for historical consistency uses many of the graphical techniques
discussed in Section 10.4 to compare the dataset with previous data compiled from the
monitoring location. The outcome of the Level 2 validation is to identify values for further
inspection. Further inspection involves performing reviews of the individual chromatograms and
datafiles described in the Level 1 validation for the data identified as questionable. Values
representing pollutant behavior outside the determined limits should be flagged for further
investigation.
10.7.3.3 Level 3 Data Validation - Parallel Consistency Checks
Tests to check for consistency with parallel datasets from the same population (region, period of
time, air mass, etc.) are used to identify systematic bias. Systematic bias is determined by
checking for the difference in average value or overall distribution values. The sign test,
Wilcoxon signed-rank test, Wilcoxon rank sum test, and inter-site correlation tests are
recommended for testing two parallel data sets. The first three tests are nonparametric and
consequently can be used for non-normal data sets, which frequently occur in air quality data.
When comparing VOCs speciation and concentration among nearby sites, observe how well the
data compare. Investigate whether differences can be explained by meteorology, photochemistry,
analytical instrumentation differences, etc.
10.7.4 Speciated VOCs Visualization Methods
10.7.4.1 Time Series Graphs
Time series plots can be used to inspect each target compound, groups of target compounds,
and/or TNMOC. These visualizations allow the identification of outliers, increased single-hour
concentrations, possible missed peak identifications, and extended periods of unusually high or
low concentrations. Experienced PAMS personnel frequently look for unusual "jumps" in the
193
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
time series plot between successive hourly data or departures from expected diurnal or seasonal
patterns. Validators should inspect time-series plots for:
• Large "jumps" or "dips" in the concentrations
• Periodicity of peaks
• Evidence of calibration gas carryover into sample hours following calibration and CCV
• Expected diurnal behavior (i.e., biogenic isoprene concentrations usually peak during
mid-day or late afternoon)
• Expected relationships among species
• High single-hour concentrations of less abundant species
Data that appear different than expected should be marked for further investigation.
10.7.4.2 Scatter Plots
Scatter plots can be used to compare pairs of target compounds or target groups to identify
outliers and excursions in the data such as the improper inclusion of calibration data in the
dataset. Prepare scatter plots of the following, at a minimum:
• TNMOC versus species groups (i.e., aromatics, paraffins)
• TNMOC versus individual species
• Benzene versus acetylene and toluene (these species typically correlate)
• Benzene versus cyclohexane (look for split in the scatter plot indicating
mi si dentifi cati on)
• Benzene versus ethane (low or missing ethane concentrations when benzene is abundant
may indicate preconcentrator trap problems)
• Species that elute close together, e.g., 2,3-dimethylbutane, 2-methylpentane, and
3-methylpentane - 2-methylpentane concentrations should always be the highest of these
three peaks in ambient air. These three compounds are prone to misidentification.
• Isomers (e.g., o-, m-, and p-xylene).
10.7.4.3 Fingerprint Plots
Fingerprint plots allow further inspection of samples previously flagged for more detailed
review. The fingerprint plot shows the compound concentration for each compound for a single
hour. Prepare the fingerprint plots with VOCs in a consistent order on the x-axis, e.g., retention
order, alphabetical order, etc.). The fingerprints can quickly be scanned, hour-by-hour, to allow
observation of diurnal changes and inspection of sample hours surrounding suspect data to
identify additional effects.
10.7.4.4 Comparison with Other Parameters
Other data collected at PAMS sites, such as meteorology data and continuous gaseous
measurements, among other sources, may be used to further investigate suspected outliers
194
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
observed in VOCs data. Following are some examples of evaluations that may clarify the cause
of observed outliers:
• Plot wind direction data on time series plots - Do extreme values occur from a consistent
wind direction?
• Produce time series plots with overlays of other criteria pollutant data, such as ozone and
NO2 - are there obvious correlations which may explain the anomaly?
• Review subsets of data, such as days with high ozone events versus days with lower
ozone concentrations.
• Investigate local industrial or agricultural operating schedules, unusual event occurrence,
etc.
• Determine local traffic patterns and understand when peak traffic levels occur.
10.8 Carbonyl Data Verification and Validation
During the data verification and validation process for carbonyls, it is useful to consider the
various sources of carbonyl compounds in the atmosphere. Aldehydes are both primary
pollutants (i.e., directly emitted into the atmosphere) and produced as secondary products of
atmospheric photochemistry that also result in ozone production. Because of this relationship, the
determination of formaldehyde and other carbonyl compounds in the atmosphere is of interest.
Natural sources of carbonyls are not typically abundant; however, aldehydes are commercially
manufactured and released into the atmosphere from anthropogenic processes. Acetaldehyde is
naturally released during biomass combustion (e.g., wildfires)5 and is produced in biomass
decomposition.6 Motor vehicle emissions are a major contributor of atmospheric carbonyls with
formaldehyde from vehicle emissions accounting for 50 to 70 percent of the total carbonyl
burden to the atmosphere. Furthermore, motor vehicles emit reactive hydrocarbons that undergo
photochemical oxidation to produce formaldehyde and other carbonyls in the atmosphere.
Similarly, carbonyls are formed during the photo-oxidation of VOCs in the presence of nitrogen
oxide. Both anthropogenic and biogenic (e.g., isoprene, pinenes) hydrocarbons result in the
formation of carbonyls, especially formaldehyde. Sources of the more abundant carbonyls are
listed in Table 10-2.
Table 10-2. Major Sources of Carbonyls in the Atmosphere
C'iirhonyl C ompound
Major Source(s)
Comments
I'ormaldchx dc
l ucl combuslion
kc\ photochemical reaction producl (secondary
reaction product)
Acetone
Surface coating
most abundant VOC in landfill emissions and a
product of photochemistry
Acetaldehyde
Fuel combustion
key photochemical reaction product (secondary
reaction product)
195
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
Carbonyls data are to be verified and validated prior to reporting to AQS. The following sections
provide procedures and context for conducting these steps and the acceptance criteria for each
performance parameter are listed in Table 10-3.
Due to the laboratory component of extraction and analysis, carbonyls data verification and
validation will follow a different workflow than the other PAMS Required Site parameters. The
monitoring agency is responsible for the sample collection details and the final data validation as
it pertains to the site. The ASL is responsible for technical review of the extraction and analysis
data. The ASL may perform additional data verification and validation activities depending on
the arrangement between the ASL and monitoring agency. Irrespective of the convention, the
data verification and validation responsibilities of the ASL and the monitoring agency should be
clearly defined as to ensure that all aspects of sample collection, analysis, and data reporting
described in the following section are addressed.
10.8.1 Site Operator Verification Activities
The site operator should ensure that the sample identifier information and field collection details
are properly and completely documented on the sample collection form and/or sample COC and
that they conform to the details prescribed in the sample collection SOP during sample setup and
collection activities. Sample collection and handling details to document include:
• collection date
• start and stop times of collection
• starting and ending flow rates
• duration of collection
• dates of sample setup and retrieval
• sample shipment and storage records
• sampling instrument flow calibration records
Site operators should document observations and unusual events (e.g., situations that may
interfere with the sample result, procedural deviations) at the site or with the samples that may
impact sample integrity or sample measurement. Prior to releasing the samples to the shipping
courier or to the laboratory, the field operator should again verify the sample collection details
and COC details are documented completely and accurately, particularly that sample identifier
and collection information is correct for field blanks, trip blanks, and duplicate or collocated
samples. Site operators are encouraged to maintain a copy of sample collection forms and COC
forms for subsequent validation activities.
10.8.2 ASL Verification and Validation Activities
10.8.2.1 ASL Sample Receipt
At the laboratory, the sample receipt custodian will complete the custody transfer details and
review the sample collection records accompanying the samples to ensure the documentation is
complete and reasonable. Inconsistencies (such as sample collection times or volumes that do not
196
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
seem reasonable) or missing information should be discussed with the field operator to complete
or correct the record, as appropriate. The sample custodian should assign a unique identifier,
store the samples within the laboratory, and document the sample storage details (refrigerator
identifier and maintained temperature range).
10.8.2.2 ASL Sample Extraction
The extraction technician will review the sample collection and custody details when preparing
the extraction batch to ensure that incomplete or inconsistent information has been corrected and
that sample identifiers match. The extraction chemist should document the extraction details
sufficiently to ensure that materials used (e.g., extraction solvent, volumetric flasks, spiking
solutions) are traceable and that activities (e.g., extract final volume, extract storage) may be
reconstructed. Problems or nonconformances with established procedures are to be documented
to notify the instrument analyst.
10.8.2.3 ASL Sample Analysis
The analyst should review the sample collection and custody records as well as the extraction
batch records for notes which may impact the analysis and to ensure records are complete and
are associated with the extract sample identifiers. The analyst should document materials utilized
(e.g., calibration stocks, mobile phases, volumetric delivery pipettes) and procedures followed
(e.g., preparation of calibration standards, analytical methods programs, reagent preparation) to
analyze the sample extracts as well as problems or nonconformances with established
procedures.
Once analysis is completed, the analyst should perform initial (self) review on the analytical data
and document adjustments and corrections as needed. This review includes examination of the
analytical data and QC sample data to ensure acceptance criteria have been met. The analyst may
reanalyze extracts as needed to verify questionable results. Preparation and use of a checklist are
recommended to ensure that appropriate analysis procedures and criteria have been addressed.
Finally, the analyst should compile a data package which typically includes the associated
sample collection and custody forms (or copies thereof), sample extraction records, analysis
instrument data, and checklist detailing a summary of the data reviewed and any comments or
notes.
10.8.2.4 ASL Overall Technical Review
Once the analyst completes their initial review and data package assembly, the data package is
then ready for technical review. To streamline the technical review and ensure aspects are not
overlooked, a checklist (which can include many of the same details as the analyst checklist) is
recommended. Technical review should include the following:
1. Review of the sample collection and custody records
o custody transfer records are complete and reasonable (transfers are chronological)
o sample transport, receipt, and laboratory storage temperatures were < 4°C
o sample collection details are within specifications
¦ start and stop times
¦ beginning and ending, or average flow rates
197
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
¦ total collected volume
¦ sample setup and retrieval
o sample collection or custody procedural deviations are documented
o site operator notes for unusual events or sampling problems
2. Review of sample extraction details
o Sample extract storage temperatures were < 4°C
o Preparation of extraction batch QC conforms to specifications
¦ Method blank
¦ Laboratory control sample and laboratory control sample duplicate
¦ Extraction solvent method blank
o Sample extract volumes
o Extraction solvent lot and expiration
o Records are traceable by initials and date
o Spiking solutions are traceable - review:
¦ preparation records
¦ dilution calculations
¦ calibration records for measurement devices (volumetric delivery pipettes,
analytical balance calibration, etc.)
o Sample identifiers are accurate
o Sample holding times (<14 days from end of collection) are met
3. Review of analysis data
o Sample extract holding times (<30 days from extraction) are met
o Calibration standards preparation records are traceable and suitable
¦ standards are within expiration
¦ certificates of analysis are available
¦ dilution calculations are verified
¦ calibration records for measurement devices (volumetric delivery pipettes,
analytical balance calibration, etc.)
