EPA/600/R-21/287 | November 2021 www.epa.gov/research
/ Polyurethane Foam Insulati
nj .pcay Application in a Full
tfist Chamber I
ng'and Foil
Officepf Research and Development
Center for Environmental Measurement and M<
-------
EPA Project Data Summary Report
Office of Research and Development
Center for Environmental Measurement and Modeling
Air Methods and Characterization Division/Source and Fine Scale Branch
Development of a Test Method to Characterize Emissions from
Spray Polyurethane Foam Insulation During and Following
Spray Application in a Full-Scale Test Chamber
EPA CEMM Technical Lead: Mark Mason
EPA/600/R-21/287
November 16, 2021
Prepared by:
Mark Mason, Dale Greenwell, Mark Barnes, Ken Krebs
US EPA CEMM AMCD
Gary Folk, John Ulrich
Jacobs Technology Contract EP-C-15-008
SPFI Methods Development
Page 2
-------
Disclaimer
This document is for informational purposes only and is not intended as legal advice. The
contents are for general informational purposes and should not be construed as legal advice
concerning any specific circumstances. You are urged to consult legal counsel concerning any
specific situation or legal issues. This document does not address all federal, state, and local
regulations, and other rules may apply. This document does not substitute for any EPA
regulation and is not an EPA rule. Any mention of trade names, manufacturers or products does
not imply an endorsement by the United States Government or the U.S. Environmental
Protection Agency. EPA and its employees do not endorse any commercial products, services,
or enterprises. The findings and conclusions in this report have not been formally disseminated
by the Agency and should not be construed to represent any Agency determination or policy.
SPFI Methods Development
Page 3
-------
Contents
LIST OF TABLES 10
LIST OF FIGURES 12
ACRONYMS AND ABBREVIATIONS 15
EXECUTIVE SUMMARY 17
LIMITATIONS OF THE STUDY 18
1.0 INTRODUCTION 20
1.1 Nature of the Problem