¦ standards storage is < 4°C
¦ records are traceable by initials and date
o Calibration curves preparation
¦ Nominal concentrations entered properly in CDS calibration table
198
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
¦ Curves meet acceptance criteria for linearity, intercept, and evaluation
against the nominal concentration (criteria listed in Table 5.5)
o Analytical sequence is appropriate
¦ Solvent blank begins the day's analysis
¦ Calibration is established day of analysis or verified by CCV
¦ Extraction batch QC samples (ESMB, MB, LCS, LCSD)
¦ Sample replicate once per batch
¦ CCV every 12 hours of analysis
¦ Sample identifiers are accurate and traceable
o Compound identification is appropriate
¦ Peak signal to noise is > 3:1
¦ RTs are within the defined RT window
o Peak integration is appropriate and consistent
¦ Automatic integration is reasonable
¦ Coelutions are addressed
¦ Manual integration, where needed, is appropriate, consistent, conforms to
monitoring agency procedures and policies, and justified (preferably with
documented analyst rationale)
o Dilutions are calculated properly and the MDL is adjusted for (multiplied by) the
dilution factor
o Field, extraction, and analysis QC samples meet criteria (listed in Table 5.5)
o Procedural or acceptance criteria failures and corrective actions are documented
and justified
o Data reasonability checks performed:
¦ DNPH is present in each cartridge extract (ambient sample, FB, TB, MB,
LCS, LCSD) - typically twice the intensity of the next largest peak
¦ Formaldehyde is detected in ambient samples (typically > 0.3 ppbC)
¦ Carbonyl concentrations in ambient samples are reasonable with those
expected at the site
o Subsequent calculations for in-air concentrations are correct
o Sample results are qualified appropriately for QC failures or procedural deviations
Laboratories may have additional review requirements including additional levels of technical
review, data quality review, or quality assurance assessments before the carbonyls data can be
released to the monitoring agency.
199
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
10.8.3 Carbonyls SLTMonitoring Agency Data Verification and Validation
Once the ASL has reported data to the SLT monitoring agency, the monitoring agency data
validator can complete the data verification activities and begin validation. Sample data from the
ASL may be provided in a number of different formats, depending on the arrangement between
the ASL and monitoring site, which may include the following conventions:
• simply a mass of target analyte per cartridge (|ig/cartridge) for the submitted samples
• the mass of target analyte normalized to the collected air volume (|ig/m3) for the
submitted samples
• the mass of target analyte normalized to the collected air volume (|ig/m3) in files ready
for AQS upload for the submitted samples
• the mass of target analyte normalized to the collected air volume in files ready for AQS
upload and upload the data to AQS for the submitted samples
Irrespective of the ASL convention for reporting carbonyls field-collected sample (ambient
samples and field QC samples such as field blanks and trip blanks) data, the monitoring agency
will need to perform subsequent validation on the carbonyls data. The ASL will ensure that the
reported values meet the acceptance criteria prescribed in the method and will flag or invalidate
data that are compromised per the laboratory QAPP and appropriate SOPs. In addition to the
field collected sample data, the ASL may also provide the associated extraction and analytical
batch QC data. The monitoring agency data validator should examine these QC data during data
validation. Data that are identified as questionable or unsuitable should be discussed with the
ASL.
10.8.3.1 Manual Inspection of Carbonyls Collection Data
The field operator should have performed initial review of the sample collection data as
described in Section 10.8.1 prior to the sample shipment to the ASL. Even if the ASL is
reviewing sample collection data, the monitoring agency should also perform technical review of
the sample collection and COC forms in concert with the ASL analysis data as part of the data
validation. Note that such a technical review is to be performed for each field sampling event to
include:
• samples collected on the proper date per the national sampling calendar (unless a make-
up sample)
• leak check performed and passed criteria (as applicable)
• pre-sampling purge performed (as applicable)
• sampling start and stop times within ±15 minutes of beginning of the scheduled hour
(adjusted for clock offset error)
• starting, ending, and average flow rates reasonable (e.g., ±10% of setting)
• duration of collection within 460 to 500 minutes
• sample shipment and storage records - samples stored refrigerated upon retrieval and
shipped on ice packs
200
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
• comments on collection forms and/or COC forms that may impact data
• COC forms are complete
• associated field QC samples meet criteria
If the data validator discovers problems or errors that impact the results, the data validator may
need to contact the field operator for clarification or to rectify the problem. Depending on the
severity of the issue, the ASL may need to be notified to amend reported data. The data may be
corrected effectively, flagged, or invalidated. Refer to Table 10-3 for specific guidance.
10.8.3.2 Review of ASL Data
As with manual inspection of the field collected data, the data validator will review the data
provided by the ASL for each sampling event for:
• sample holding times compliance
• acceptable laboratory QC (if provided) and appropriate data flagging
• MDLs reported with each sample result
• comments or notes affecting data quality
If the data validator discovers problems or errors with the ASL data, the data validator should
notify the ASL as soon as practical so the issue can be corrected. The ASL should then provide
revised data to the monitoring agency, as appropriate.
10.8.3.3 Review of Supporting QC Data
In addition to performing technical review of the routine carbonyls field sampling data, the data
validator should review site logs, instrument logs, audit reports, corrective actions, and
supporting documentation for the following:
• instrument bias checks met criteria at the beginning of PAMS season
• sampler siting verified prior to PAMS season
• ozone denuder recharged/replaced at the required frequency
• sampling instrument flow calibration and calibration verification records demonstrate
acceptable performance (within ±10% of transfer standard)
• reference transfer standards within calibration
• maintenance and site logs detail unusual events
• cartridge lot background determination met criteria
• cartridge media are within the assigned expiration period
• field QC samples collected at the proper frequency
• audit findings from IP As, TSAs, or ADQs affecting data integrity or quality
• corrective actions that may impact data integrity or quality
201
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
A validation table distilling the general QC guidance and requirements for carbonyls is provided
in Table 10-3. More information on each data validation parameter can be located within the text
identified in the reference column. Each parameter is assigned a category of importance. The
categories in order of decreasing importance are:
1. Critical - Criteria will be met for reported results to be valid
2. MQO - PAMS Measurement Quality Objective to be attained to evaluate the DQO
3. Operational - Failure to meet criteria does not invalidate reported results; the results
are compromised and on a case-by-case basis may require qualification or
invalidation
4. Practical - Failure to meet criteria does not invalidate reported results; results may be
compromised, do not require qualification, but may be qualified based on the opinion
of the data validator.
Issues or problems with the verifications above and findings or corrective actions impacting data
are impetus for data flagging or qualification. Many of the common issues are listed in Table
10-3; however, for situations where a large amount of data is impacted and the guidance does not
address the specific situation, monitoring agencies should confer with their Regional
representative for how the data should be handled.
10.8.3.4 SLTMonitoring Agency Carbonyls Data Validation
Once the carbonyls data have been verified in the above steps, the monitoring agency can
conduct validation by employing the statistical and visualization tools and methods described in
Section 10.4.
10.8.3.4.1 Level 1 Carbonyls Data Validation
Carbonyls data that have undergone data verification will have been evaluated to have met
acceptance criteria but should still be examined for suitability by a data validator. Specifically,
the data validator should review approximately 5 to 10% of the sample data for transcription and
calculation accuracy and verify the criteria in Table 10-3 are met for the chosen samples.
Validators should review the following records to assess the impact on the data undergoing
validation:
• audit records
• corrective action reports
• monthly flow check results
• DNPH cartridge lot blank determinations
• sampling unit bias check results
• site logbooks
• instrument maintenance records
202
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
• field QC sample results
• laboratory QC results
These data sources should be examined as a first step in the data validation process and any
issues or problems noted should confirm that associated data have been appropriately qualified
or invalidated.
Provided the supporting data sources discussed above do not indicate anything out of the
ordinary, subsequent methods for validating carbonyls data include employing statistical tools to
characterize the central tendency and variability and data visualization renderings.
Time series plots for formaldehyde and acetaldehyde should be prepared and plots examined for
non-detect samples. Formaldehyde should be measured for all ambient (non-field blank) samples
above the 0.25 |ig/m3 MDL MQO threshold, and acetaldehyde will typically be measured above
this level for the two (of the three) sequential samples collected during daylight hours.
Concentrations of acetone are highly variable as there are numerous sources including secondary
formation (50%), biomass burning (26%), direct biogenic emissions (21%), and anthropogenic
sources (3%).5 Additionally, acetone is typically present at significant background amounts on
DNPH cartridge media, which complicates the ability to accurately measure the concentration of
acetone attributable to ambient measurements. This complication makes it difficult to discern
patterns and to assign parameters with which acetone's behavior correlates or reasonable
definitive criteria for determining outliers. Due to the significant and variable acetone
background on DNPH cartridges, ambient acetone data may frequently require qualification to
estimate the measured concentration, especially when measured in lot blank evaluations, trip
blanks, and field blanks.
Benzaldehyde may not be measured at concentrations above the MDL due to typically lower
atmospheric concentrations relative to formaldehyde and acetaldehyde, therefore atypically high
concentrations (e.g., > 0.5 |ig/m3) should be reviewed for reasonableness.
Preparing plots of the individual compounds for the three different 8-hour periods should
indicate a diurnal pattern where the carbonyls concentrations measured in the overnight sample
are lowest, particularly for formaldehyde and acetaldehyde. As atmospheric carbonyls burden is
primarily driven by mobile sources, time series overlays should be prepared with the BTEX
components from the co-collected speciated VOCs data to verify concentrations show a general
correlation of increase and decrease together. Scatter plots may be likewise useful in identifying
data that deviate from this expected relationship. Carbonyls sample values which deviate
significantly from the BTEX trend should be investigated further. A major source of
benzaldehyde is as a by-product of atmospheric toluene degradation, particularly in the presence
of NO. Benzaldehyde may be compared to daily toluene and/or NO concentrations by
preparation of scatter plots.
Validators should keep in mind that when comparing data from different measurement systems
such as speciated VOCs data, true NO2 data, and meteorological data, the data pulled in for
203
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
comparison may be in different phases of the data verification and validation process. Validators
should be aware that VOCs data may not have gone through validation and may include
qualifiers indicating the data quality are compromised. When questionable data are identified by
comparison with data from different measurements systems, the data to which they are compared
should be quickly reviewed for qualifiers to ensure the comparison is meaningful.
Scatterplots and fingerprint plots of carbonyls compounds plotted against speciated VOCs or
other carbonyls may be useful in identifying abnormal carbonyls data. Caution should be
exercised when combining 8-hour carbonyls data with hourly measurements from continuous
methods such that the validator ensures continuous data are plotted on the same time scale (i.e.,
averaged over the corresponding 8-hour period).
Carbonyls data collected as 8-hour samples may be compared generally with 24-hour samples
collected over a similar timeframe (e.g., for air toxics networks). Validators should utilize such
checks for a gross error check, as the sampling period covers a different time window - 24-hour
sampling typically begins and ends at midnight, whereas the 24-hour period for PAMS sampling
begins and ends at 04:00 a.m. Additionally, there are known differences in the performance of
the collection of carbonyls due to humidity differences. Under identical conditions over the
identical time period, a collocated 24-hour sample concentration may differ significantly from
the average concentration of three collocated sequential 8-hour samples. Such a gross error check
may identify carbonyls compounds or samples for which there are large discrepancies that
warrant further investigation. Such may be the case for carbonyls compounds that are measured
on one or more of the 8-hour samples but not on the corresponding 24-hour sample. Large, e.g.,
several-fold, discrepancies may indicate a sample identification discrepancy or measurement
error. Such a check may be used to identify data for further investigation but should not in itself
be a rationale for invalidation of data.
10.8.3.4.2 Level 2 Carbonyls Data Validation
Data validators should compare the current carbonyls dataset against historically collected
carbonyls data from the site, if available, to identify values which stand out historically.
Comparisons should involve combining the current and historic datasets as possible and
generation of simple statistics to identify extreme values and longer-term variability (e.g.,
standard deviation) that is characteristic at the site and against which outliers may be discerned.
Validators should employ the data visualization tools such as time series plots and scatter plots
as recommended for the Level 1 validation (overlays with BTEX compounds, etc). Values
identified to not conform to the expected pattern should be investigated to verify the values meet
criteria or are qualified or invalidated, as appropriate.
10.8.3.4.3 Level 3 Carbonyls Data Validation
For Level 3 data validation, carbonyls data should be compared to data generated from other
nearby sites within the airshed and collocated samples. When comparing carbonyls
concentrations among nearby sites or collocated monitors, observe how well the data compare.
Investigate whether differences can be explained by meteorology, photochemistry, analytical
instrumentation differences, etc.
204
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
Table 10-3. Carbonyls Data Validation Table
Piii'iimolcr
Dcsci'iplion iiml Kc(|iiiivd I miucno
AciT|HiiniT ( riii-riii
KiTciviiit
CsiU'Kon
Field Readiness Checks and Collection Activities
Collection Media
All field-collected samples and matrix quality control
samples
Cartridge containing silica gel solid
sorbent coated with DNPH
Section 5.5
Critical
Sample retrieval as soon as possible, not
to exceed 72 hours post-sampling.
Media Handling
All field-collected samples and all quality control
samples
Retrieved sample transported and stored
refrigerated at < 4°C, protected from light
until extraction.
Damaged cartridges (water damage or
cracked) must be voided.
Sections
5.8.1.2, 5.5.2,
and 5.5.3
Operational
Cartridge Lot
Blank Check
Analysis of a minimum of 3 cartridges or 1% of the total
lot, whichever is greater, for each new lot
Formaldehyde < 0.15 |ig/cartridge,
Acetaldehyde < 0.10 |ig/cartridge.
Acetone <0.30 jig/cartridgc.
all others <0.10 |ig/cartridge
Section 5.5.1
and Table
5-2
Critical
Sampling Unit
Clock/Timer
Verified with each sample collection event
Clock/timer accurate to ± 5 minutes of
reference, set to local standard time
Section 5.8.1.1
Operational
Check
Sample collection program verified
Sampling Unit
Leak Check
Pressurization or evacuation of internal sampler flow
paths to demonstrate as leak-free
Prior to each sample collection
Must show no indicated flow
Section 5.7.2
Operational
Sampling
Frequency
Three consecutive 8-hour samples every three days
according to the EPA National Monitoring Schedule
Samples in sequence will be valid or a
make-up sample sequence scheduled
(refer to Section 3.3.2.1)
Section 5.8.1.3
Critical and
MQO
Sampling Period
All ambient field-collected samples
460 - 500 minutes (8 hr ± 20 mins)
starting and ending within 15 minutes of
scheduled hour
Section 5.8.1.3
Critical and
MQO
Pre-Sample
Collection Purge
Prior to the first sequential sample of each sampling
event
Minimum of ten air changes just prior to
sample collection
Section 5.7.2
Practical
205
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
Table 10-3 (continued). Carbonyls Data Validation Table
Piii'iiiiHMcr
Description ;nul Kc(|iiirc(l l-"rc(|iicnc>
Accept mice (rilcrhi
Reference
Ciilcjion
Sample Receipt
Chain-of-
custody
All field-collected samples
Each cartridge will be uniquely
identified and accompanied by a
valid and legible COC with
complete sample documentation
Section
5.8.1.4
Critical
Sample Holding
Time
All field-collected samples, laboratory QC samples, and standards
Extraction: 14 days from sample
collection (cartridge storage < 4
°C)
Analysis: 30 days from extraction
(extract storage < 4 °C)
Section
5.9.2.4
Operational
Sample Receipt
Temperature
Check and
Subsequent
Storage
All field-collected samples upon receipt at the laboratory, stored
immediately in refrigerator
< 4°C (unless transport duration is
not sufficient to sufficiently cool
samples)
Section
5.5.2
Operational
HPLC Analysis
Solvent Blank
(SB)
Prior to ICAL and daily beginning CCV
All target compounds < MDLsp
Section
5.9.5.2
Operational
HPLC Initial
Multi-Point
Calibration
(ICAL)
Initially, following failed CCV, or when changes to the instrument affect
calibration response
Analysis of a minimum of 5 points covering approximately 0.01 to 3.0
|ig/mL
Correlation coefficient (r) > 0.999;
relative error for each level
against calibration curve < 20%.
Absolute value of intercept
divided by slope must not exceed
MDLsd
Section
5.9.5.2
Critical
Secondary
Source
Calibration
Verification
(SSCV)
Secondary source standard prepared at the mid-range of the calibration
curve, analyzed immediately after each ICAL
85 to 115% recovery
Section
5.9.5.3
Critical
Continuing
Calibration
Verification
(CCV)
Prior to sample analysis on days when an ICAL is not performed,
minimally every 12 hours of analysis, at the end of the analytical
sequence - recommended following analysis of every 10 field-collected
samples
85 to 115% recovery
Section
5.9.5.4
Critical
Extraction
Solvent Method
Blank (ESMB)
An aliquot of extraction solvent delivered to a volumetric flask. One with
each extraction batch of 20 or fewer field-collected samples.
All target compounds < MDLsp
Section
5.9.4.1
Operational
206
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
Table 10-3 (continued). Carbonyls Data Validation Table
Piii'iiiiHMcr
Description iiml Kc(|iiirc(l l-"rc(|iicnc>
Accept mice Crilcrhi
Reference
Ciilcjion
Method Blank
(MB)
Unexposed DNPH cartridge extracted as a sample
One with every extraction batch of 20 or fewer field-collected samples
All target compounds < MDL
Section
5.9.4.1
Operational
Laboratory
Control Sample
(LCS)
DNPH cartridge spiked with known amount of target analytes at
approximately the lower third of the calibration curve, twice quarterly -
recommended with every extraction batch of 20 or fewer field-collected
samples
Formaldehyde recovery 80-120%
of nominal spike
All other compounds recovery 70-
130% of nominal spike
Section
5.9.4.1
Operational
Laboratory
Control Sample
Duplicate
(LCSD)
Duplicate LCS to evaluate precision through extraction and analysis,
twice quarterly - recommended with every extraction batch of 20 or
fewer samples
Formaldehyde recovery 80-120%
of nominal spike
All other compounds recovery 70-
130% of nominal spike
Precision < 20% RPD of LCS
Section
5.9.4.1
Operational
Retention Time
(RT)
Every injection
Each target carbonyl's RT within
± 3s or ± 2% of its mean ICAL
RT
Section
5.9.5.6
Critical
Replicate
Analysis
A single additional analysis of a field-collected sample extract
Once with every analysis sequence of 20 or fewer samples
Precision < 10% RPD for
concentrations >0.5 |ig/cartridge
Section
5.9.5.5
Operational
Field Blank
Minimally twice per month, sample cartridge installed in sampling
channel for approximately five minutes to expose the cartridge to the
field conditions of the co-collected field sample
Formaldehyde < 0.30 |ig/cartridge.
Acetaldehyde < 0.40 jig/cartridgc.
Acetone < 0.75 jig/cartridgc.
Sum of all other target compounds
< 7.0 ug/cartridge
Section
5.8.2.1
Operational
Collocated
Sample
Collection
Field sample collected through a separate inlet probe from the primary
sample
10% of primary samples for sites performing collocated sample
collection
Precision < 20% RPD of primary
sample for concentrations >0.5
|ig/cart ridge
Section
5.8.2.3
Operational
Duplicate
Sample
Collection
Field sample collected through the same inlet probe as the primary
sample
10% of primary samples for sites performing collocated sample
collection
Precision < 20% RPD of primary
sample for concentrations >0.5
|ig/cart ridge
Section
5.8.2.4
Operational
207
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
Table 10-3 (continued). Carbonyls Data Validation Table
Piii'iiiiHMcr
Description iiml Kc(|iiirc(l l-"rc(|iicnc>
Accept mice Crilcrhi
Reference
Ciilcjion
DNPH
Chromatography
Evaluation
All cartridges
DNPH peak must be present
Section
5.9.5.7
Critical
For all field-collected cartridges
DNPH must be > 50% of the
DNPH area in the laboratory QC
samples
Operational
Laboratory Readiness and Proficiency
Proficiency
Testing
Blind sample submitted to each laboratory to evaluate laboratory bias
Twice per calendar year, once prior to PAMS season and toward the end
of PAMS season1
Each target compound within ±
25% of the assigned target value
Section
3.3.4.1
Critical and
MQO
Method
Detection Limit
Determined annually prior to PAMS season and when method changes
alter instrument sensitivity
MDL will be:
Formaldehyde < 0.25 |ig/m3
Acetaldehyde < 0.25 ug/m3
Section
5.6
MQO
Stock Standard
Solutions
Purchased stock materials for each target carbonyl
All standards
Certified and accompanied by
certificate of analysis
Section
5.9.2.2
Critical
Working
Standard
Solutions
And Sample
Extracts
Storage of all working standards and extracts
Stored at < 4°C, protected from
light
Section
5.9.2.4
Operational
Sampling Unit Testing and Maintenance
Field Sampler
Flow Rate
Calibration
Calibration of sampling unit flow controller
Minimally annually, prior to PAMS season
Flow set to match a certified flow
transfer standard
Table
5.7.1.2
Critical
Ozone Scrubber
Recharge
Recharge ozone scrubber with KI
Minimally every other PAMS season
Scrubber capacity sufficient to be
effective (ozone removal > 95%)
for 6 months of 24-hour sampling
every third day.
Section
5.4
Critical
Sampling Unit
Non-biasing
Certification
Verification with humidified zero air or nitrogen that the sampling unit
does not contribute to positive bias
Prior to field deployment and annually thereafter, or when flow path
components are repaired or replaced
<0.2 |ig/cartridge more than the
co-collected reference sample for
each target carbonyl
Section
5.7.1.1
Operational
Sampling Unit
Flow
Calibration
Check or Audit
Verification of sampling unit flow rate
Immediately following calibration and minimally monthly thereafter
Flow within ± 10% of certified
primary or transfer standard flow
and design flow
Section
5.7.1.2
Operational
208
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
Table 10-3 (continued). Carbonyls Data Validation Table
Piii'iiiiHMcr
Description iiml Keijiiireri l-'re(|iiene>
AivrpliiniT ( rilerhi
Reference
Ciilejion
Site Specifications and Maintenance
Sampling Unit
Siting
Verify conformance to requirements
Annually
270° unobstructed probe inlet
Inlet 2-15 meters AGL and > 1
meter from any supporting
structure
>10 meters from drip line of
nearest tree
Collocated sampling inlets spaced
< 4 meters from primary sampling
unit inlet
Section
3.3.1.2
Critical
Sample Probe
and Inlet
Sample probe and inlet materials composition
Annually
Chromatographic grade stainless
steel, PTFE or PFA Teflon®, or
borosilicate glass
Section
5.7.3
Critical
Sample Inlet
Filter
Particulate filter maintenance
Minimally annually prior to PAMS season, if equipped - commensurate
with site particulate matter conditions
Clean or replace the inline
particulate filter (if equipped)
Section
5.7.3
Operational
Sampling Inlet
and Inlet Line
Cleaning
Sample inlet and inlet line cleaning or replacement
Minimally annually prior to PAMS season - More often in areas with
high airborne particulate levels
Cleaned with distilled water or
replaced
Section
5.7.3
Operational
Data Reporting
Data Reporting
to AQS
Reporting of all results a given calendar quarter
Quarterly, within 180 days of end of calendar quarter
All field-collected sample
concentrations reported including
concentrations below MDL.
Field QC sample and laboratory
replicates will be reported.
Section
11.2
Operational
AQS Reporting
Units
Units as specified with each quarterly submission to AQS
|ig/m3 or ng/m3 at EPA standard
conditions of 25°C and 760
mmHg
Table 11.1
Critical
Data
Completeness
Valid samples compared to scheduled samples
Each PAMS season
> 85% of scheduled samples
Section
3.3.2
MQO
209
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
10.9 Data Verification and Validation of Ozone and Nitrogen Oxides
EPA has established data validation guidelines and procedures for data verification and
validation for continuous gaseous criteria pollutant (ozone and oxides of nitrogen) monitoring as
prescribed in the validation templates in the QA Handbook Volume II, Appendix D, Revision 1,
March 2017, available at the following link on AMTIC:
https://www3.epa.gov/ttnamtil/files/ambient/pm25/qa/APP D%20validation%20ternplate%20ve
rston%2< >0 » :0l i'« i%20AMTIC%20R^ tpdf
In general, monitoring agencies will have established data verification and validation procedures
for these parameters as part of their criteria pollutant monitoring programs at their NCore
monitoring sites.
This section will briefly describe some of the aspects of data verification and validation
described in Sections 10.1 and 10.2 that apply to the criteria gaseous monitoring, which include:
• Timely review of data to correct problems to limit the amount of data affected
• Instrument operators and technical reviewers are familiar with the instrument behavior
and outputs, typical measured concentrations, typical interferences, calibration routines,
and quality control checks and acceptance criteria
• Instrument operators should frequently check on the instruments to ensure they are
operating properly and that data are being recorded - whether by a site visit or remote
login to the instrument or DAS
• Data validation should involve calculating simple statistics and visualizing data
• Review of the site log and/or instrument log for unusual events and maintenance
• Review of independent audits and corrective actions that would indicate ongoing issues
impacting data
• Documentation of notes, problems, and observations by the data reviewer and validator
For the continuous gaseous criteria measurements, when a zero or span check fails criteria, the
ambient measurements recorded should be invalidated back to the most recent passing valid QC
check. Data following an analyzer malfunction or period of non-operation should be invalidated
until the next calibration or zero and span QC checks unless the QC checks meet criteria and the
instrument has not been adjusted over the time in question.
For continuous analyzers with on-board or external DAS, including true NO2 and O3, the DAS
may include software functions to aid in detecting changes in operating conditions (e.g.,
sensitivity changes, equipment degradation or malfunction, etc.). As these functions are
dependent on the specific instrument model and/or DAS installed, please refer to the
manufacturer manual for further information. Such automated validation programs may include
the ability to flag data according to:
• A user-defined maximum concentration
• A user-defined minimum concentration
210
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
• A maximum rate of concentration change
• Quality control check failures
• Communication errors
10.9.1 Ozone
Monitoring agencies should have established guidelines for data review, verification, and
validation of ozone. What follows is a summary discussion of ozone data review and validation
information described within the EPA QA Handbook Volume II, Appendix D, Revision 1,
March 2017.
Ozone is a secondary pollutant produced from reactions of NOx and VOCs in the presence of
sunlight. Most of the nitrogen oxides emitted into the atmosphere are emitted as NO. If O3 is
present where NO is emitted, the O3 levels will be reduced due to NO scavenging (titration).
However, in the presence of VOCs, ozone will form and accumulate - over a period of a few
hours or over several days, depending on meteorological and other environmental conditions.
Typically, a diurnal profile of ozone will show moderate levels overnight, a slight dip as NO
levels increase, followed by a steady rise to the maximum ozone value, generally often observed
in mid-afternoon, followed by a decay in the early evening when the lack of sunlight, NO, and
VOCs limits the production of ozone. Diurnal profiles will vary greatly by site, depending on
location, emission sources, and meteorological conditions; they can, however, be useful in
identifying unusual/suspect values. Comparison of values to those measured at nearby sites is
another useful screening approach. Note that differences observed between nearby sites may be
real and should be explainable. For example, if the nearby "buddy" site is further upwind (i.e.,
closer to a NOx source), the ozone levels may be lower due to titration.
Although the relationship between O3 and N0/N02/N0x/N0y is not definitively predictable, it
can be quite useful to plot O3 data with N02/N0x/N0y. For example, a short-term drop in O3 that
appears at the same time as a spike in NOx is likely to be real. Instrument or data transmission
issues may be detected by examining the number of hours where the measured ozone
concentration is zero. Similarly, instrument issues and/or titration can be indicated by the
majority of non-zero values that are less than 30 ppb. Additional screening criteria for ozone are
included in Table 10-4.6
Tab
e 10-4. Example Screening Criteria for Ozone
Screening Check
Crileriii lor l-'iirthcr Imesligiilion
Maximum
> -170 to 225 ppb
Minimum
< -5 ppb
Check hourly O3 data for shifts in baseline concentrations
Rate of change
> 50 to 60 ppb/hr
Nearby sites
Within ±50 ppb
Sticking Check
> 40 ppb for > 4 consecutive hours
Co-pollutant
Relationship with NO and NOx should be conform to expectations
211
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
10.9.2 Nitrogen Oxides, including True NO2
Monitoring agencies should have established guidelines for data verification and validation of
oxides of nitrogen. What follows is a summary discussion of oxides of nitrogen data review and
validation information described within the EPA QA Handbook. 7
In the atmosphere, NO2 is generally produced through the reaction of emitted NO with ozone; as
such, it can be difficult to establish a relationship between the two compounds. However, it can
be useful to plot collocated ozone with NO/NOx/NOy, in a time series plot. The following
questions should be considered when screening NO/NOx/NOy data:
• Are NO/NO2 concentrations high in the morning and evening?
• Are there any negative concentrations?
• Do measured values correlate with wind direction for upstream sources?
Additional screening criteria for true NO2 will vary by site; Table 10-5 presents example criteria
that can be used as a starting point for developing site-specific criteria.6
Table 10-5. Example Screening Criteria for N02/N0/N0x/N0y
Screening Check
Critcriji lor l-'urlhcr ln\csli»;ilion
Maximum
>700 ppb urban
>300 ppb rural
Minimum
<-1 ppb
Rate of change
>30 ppb/hr
Sticking Check
Any non-zero value for > 4 consecutive hours
Co-pollutant
NO should not exceed NOx or NOy
NO2 should not exceed NOx or NOy
10.10 Verification and Validation of Routine Meteorological Measurements
Ambient air pollution data are linked to meteorological data, and it is strongly recommended that
meteorology data be verified and validated at the same time as pollution data. The data
verification and validation process for routine meteorological data is described in detail in the
EPA QA Handbook, Volume 4, January 2008, available at the following link on AMTIC:
https://www3.epa.gov/ttnamtil/files/ambient/met/Volume IV Meteorological Measurements.pdf
10.10.1 Routine Meteorology Data Verification
Verification of routine meteorology data includes three basic aspects:
1. routine inspection by site operators of instruments, DAS communication, and data
reasonableness
2. automated checks, analysis, and verification performed by the DAS (or similar automated
system)
212
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
3. technical review of data where calibration data, QC check data, and routine data are
reviewed for correctness, completeness, and compliance with the established procedures
10.10.1.1 Site Operator Routine Checks
Data from certain meteorological parameters can be verified visually. Site operators should
check measured meteorological data against manual/visual observations daily when onsite. For
example, under windy conditions, a cup anemometer and vane system at a monitoring station
should move according to the conditions. Rainfall can be manually inspected using a standard
residential precipitation gauge while the site operator is present. Temperature and barometric
pressure readings can be checked against measurements from other instruments at the site such
as PM2.5 monitors to ensure readings are reasonable.
When on site, site operators should perform the following visual checks:
1. Verifying equipment is performing properly and generating reasonable measurements
a. instruments are communicating properly with the DAS
b. mechanical instruments such as wind instruments and precipitation instruments
are moving according to conditions and registering reasonable data with the DAS
c. Solar radiation and UV radiation instrument measurements are reasonable for the
conditions (sunny, overcast, etc.)
d. Hygrometer readings are reasonable for the conditions (rainy, drought, etc.)
2. The site operator should review the meteorological data collected since the last site visit:
a. Ensure data have been collected for all parameters for all hours
b. Perform a quick visual inspection of the data to look for anomalies from the
following expectations:
i. Do higher relative humidity values correspond to relatively high
temperature values?
ii. Do temperature reading changes generally transition smoothly?
iii. Are precipitation measurements indicated for recent rain or snow events?
iv. Do solar radiation, UV radiation, and mixing layer height indicate a
general diurnal pattern?
Operators should notate observed meteorological conditions when on site and document the
checks performed above. An example checklist for routine onsite checks of meteorological
instruments is shown in Figure 10-7. This example checklist and a description of what each
check entails can be found in the EPA QA Handbook for Air Pollution Measurement Systems
Volume IV: Meteorological Measurements Version 2.0, Section 10. Note the form includes a
signature block for a reviewer and the completed forms should be reviewed as part of technical
data review. Monitoring agencies are encouraged to use this example form, or similar form
tailored to the instruments at the site, to record onsite meteorological checks.
213
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
Weekly Quality Control Cheek Sh
Meteorological Instrumarits
set
Site
Month/Year
Site Number
Technician
Date:
Tower Check:
Crmsarm attained wrth north?
WS nips nfeay?
WD «art#» nktr/>
VWS prooeier cfcaw?
T & RH shHd oka/?
Solar radiation okav?
UV radiation okav?
Wind Check:
WS estimate 111
WS DAS (nvs\
mpn
mps
mps
mps
rtsps
mps
WS Chart %
%
%
*
%
%
%
WD estimate (9\ rri«rj>
OTU
dteg
dag
i§S
dag
J-L
WD DAS (rteqt
cleg
cleg
dag
dag
dag
dag
WD Chsml %
%
%
*
%
%
%
Temperature Check:
T drv Itfea C\
fegC
SftgC
d«yC
tte§C
d*gC
T wtt fitea C.\
¦eteaC
dtcsC
AaoC
deuC
daoC
AwC
RH % calculation 13)
%
%
%
%
%
%
Tt»mp HAS Mma CI
tfeaC
esmc
dsoC
tlsoC
ttooC
RH % flAfi
%
%
*
%
*
%
Sky Check:
Skv condition (41
SRDDAR fmMiffl
1IV DAS fmb)
All comments must be noted in the station tog.
(1) WS estimate
C • calm (0-1 nips)
L = light (1-3 mps j
M = moderate (4-6 mps)
S = strong (> 6 mps)
(2) WD esimate
N. NE. E,SE
S, SW. W, NW
(3) Calculated RH
From graph
on reverse side
(4) Sky condion (choose 1 or more)
CLR (clear) f (tog)
PC (partlycloudy) S (smog)
CLDY (ctady) H (naze'!
OVC (overcast) R (ram)
RwiMBd bw
Date
Figure 10-7. Example Meteorological Sensor Visual Checklist8
214
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
Validators will utilize the information recorded on these forms when validating the data collected
by the DAS.
In addition to the above checks, site operators should review documentation of meteorology
instrument calibration to ensure they have recorded calibration activities and reference standards
appropriately per the monitoring agency SOP and complete any information gaps as necessary.
10.10.1.2 Data Verification Performed by DAS
Monitoring agencies should be conducting data verification activities through checks facilitated
through the site DAS. The DAS is configured in advance (e.g., at setup) to establish limits for
each parameter to automatically flag data meeting certain criteria; typically these are alarm
thresholds established to flag high values, low values, high rates of change, or when data are
incomplete. The DAS should be configured such that separate alarm limits are set for
instantaneous data and hourly averages. These thresholds should be updated and refined based on
the season to include historical data for the season.
The suggested DAS automated alarm parameters are shown below for meteorology data:
• Values exceeding DAS maximum reading (hardware driven)
• Values less than the DAS minimum reading (hardware driven)
• Consecutive values exceeding a given maximum rate of change
• Values exceeding a seasonal maximum reading (user defined)
• Values exceeding a seasonal minimum reading (user defined)
• Hourly averages with less than 45 minutes of collected measurements
Monitoring agencies should establish additional parameters and automated checks as appropriate
that aid in identifying potential problematic data according to the individual site conditions. Site
operators should be performing such checks during routine review and technical reviewers
should verify these checks were performed, and if not, conduct them. The automated flags are
useful in identifying data that could be problematic, but these data may still be valid for
reporting. The reviewer should evaluate whether the flagged data represent meteorological
conditions at the site and whether such flagged data should remain flagged. In some cases,
extreme meteorological conditions can occur rapidly, and the data may reflect real conditions.
For example, if a thunderstorm moves through the site, winds can transition from calm to quickly
reach 20 to 30 m/s within seconds and be flagged by the DAS as exceeding a pre-defined rate of
change threshold. The reviewer should document the rationale for addressing validity of
automatically flagged data and alert data validators that such data were reviewed.
10.10.1.3 Technical Review of Meteorology Data
Technical review of meteorology data entails verifying the traceability of the meteorology data
from calibration and ensuring the data collection processes are correct and accurate according to
the monitoring agency SOPs. Technical reviewers will review the calibration data prior to
reviewing the collected environmental data to ensure the documentation trail is traceable,
calculations and transformations are accurate, and to ensure:
215
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
• Calibrations were performed at the required frequency and prior to generation of data
• Calibration procedures were performed in the proper order as prescribed in the SOP
• Calibrations include the proper number of calibration points
• Calibration ranges bracket the expected range of reported measurements
• Reference standards were within their certified calibration dates (where applicable)
• Linearity checks or operational checks were performed (where applicable) to ensure that
the measurement system was stable when the calibrations were established
Once the calibration data are verified, technical reviewers should review documentation of QC
checks for compliance with acceptance criteria, site logs, instrument maintenance logs, and
routine site visit check sheets for completeness and indications of unusual events that may
impact data. Technical reviewers should also review routine meteorology data for completeness
and verify that missing data or data flagged by the site operator or automatic DAS assessments
are appropriately qualified or invalidated. Technical reviewers should document the scope of
their review (parameters, date ranges, etc.), the materials reviewed (logs, electronic data, etc.),
and activities conducted during technical review so these records are available for data
validation.
10.10.2 Meteorology Data Validation
Once the data verification process has been completed for meteorology data, the data are
presumed to be technically correct and compliant with the established policies and procedures.
Data that do not meet technical acceptance criteria will have been qualified (flagged) or
invalidated, as appropriate per the monitoring agency policies and procedures. The monitoring
agency can then perform Level 1 data validation steps to examine the dataset for internal
consistency.
10.10.2.1 Level 1 Validation of Meteorology Data
Evaluating the internal consistency of the meteorology data should involve generation of simple
statistics to characterize the central tendency and the variability of the data. These statistics are
useful in identifying measurement values that appear to be abnormally high or low. As
previously discussed, meteorological parameters will fluctuate seasonally, and in general will
demonstrate a diurnal pattern, particularly temperature, solar radiation, UV radiation, and mixing
layer height.
Validators should prepare visualization plots of the meteorology data to aid in identifying
extreme values that warrant further investigation. Time series plots are particularly useful in
verifying an expected diurnal pattern or rapid changes in values that are not expected.
Instrument zero drift may be indicated when the daily minimum values deviate (increase or
decrease) from the expected minimum value over a period of several days. Validators should
review several consecutive days or weeks of data collected in the early morning hours (e.g., 3
a.m. to 4 a.m.) when winds are light and variable, solar radiation is minimal, and temperatures
216
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
are typically at their daily minimum. Reviewing daily minimum data over such a time period
may reveal a drift in the instrument baseline.
Preparing overlays of related meteorological and pollutant parameters on timeseries plots is
strongly recommended, including:
• Temperature and RH (direct relationship)
• Precipitation and RH (direct relationship)
• Ozone and ambient temperature (direct relationship)
• Wind speed and wind direction
These parameter combinations should also be plotted as scatter plots to identify pairs of data that
deviate from the expected relationship.
In addition, validators should examine values that standout from visual screening and data
flagged by automated DAS screening to confirm they are consistent with a meteorological cause.
Site operator and technical reviewer notes should indicate that data flagged by automated DAS
screening have been reviewed and the status of validity. Validators should review calibration
data and calibration check/QC check data to ensure calibrations were conducted at the proper
frequency, adjustments (recalibration) were made when instruments are out of tolerance, and
data since the last passing calibration or calibration check were appropriately flagged. Suspect
data values may be more closely reviewed and compared to NWS data from nearby stations.
Validators should document the scope of the data validated (parameters and date ranges), record
observations and actions taken to further investigate suspect data values, and document the
outcomes of changes, corrections, or data status changes (such as qualification or invalidation).
10.10.2.2 Level 2 Validation of Meteorology Data
For Level 2 validation of meteorological data, the data are compared to historical data at the site
to investigate values that stand out historically. It is useful to compare data seasonally by
generating simple statistics of historical data and verifying that current measurements are in line
with the historical data for the central tendency and variability. Preparing overlays of time series
plots of the parameter pairs listed in 10.10.2.1 is useful to identify values that stand out
seasonally. Significant deviations from historical measurements may be explained as unusual
weather events, or may indicate instrument or data processing problems requiring further
investigation and potential data correction, qualification, or invalidation.
10.10.2.3 Level 3 Validation of Meteorology Data
Once meteorology data have been validated for Level 1 and Level 2, they should be compared to
data from nearby sites, if possible, to investigate systematic bias. While variation is expected due
to the unique topographical and geographical nature of each monitoring site, known relationships
between sites can be examined. Such relationships may include typical wind directions,
temperature gradients, relative humidity conditions due to precipitation run off, and proximity to
urban environments or other variables that influence the local microclimates.
217
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
Comparison of site meteorology data should involve evaluating simple statistics, parameter
changes seasonally, and historic data at the sites. Discrepancies between sites are to be expected,
but may make extreme values or trends apparent which should prompt closer examination of
meteorology data.
10.10.2.4 Reporting Validated Data to A QS
Once data validation activities are complete and the data have been verified to be appropriately
qualified or invalidated and translated into AQS format, the data may be uploaded to AQS. Once
uploaded, the uploaded data should be retrieved and parity checks should be performed to ensure
the upload was accurate and complete. Monitoring agencies should maintain documentation of
such verifications.
10.11 Using Surface Meteorology Measurements for Data Validation
A key advantage to having meteorological data collected onsite is the ability to correlate the
occurrence of peak pollutant concentrations to wind conditions. Data analysis of the collected
pollutant data will be greatly enhanced by knowing whether winds are calm, parallel to a main
pollutant source, or at any other angle positioning the monitoring site relatively upwind or
downwind of a known source. In addition, plots of pollutants such as NO2 versus wind speed will
generally reveal higher concentrations with lower wind speeds due to reduced dilution of source
emissions. Similarly, preparation of pollution roses which plot pollutants as a function of wind
direction can reveal patterns in pollutant concentrations from specific wind sectors due to source
influences.
10.12 References and Further Reading
1. US EPA. (November 2002). Guidance on Environmental Data Verification and Validation.
EPA QA/G-8 (EPA Publication No. EPA/240/R-02/004). Office of Environmental
Information, Washington, DC. Available at (accessed November 2017):
https://www.epa.gov/sites/prodiiction/files/2015-06/dociiments/g8-final.pdf
2- http://www.ladco.org/reports/workshops/2011/data validation/presentations/session 1 intro
f
3. Guidance for Data Quality Assessment: Practical Methods for Data Analysis: EPA QA/G9:
QAOO Update: EPA/600/R-96/084. July 2000. Available at (accessed March 2018:
https://www.epa.gOv/sites/production/files/2.015-06/docum.en.ts/g9-final.pdf
4. MacGregor, I.C., Seay, B.S., Selection ofan MDL Measurement Quality Objective for
Formaldehyde Monitoring in the PAMS Network Based on Historical Measurement Results.
US EPA OAQPS Work Assignment 5-07, Project # 100095502, Contract EP-D-13-
00510/31/2017
5. Environment Canada and Health Canada. (1999). Canadian Environmental Protection Act.
Priority Substances List Assessment Report: Acetaldehyde. ISBN 0-662-28654-5. Cat. No.
En40-215/50E. (accessed February 2019).
http://publications.gc.ca/site/archivee-
arcMved.html?url=fa. ftp://pubtications.gc.ca/collections/Collection/Em -50E.pdf
218
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
6. Kumar, A., Alaimo, C.P., Horowitz, R., Mitloehner, F.M., Kleeman, M.J., Green, P.G. 2011.
Volatile Organic Compound Emissions from Green Waste Composting: Characterization and
Ozone Formation. Atmospheric Environment Vol. 45: pp 1841-1848. 2011 (accessed
February 2019).
https://pdfs.semanticscholar.ore/1473/t7f0780c407a510a56d70bed59afcc9da402.pdf
7. Singh, H.B., O'Hara, D., Herlth, D., Sachse, W., Blake, D.R., Bradshaw, J.D., Kanakidou,
M., Crutzen, P.J. 1994. Acetone in the Atmosphere - Distribution, Sources, And Sinks. J.
Geophys. Res.-Atmospheres. 99(D1): 1805-1819. 1994
8. Hafner, H. R. & Cavender, K, 2006, 'Level 2 Data Validation: Whoops! What the Grass
Roots Level Missed', Presented at The 2006 National Air Monitoring Conference, Las
Vegas, NV, November 6-9, 2006 (accessed March 2018).
https://www3.epa.eov/ttnamtil/files/2006conference/hafnerdat.pdf
9. US EPA. (January 2017) QA Handbook for Air Pollution Measurement Systems: "Volume
II: Ambient Air Quality Monitoring Program" EPA-454/B-17-001 (accessed March 2018).
10. US EPA. (March 2008). Quality Assurance Handbook for Air Pollution Measurement
Systems Volume IV: Meteorological Measurements Version 2.0 (Final). EPA-454/B-08-002.
Office of Air Quality Planning and Standards, Research Triangle Park, NC. Available at
(accessed March 2018):
epa. gov/ttnamti 1/files/ambient/p.
If
.epa.eov/ttnamtil/files/ambient/met/Voliime IV Meteorological Measurement
s.pdf
219
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
11.0 REPORTING DATA TO AQS
Following completion of data verification and validation activities, data are to be reported to
AQS. Data are to be reported to AQS within 180 days of the end of the calendar quarter in which
the measurements were made. During data verification and validation, data which are valid, but
which may not have met quality control criteria or are otherwise compromised, will have
appropriate qualifier codes added to the data so data users querying data in AQS are informed of
any data quality issues. Monitoring agencies should have staff responsible for coding the air
monitoring data for AQS and uploading the coded data.
11.1 Coding Ambient and Quality Assurance Data for AQS
This section covers reporting of ambient and QA data for carbonyls, speciated VOCs, and
meteorology data, as applicable. Monitoring agencies should follow the current approved
procedures for the NCore program for reporting the continuous gaseous parameters (ozone, true
N02, NO, and NOy) to AQS.
Monitoring agencies will need to amend or add site information and monitors at the PAMS
Required Sites to AQS; however, this is a relatively routine function for monitoring agencies to
perform, therefore this section will focus on coding and uploading routine ambient and QA data
to AQS and will not address AQS transactions used to setup sites and input basic site information
or to establish monitors. Additionally, each PAMS Required Site will have data handling
practices and procedures defined (in a QAPP, SOP, or similar), which may involve alternative
data coding practices for interim database submission, such as is needed for state-operated or
regionally-operated databases from which data are subsequently coded for AQS submission. As
such, this section describes general aspects for coding data for AQS input.
Briefly, AQS accepts data transactions, or inputs, from monitoring agencies for air monitoring
data in a pipe-delimited format. These transactions must be programmed in a specific way for
AQS to accept the information. The information contained in each data string consist of the
following types of information: codes, dates, numeric data, and alphanumeric data. Definitions of
these information types are detailed in the AQS Data Coding Manual, Version 3.6, (February 2,
2018), available at the following link:
https://www.epa.gov/sites/production/files/2018-
02/documents/user data coding manual feb 2018 O.pdf
Each data string, or transaction, consists of a series of fields, each separated by a pipe, "|" to
indicate the end of a field and the start of the next field. Depending on the transaction type, some
fields may be required and the information in the field must meet specific criteria as defined in
the business rules defined in the AQS Data Coding Manual.
The EPA has developed an AQS Transaction Generator program that will run in the Windows
operating system. This tool facilitates the creation of the AQS transactions and verifies
compliance with the AQS data and business rules to ensure the coded transaction will
220
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
successfully upload to AQS. Note that users will need to have administrator rights on their PC to
install the program, available at the following link:
https://www.epa.gov/aqs/aqs4ransaction-generator
Guidance for coding PAMS QA data for submission to AQS is provided in Appendix B.
11.2 Reporting PAMS Parameters to AQS
PAMS Required Site monitoring agencies are required to report data for each of the priority
chemical parameters listed in Table 2-2 and the meteorological parameters listed in Table 2-3.
Monitoring agencies are also encouraged to report data collected for those chemical parameters
listed as optional in Table 2-2. Careful attention must be paid to coding of data uploaded to AQS
to ensure that the five-digit parameter code is accurate and that the associated units comply with
those units AQS accepts. Monitoring agencies are highly encouraged to employ software (e.g.,
from a DAS, LIMS, or similar) or spreadsheet programs in which the various AQS codes and the
data outputs have been validated. Prior to submission of data to AQS, the monitoring agency
should have completed data validation and performed a spot check of the dataset to ensure that
the parameter code, parameter occurrence code (POC), unit code, method code, and any
associated qualifier or null codes are properly assigned. Data which are miscoded may not be
identified properly and may result in underestimation of completeness or may be rejected by
AQS.
PAMS Required Sites will likely have numerous monitors collecting data for monitoring
programs besides PAMS. Each individual monitor of a given type (speciated VOCs, carbonyls,
true NO2, or meteorology) and duplicate samples collected from a single monitor are assigned a
POC by the monitoring agency. There is no guidance on how monitoring agencies assign POCs
and discussion with monitoring agencies have indicated several monitors can be assigned the
same POC. Data uploaded to AQS indicate the assigned POC, but the POC does not indicate
whether the associated data are from a primary monitor, duplicate sample from the primary
monitor, or collocated monitor or sample. Due to the ambiguous nature of POC assignment, the
monitoring agency should prescribe and maintain a legend of POCs for minimally each of the
monitor types required for PAMS Required Site parameters. Monitoring agencies are encouraged
to include the POC assignments in their ANP and/or program QAPP.
AQS instructions for data upload are described in the AQS User Guide and additional AQS
manuals and guides available at the following URL:
http://www3.epa.gov/ttn/airs/airsaqs/manuals/
Additional assistance is available by calling the AQS help line at (866) 411-4372.
11.3 AQS Reporting Units
Data may be coded with associated units for AQS upload for PAMS parameters in any
appropriate unit accepted for that parameter by AQS. Recommended units for reporting data to
221
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
AQS for each parameter are shown in Table 11-1.
Table 11-1. AQS Parameters and Recommended Reporting Units
Piii'iiiiHMcr
AQS
Piii'iimeler
Cwk'
Dui'iilinn
AQS
Dui'iilinn
Code
Recommended
Reported I nil
AQS
I nil
Code
Speciated VOCs by Auto-GC
refer to
Table 2-2
hourly average
1
ppbC
78
True NO2
42602
hourly average
1
ppm
7
Ozone
44201
hourly average
1
PPb
8
Carbonyl Compounds by TO-11A
refer to
Table 2-2
8-hour average
|ig/m3 at 25°C
1
Ambient Temperature
62101
hourly average
1
°C
17
Relative Humidity
62201
hourly average
1
% relative humidity
19
Barometric Pressure
64101
hourly average
1
millibar (hPa)
16
Wind Speed
61103
hourly average
1
m/s
11
Wind Direction
61104
hourly average
1
degrees compass
14
Solar Radiation
63301
hourly average
1
Watt/m2
79
Ultraviolet Radiation
63302
hourly average
1
Watt/m2
79
Precipitation
65102
hourly average
1
mm
29
Mixing Layer Height
61301
hourly average
1
m
58
Once uploaded to AQS, data may be queried through established reports and different units may
be specified; AQS converts to various units specified by the user, as appropriate. For example,
carbonyls data may be reported to AQS in |ig/m3 but may be converted when queried from AQS
as ppbC, ppbV or ng/m3.
11.4 Corrections to Data Uploaded to AQS
If it is discovered during data validation, as a result of corrective action, or through other means
that erroneous data have been reported to AQS, the data should be corrected and the updated data
uploaded to AQS. Situations where this may occur could result in previously acceptable data
being invalidated as a result of an audit, or data that were initially incorrectly invalidated could
be deemed valid. Monitoring agencies should notify EPA Region staff when a significant amount
(as determined by the monitoring agency) of data are discovered and require updating in AQS.
Monitoring agencies should coordinate with the EPA Region to correct the records in AQS, as it
is important to ensure that data end users are notified of data that may have been updated due to
the potential impact on decision-making.
11.5 AQS Qualifiers
The monitoring agency should identify compromised data within AQS by addition of a qualifier
or combination of qualifiers. Qualifiers associated with PAMS data are indicated in Table 11-2
below. Note that at the time this TAD was published, qualifiers for specific situations were not
222
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
available in AQS; however, EPA periodically updates the AQS qualifier list which is published
at the following link:
https://aqs.epa.eov/aqsweb/dociiments/codetables/qualifiers.html
Compromised data should either be flagged or invalidated in AQS as described below.
Flagging Data in AQS: Compromised monitoring data will be flagged in AQS only if the data
are considered valid for most purposes and uses. AQS permits users to label each data point with
up to 10 QA Qualifiers and/or Informational (INFORM) Qualifiers.
Invalidating Data in AQS: Data of uncertain origin, data with unacceptable levels of
uncertainty, or data which are known to not be an ambient measurement will not have an
associated measurement value included in AQS. Such data may be the result of instrument
failure, known instrument contamination, irrecoverable data corruption, or measurements
associated with failed routine QC checks, calibration, or determination of MDLs or instrument
troubleshooting. Invalid data are reported to AQS with a Null (NULL) Code Qualifier which
eliminates the associated measurement value and indicates the reason for the invalidation. AQS
accepts a single NULL qualifier and does not permit addition of other qualifiers to the
transaction string.
As discussed further below, data should be qualified and estimated with descriptive QA and
INFORM flags where the data are compromised but remain valid. Incorrect use of null codes
eliminates the measurement value in the AQS transaction string.
Note: EPA intends to addfunctionality to AQS to automatically add QA qualifier flags to low
concentration pollutant data according to their proximity to the MDL.
Table 11-2. AQS Qualifiers for PAMS
Qiiiilil'k'r
( ixk-
AQS Qiiiilil'k'r Disi i iplinn
QiiiililliT
T\ |K"
Commi'iil
1
Deviation from a CFR/Critical
Criteria Requirement
QA
Substitute a descriptive QA qualifier where possible
2
Operational Deviation
QA
Substitute a descriptive QA qualifier where possible
3
Field Issue
QA
Substitute a descriptive QA qualifier where possible
4
Lab Issue
QA
Substitute a descriptive QA qualifier where possible
5
Outlier
QA
7
Below Lowest Calibration Level
QA
DI
Sample was diluted for analysis
QA
Applies to carbonyls only
EH
Estimated; Exceeds Upper Range
QA
FB
Field Blank Value Above Acceptable
Limit
QA
HT
Sample pick-up hold time exceeded
QA
Applies to carbonyls only
LB
Lab blank value above acceptable
limit
QA
Applies to carbonyls and speciated VOCs
223
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
Table 11-2 (continued). AQS Qualifiers for PAMS
LJ
Identification of Analyte is
Acceptable; Reported Value is an
Estimate
QA
Probably the most common qualifier when an estimate is
needed
LK
Analyte Identified; Reported Value
May Be Biased High
QA
Use in place of LJ when direction of bias is known
LL
Analyte Identified; Reported Value
May Be Biased Low
QA
Use in place of LJ when direction of bias is known
MD
Value less than MDL
QA
ND
No Value Detected
QA
The analyte was not positively identified - should
accompany a measurement value of 0
NS
Influenced by nearby source
QA
Rare - in most situations such data should be invalidated
QX
Does not meet QC criteria
QA
SQ
Values Between SQL and MDL
QA
The SQL is defined as 3.18-fold the MDL value
ss
Value substituted from secondary
monitor
QA
Rare - most sites will not have collocated instruments
sx
Does Not Meet Siting Criteria
QA
Should require invalidation, but no associated null code
exists
TB
Trip Blank Value Above Acceptable
Limit
QA
Applies to carbonyls only
IT
Transport Temperature is Out of
Specs.
QA
Applies to carbonyls only
V
Validated Value
QA
Data should be validated when uploaded to AQS, this
code is not necessary but may identify suspect values that
have gone through additional scrutiny
VB
Value below normal; no reason to
invalidate
QA
AC
Construction/Repairs in Area
NULL
AD
Shelter Storm Damage
NULL
AE
Shelter Temperature Outside Limits
NULL
AF
Scheduled but not Collected
NULL
AG
Sample Time out of Limits
NULL
AH
Sample Flow Rate out of Limits
NULL
Would rather qualify than invalidate
AI
Insufficient Data (cannot calculate)
NULL
Should be used in situations where data were collected
for < 75% of the hour or the sampling period for VOCs is
< 40 minutes
AM
Miscellaneous Void
NULL
Substitute a more descriptive code where possible
AN
Machine Malfunction
NULL
AP
Vandalism
NULL
AQ
Collection Error
NULL
AR
Lab Error
NULL
AS
Poor Quality Assurance Results
NULL
Would rather qualify than invalidate, severity dependent
AT
Calibration
NULL
Applies when data represent instrument calibration
AU
Monitoring Waived
NULL
AV
Power Failure
NULL
AW
Wildlife Damage
NULL
AX
Precision Check
NULL
Applies when data represent instrument precision check
AY
QC Control Points (zero/span)
NULL
Applies when data represent instrument QC checks
AZ
QC Audit
NULL
Used for analysis of the VOCs PT sample and TTP audits
for ozone and NO2
BA
Maintenance/Routine Repairs
NULL
224
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
Table 11-2 (continued). AQS Qualifiers for PAMS
BB
Unable to Reach Site
NULL
BE
Building/Site Repair
NULL
BH
Interference/co-
elution/misidentification
NULL
Applies to auto-GC parameters only
BI
Lost or damaged in transit
NULL
Applies to carbonyls only
BJ
Operator Error
NULL
BK
Site computer/data logger down
NULL
DA
Aberrant Data (Corrupt Files,
Aberrant Chromatography, Spikes,
Shifts)
NULL
DL
Detection Limit Analyses
NULL
MC
Module End Cap Missing
NULL
Applies to carbonyls only
SC
Sampler Contamination
NULL
TC
Component Check & Retention Time
Standard
NULL
TS
Holding Time or Transport
Temperature is Out of Specs.
NULL
Would prefer to use QA qualifier instead
XX
Experimental Data
NULL
Used for troubleshooting, instrument conditioning, etc
IC
Chem. Spills & Indust Accidents
INFORM
Rare
ID
Cleanup After a Major Disaster
INFORM
Rare
IE
Demolition
INFORM
Rare
IH
Fireworks
INFORM
Rare
II
High Pollen Count
INFORM
Rare
IJ
High Winds
INFORM
Rare
IK
Infrequent Large Gatherings
INFORM
Rare
IM
Prescribed Fire
INFORM
Rare
IP
Structural Fire
INFORM
Rare
IQ
Terrorist Act
INFORM
Rare
IR
Unique Traffic Disruption
INFORM
Rare
IS
Volcanic Eruptions
INFORM
Rare
IT
Wildfire-U. S.
INFORM
Rare
J
Construction
INFORM
Rare
11.5.1 AQS Qualification for Low Concentration Data
Concentration data uploaded to AQS will be qualified/flagged according to whether they are
above or below the sample quantitation limit (SQL) or method detection limit (MDL) thresholds
(refer to Section 3 for further information regarding MDL and SQL). Concentration data less
than the determined MDL are to be flagged with the QA qualifier code MD, values greater than
or equal to the MDL but less than the SQL (3.18-fold the MDL) are to be flagged using the QA
qualifier code SQ. All concentration values for qualitatively (positively) identified analytes, even
those less than MDL, are to be reported to AQS and should not be censored by substitution of
one half the MDL, by replacement with 0, or by any other substitution method. Negative
concentrations should not be translated to zero for reporting purposes. Where qualitative
identification acceptance criteria are not met for a given parameter, its concentration must be
225
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
reported as zero and flagged as ND. The convention for reporting concentration data and the
associated QA flags are shown below in Table 11-3.
Table 11-3. AQS Quality Assurance Qualifier Flags for Various Concentrations Compared
to a Laboratory's MDL and SQL
( oncoilll'illioil I.C'M'I
Value
Associiilcd M)S i)\ l-'hiu
> SQL
measured concentration
no flag
> MDL and < SQL
measured concentration
so
< MDL
measured concentration
MD
Parameter not qualitatively identified
0
ND
The MDL for a given parameter is to be reported to AQS along with the measured concentration
to be a valid AQS transaction string. For carbonyls parameters, the reported MDL should ideally
be normalized to the collected air volume for the respective air sample. For example, the target
collected air volume for carbonyls sampling at 1.0 L/min is 0.48 m3 and the formaldehyde MDL
is 0.098 |ig/m3 for this target volume. For a total collected sample volume of 0.42 m3, the MDL
is normalized as follows (MDL increases by the -12% to account of the reduced sample
volume):
0.098 f^g/m3 • 0.48 m3 = 0.11 |ig/m3
0.42 m3
226
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
APPENDIX A
EPA ROUNDING GUIDANCE
Provided by EPA Region IV
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
Rounding Policy for Evaluating NAAQS QA/QC Acceptance Criteria
The following outlines EPA's Rounding Policy for evaluating Quality Assurance / Quality
Control (QA/QC) acceptance criteria. This policy is being provided to air monitoring
organizations in order to ensure consistency across the country in the validation of monitoring
data that is used for demonstrating compliance with the National Ambient Air Quality Standards
(NAAQS).
EPA's interpretation of standard rounding conventions is that the resolution of the measurement
device or instrument determines the significant figures used for rounding. The acceptance criteria
promulgated in the appendices of 40 CFR Part 50, or otherwise established in EPA guidance
documents, are not physical measurements. As an example, the quality control (QC) acceptance
criterion of ±5% stated in the fine particulate matter regulations (40 CFR Part 50, Appendix L,
Section 7.4.3.1) is not a measurement and, as such, does not directly contribute to either the
significant figures or to rounding. However, the flow rate of the sampler - measured either
internally by the flow rate control system or externally with a flow rate audit standard - is a
measurement, and as such, will contribute to the significant figures and rounding. EPA's position
is that it is not acceptable to adjust or modify acceptance criteria through rounding or other
means.
Example using PM2.5 Sampler Design Flow Rate
40 CFR Part 50, Appendix L, Section 7.4.3.1 defines the 24-hour sample flow rate acceptance
criterion as ±5% of the design flow rate of the sampler (16.67 liters per minute, LPM). The QC
acceptance criterion of ±5% stated in regulation is not a measurement and, therefore, does not
contribute towards significant figures or rounding. The measurement in this example is the flow
rate of the sampler. PM2.5 samplers display flow rate measurements to the hundredths place
(resolution) - e.g., 16.67 LPM, which has 4 significant figures. Multiplying the design flow rate
(16.67 LPM) by the ±5% acceptance criterion defines the acceptable flow regime for the
sampler. By maintaining 4 significant figures - with values greater than 5 rounding up - the
computations provide the following results:
• The low range is -5% of the design flow: 0.95x16.67=15.8365-15.84
• The upper range is+5% of the design flow: 1.05x16.67=17.5035-17.50
Rounding in this manner, the lower and upper acceptance limits for the flow rate measurement
are defined as 15.84 and 17.50 LPM, respectively.
40 CFR Part 58, Appendix A, Section 3.2.1 requires monthly PM2.5 flow rate verifications. The
verification is completed with an independent audit standard (flow device). The monthly check
includes a calculation to ensure the flow rate falls within ±5% of the design flow rate (see
A-l
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
Method 2.12, Section 7.4.7). Therefore, flow rates obtained during monthly flow rate verification
checks should measure between 15.84 - 17.50 LPM, as defined above.
Measurements, in general, are approximate numbers and contain some degree of error at the
outset; therefore, care MUST be taken to avoid introducing additional error into the final results.
With regards to the PM2.5 sampler's design flow rate, it is not acceptable to round the ±5%
acceptance criterion such that any calculated percent difference up to ±5.4% is acceptable -
because rounding the acceptance criterion increases the error in the measurement. It is important
to note that the PM2.5 sampler MUST maintain a volumetric flow rate of approximately 16.67
LPM in order for its inertial separators to appropriately fractionate the collected ambient air
particles.
Flow rates greater than 5% of the nominal 16.67 LPM will shift the cut point of the inertial
separator lower than the required aerodynamic diameter of 2.5 microns and, thus, block the
larger fraction of the PM2.5 sample from being collected on the sample filter. Conversely, as the
sampler's flow rate drops below -5% of the nominal 16.67 LPM, the inertial separator will allow
particulate matter with aerodynamic diameters unacceptably larger than 2.5 microns to be passed
to the sample filter. Therefore, it is imperative that the flow rate of the sampler fall within the
±5% acceptance criterion.
A Note on Resolution and Rounding
Measurement devices will display their measurements to varying degrees of resolution. For
example, some flow rate devices may show measurements to tenths place resolution, whereas
others may show measurements to the hundredths place. The same holds true for thermometers,
barometers, and other instruments. With this in mind, rounding should be based on the
measurement having the least number of significant figures. For example, if a low-volume PM10
sampler displays flow rate measurements to the tenths place (3 significant figures), but is audited
with a flow device that displays measurements to the hundredths place (4 significant figures), the
rounding in this scenario will be kept to 3 significant figures.
Table 1 below lists some examples of NAAQS regulatory QA/QC acceptance criteria with
EPA's interpretation of the allowable acceptance ranges, as well as a column that identifies
results that exceed the stated acceptance limits. Table 1 is not a comprehensive list of ambient air
monitoring QA/QC acceptance criteria. Rather, Table 1 is provided to demonstrate how EPA
evaluates acceptance criteria with respect to measurement resolution.
The validation templates in the QA Handbook Vol II will be revised to meet this policy.
If you have any questions regarding this policy or the rounding conventions described, please
contact your EPA Regional Office for assistance.
A-2
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
Table 1: Examples of Quality Control Acceptance Criteria
Regulatory
Method
Requirement
Method
Acceptance
Criteria
Typical
Measurement
Resolution
Acceptance Range
(Passing Results)
Exceeding
QA/QC Check
Shelter
Temperature
20 to 30°C or
FEM op. range
1 Decimal, 3
SF*
20.0 to 30.0°C or
FEM op. range
< 19.9°C
>3o.rc
PM2.5 Design
Flow (16.67 1pm)
±5%
2 Decimal, 4
SF
15.84 to 17.501pm
<-5.1%
>+5.1%
PM2.5 Transfer
Standard
Tolerance
±4%
2 Decimal, 4
SF
-4% Audit
Std
Sampler
Display
+4%
Audit Std
<-4.1%
>+4.1%
15.84
16.47
16.00
16.67
1 7 34
16.80
17.50
PM2.5 Lab:
Mean Temp
24-hr Mean
20 to 23°C
1 Decimal, 3
SF
20.0 to 23.0°C
< 19.9°C
>23.1°C
PM2.5 Lab:
Temp Control
SD over 24-hr
±2°C
1 Decimal, 3
SF
±2.0°C
<-2.rc
>+2.rc
PM2.5 Lab:
Mean RH
24-hr Mean
30% to 40%
1 Decimal, 3
SF
30.0% to 40.0%
< 29.9%
>40.1%
PM2.5 Lab:
RH Control
SD over 24-hr
±5%
1 Decimal, 3
SF
±5.0%
<-5.1%
>+5.1%
PM2.5 Lab:
Difference
in 24-hr RH
Means
±5%
1 Decimal, 3
SF
±5.0%
<-5.1%
>+5.1%
*SF = Significant Figures
A-3
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
APPENDIX B
AQS Coding Guidance For
PAMS Quality Assurance Data
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
APPENDIX B: Reporting PAMS QA Data to AQS
PAMS QA data reported to AQS includes Blanks and Precision Field QC (Collocated and
Duplicate) and Laboratory Samples (Analysis Replicate).
BLANK SAMPLE REPORTING
Blank samples for the PAMS program are analyzed for carbonyls by EPA Compendium Method
TO-11A and speciated VOCs by auto-GC.
Carbonyls Blank Sample Reporting
Blank samples for carbonyls in the PAMS program consist of field blanks, trip blanks, lot blanks,
laboratory method blanks, and exposure blanks. Monitoring agencies are to report field blank,
trip blank, and lot blank data to AQS. Optionally, monitoring agencies may also report
laboratory method blanks and exposure blanks.
To report blank data, submit a raw blank (RB) transaction for each blank sample. The Blank
Types for the various blanks are:
For example, for a field blank, the Blank Type field is entered as "FIELD" (bold in example
below).
RB|I|11|222|3333|44444|9|7|454|888|FIELD|2015010110 0:0 0|0.0463| | | | | | | | | | | |0.0001 |
Speciated VOCs Blank Sample Reporting
Blank samples for speciated VOCs in the PAMS program consist of daily system blanks. These
blanks are reported similarly to the various blanks collected and analyzed for carbonyls, except
the Blank Type is "LAB" to indicate an analysis blank.
PRECISION SAMPLE REPORTING
Duplicate and replicate analyses are defined and reported in the PAMS and NATTS programs for
carbonyls. Collocated data reporting is used in both the SLAMS and NATTS programs for
collocated monitors. The purpose of this section is to clarify how data from these assessments
should be reported to AQS using QA transaction formats. (Please note, the old AQS "RA" and
"RP" transactions have been retired and can no longer be used to report data.) The goal is to
provide consistent reporting terms and procedures to allow the data to be universally understood.
Field blank:
Trip blank:
Lot blank:
FIELD
TRIP
LOT
LAB
Laboratory Method Blank:
Exposure Blank:
FIELD 24HR
B-l
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
Simplified schematics are included in this article for illustrative purposes and do not address
specifics related to different sampling approaches or methodologies.
The AQS transaction formatting descriptions are not repeated in this document, but may be
found on the AQS website:
https://aqs.epa.gov/aqsweb/documents/TransactionFormats.html
Collocated Samples
Collocated samples are samples collected simultaneously at the same location using two
completely separate sampling systems, each with a separate inlet probe to the ambient sampled
atmosphere. The allowable distance between inlet probes is defined in regulations or program
guidance. Both of the monitors (each designated by a separate AQS Parameter Occurrence Code
- POC) have been established in AQS already for the site. The samples are collected and
analyzed separately. Each is reported as a sample value for the appropriate monitor.
Collocated Sample Schematic
Collocated Samples
Collocated samples are samples collected simultaneously at the same
location using two completely separate sampling systems, each with a
separate inlet probe to the ambient sampled atmosphere.
Primary
Monitor N
Collocated
Monitor C
0
0
Two completely separate sampling systems at same location, two different samples analyzed.
Collocated Sample Reporting Instructions
For AQS to automatically create the 'precision pair' for the primary and collocated samples, the
monitors must be identified to the system as QA collocated. One monitor must be designated as
the QA primary. If using transactions, the Monitor Collocation Period (MJ) transaction is used.
(If using the AQS application, the "QA Collocation" tab on the Maintain Monitor form may be
used to enter these data.) The collocation data must be entered for both monitors, with one
indicated as the primary, and the other indicated as the collocated (not the primary). In the
example below, the primary monitor is indicated by the bolded ' Y' (yes, this is the primary) in
the Primary Sampler Indicator in the first MJ string and the collocated monitor by the bolded 'N'
(no, this is not the primary) in the Primary Sampler Indicator in the second MJ string.
Once the monitors have been identified as collocated, there are no additional reporting
requirements; simply report the raw data from each monitor (From the schematic, value 'a' from
B-2
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
the primary monitor 'N' and value 'b' from the collocated monitor 'C'). Once this is done, AQS
will know to pair data from these two monitors for the date range specified.
A set of transactions must be created for each time period the monitors are operating together.
The transactions have a begin date and end date for the operational period. The end date may be
left blank if the collocation period is still active (as indicated in the example below). To define a
collocation, submit two MJ transactions (example below with differences bolded and where
primary monitor 'N' is POC 5 and collocated monitor 'C' is POC 9):
MJ|I11112 2 2|3333144444|5|20150101| |3|Y
MJ|I11112 2 2|3333|44444|9|201501011 |3|N
Report two Raw Data (RD) transactions for each time sample data are to be reported from both
monitors; one for each monitor (POC). (In this example, sample 'a' is 0.0463 from monitor 'N'
(POC 5) and sample 'b' from monitor 'C' (POC 9) is 0.0458):
RD|I11112 2 2|3333144444|5|7|454|888|20150101|00:00|0.0463| |6| | | | | | | | | | | |0 . 00011
RD|I11112 2 2|3333|44444|9|7|454|888|20150101100:00|0.0458| |6| | | | | | | | | | | |0.00011
Since there are two monitors involved, each sample is reported for its appropriate POC and there
will be an RD transaction for every time there is a valid sample from each monitor (e.g., two per
day in this scenario). If the sample value from one POC is not available, report a Null data
qualifier code for that monitor (that is, do not report the sample value from the collocated
monitor as being from the primary POC).
Duplicate Samples
Duplicate samples are two (or more) samples collected simultaneously using one or more
sampling units sharing a common inlet probe to the ambient atmosphere and the collected
samples are analyzed separately. This simultaneous collection may be accomplished by "teeing"
the line from the flow control device (sampling unit) to the media (e.g. DNPH cartridges), and
then doubling the collection flow rate, or may be accomplished by collecting one discrete sample
via two separate flow control devices (sampling units) connected to the same inlet probe, such as
two sampling units connected to a common manifold inlet.
B-3
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
Duplicate Samples Schematic
Duplicate Samples
Duplicate samples are two (or more) samples collected simultaneously
using one or more sampling units sharing a common inlet probe to the
ambient atmosphere and the collected samples are analyzed separately
or
Primary
Monitor N
Collocated
Monitor C
0
&
One sampling system inlet probe, two different samples analyzed.
Duplicate Sample Reporting Instructions
In this case, there is only one inlet probe involved but with multiple samples. Since only one
inlet probe is involved, all data should be reported for the same POC.
First, report the raw data as you normally would via the RD transaction. Report just one value,
the one for the sample obtained through the 'primary' hardware (the normal flow path or normal
cartridge, etc. as defined by the monitoring organization convention - typically this would be
sample 'a'). In this case, if sample 'a' comes from the primary hardware and has a value of
54.956, you would report:
RD |I11112 2 2|3333|44444|5|7|454|888|20150101|00:00|54.9561 |6| | | | | | | | | | | |0.00011
If the primary value is null for some reason, the duplicate value may be reported as the sample
value for this POC in the RD transaction. In this case, there is not a valid duplicate assessment
to report. If all duplicates are null, an RD transaction with no sample value and a Null data
qualifier code should be reported.
Each of the duplicate sample values is then also reported via the QA - Duplicate transaction.
This transaction has room for up to 5 duplicate sample values. Report them in any order, starting
with 1 and proceeding through the number of samples. In the schematic, there are two samples
(a 'primary' and a 'duplicate') so sample value 'a' would be reported as Duplicate Value 1 and
sample value 'b' would be reported as Duplicate Value 2. The same value reported on the Raw
Data transaction must be one of the values reported on the QA - Duplicate transaction.
B-4
-------
PAMS Required Site Network TAD
EPA-454/B-19-004
April 2019
Note that there is no sampling time reported on the QA - Duplicate transaction. Instead, there is
an Assessment Date and an Assessment Number. If multiple duplicate samples are performed on
the same day, label the first with Assessment Number = 1, the second with Assessment Number
= 2, and so on. Also note that all values must be reported in the same units of measure.
Here is an example QA - Duplicate transaction (with sample 'a' = 54.956 and sample 'b' =
51.443 - Assessment Number ' 1' bolded):
QA| I|Duplicate|999|11|222|3333|44444|5|20150101|11454|888|54.956|51. 443 | | | |
Replicate Analysis
A replicate assessment is a separate analysis or multiple separate analyses of one discrete sample
(a carbonyls sample extract) to yield multiple measurements from the same sample.
Replicate Analysis Schematic
Replicate Samples
T
Monitor N
A replicate assessment is a separate analysis or multiple separate
analyses of one discrete sample (carbonyls extractto yield multiple
measurements from the same sample.
One sampling system, one sample, multiple analyses.
Replicate Sample Reporting Instructions
Again in this case, there is only one AQS monitor (POC) involved and one single sample,
however multiple analyses of the sample.
First, report the raw data as you normally would via an RD transaction. Report just one value,
according to your laboratory's convention for reporting replicate data (e.g. the first replicate). In
this case, if you have chosen replicate 'a' as your raw data value and it has a value of 0.844, you
would report:
RD|I|11|222|3333|4 44 44|5|7|454|8 88|2 0150101|00:00|0.8441|6||||||||||||0.0001|
If the normally reported value is null for some reason, one of the other replicate values may be
reported as the sample value for this POC in the RD transaction. If only one of the replicate
values remains valid, there is not a valid replicate assessment to report. If all replicates are null,
an RD transaction with no sample value and a Null data qualifier code should be reported.
Once the RD transaction is completed, if two or more replicates are valid, these are reported via
the QA - Replicate transaction. This transaction has room for up to 5 replicate sample values.
B-5
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
Report them in any order, starting with 1 and proceeding through the number of samples. In the
schematic above there are three replicates 'a', 'b', and 'c' for monitor N, thus analytical value 'a'
would be reported as Replicate Value 1, analytical value 'b' would be reported as Replicate
Value 2, and analytical value 'c' would be reported as Replicate Value 3.
Note that there is no sampling time reported on this transaction. Instead, there is an Assessment
Date and an Assessment Number. If multiple replicate samples are collected on the same day,
label the first with Assessment Number = 1 (indicated below in bold), the second with
Assessment Number = 2, and so on. Also note that all values must be reported in the same units
of measure.
Here is a sample QA - Replicate transaction (if sample values 'a', 'b', and 'c' are 0.844, 0.843,
and 0.792, respectively):
QA|I|Replicate|999|11|222|333|44444|5|2 0210101|1|454|888|0.844|0.843|0.7 92|||
Combining Duplicates and Replicate Analysis
It is possible to collect duplicate samples simultaneously and perform replicate analyses of these
duplicate samples. This is often referred to as a duplicate/replicate sample. In this case (see
schematic below), there are two duplicate samples, ' 1' and '2'. Duplicate Sample ' 1' has three
replicates: 'a', 'b', and 'c\ Duplicate Sample '2' has three replicates: 'd', 'e', and 'f.
Replicates of Duplicate Samples Schematic
Duplicate / Replicate Samples
Monitor N
It is allowable (but not a requirement) to perform replicate analyses of
duplicate samples. In this case two duplicate samples are collected,
and each is "replicate" analyzed three times.
One sampling system, two samples, multiple analyses of each.
Duplicate/Replicate Reporting Instructions
This scenario requires the reporting of an RD transaction, a QA - Duplicate transaction, and a
QA - Replicate transaction to AQS.
For the RD transaction, follow the same rules to report the value from the primary (normal)
hardware (this would typically be sample ' 1', replicate 'a') and operations procedure path if
possible; follow the convention established by the laboratory. If the normal hardware path yields
sample 'la' you would report (in this case the value is represented by the "a" in the appropriate
place, with spaces for clarity):
B-6
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
RD | I | 11122 2|3333|44444|5|7|454|888|20150101100:00| a | |6| I I I I I I I I I I I 0.00 011
For the QA - Duplicate transaction: select one of the replicate analyses each from the primary
and duplicate sample (using the convention established by the laboratory) and report those on the
QA - Duplicate transaction. If the values to be reported are 'la' and '2d', the record would look
like this (again, values are represented by 'a' and'd', spaces added for clarity):
QA|IIDuplicate|999|11|222|333|44444|5|20210101|1|454|888| a | d ||||
There are only two duplicate samples (one pair) in this case because only two paths were
assessed. (That is, you are not allowed to cross-multiply the replicate analyses to create
additional duplicate assessments [pairs].)
For the replicate transaction: report this as two assessments. Assessment Number 1 for the day
would include the values for replicates 'a', 'b', and 'c'. Assessment Number 2 for the day would
include values for replicates 'd', 'e', and 'f.
The example transactions, using letters in place of the values:
QA|IIReplicate|999|11|222|333|44444|5|20210101|1|454|888| a | b | c |||
QA|IIReplicate|999|11|222|333|44444|5|20210101|2|454|888| d | e | f |||
Combining Collocated Samples and Replicate Analysis
It is also possible to make replicate analyses of collocated samples. These are sometimes
referred to as collocated replicate samples.
Replicates of Collocated Samples Schematic
Collocated Replicate Samples
Primary
Monitor N
Collocated replicate samples are also possible.
Collocated
Monitor C
f
Two completely separate sampling systems, two different samples analyzed multiple times.
Collocated Replicate Reporting Instructions
Since collocated monitors report all data independently, report these data for each monitor (e.g.,
under its own POC) according to the replicate reporting instructions.
B-7
-------
PAMS Required Site Network TAD EPA-454/B-19-004
April 2019
Reporting of Proficiency Test Sample Results
Monitoring sites analyzing proficiency test (PT) samples for speciated VOCs and analytical
support laboratories (ASLs) analyzing PT samples for carbonyls should report their results to
AQS.
Proficiency Test Sample Reporting Instructions
For the QA - Lab Proficiency Test transaction: The AQS transaction string should be composed
as follows:
QA|I|Lab Proficiency Test|9999|8888|43502|20180101|1|077|1.21|1.17|
where the items described in the sequence are defined as:
• QA = quality assurance transaction
• I = action indicator (insert)
• Lab Proficiency Test = assessment type
• 9999 = performing agency
• 8888 = primary quality assurance organization (PQAO)
• 43502 = parameter code (formaldehyde in this example)
• 20180101 = assessment date in format YYYYMMDD
• 1 = assessment number (should be 1 unless multiple assessments for the same parameter
on the same date)
• 077 = units (micrograms in this example)
• 1.21 = laboratory response value
• 1.17 = assessment mass (assigned PT value)
B-8
-------
United States Office of Air Quality Planning and Standards Publication No. EPA-454/B-19-004
Environmental Protection Air Quality Assessment Division April 2019
Agency Research Triangle Park, NC
------- |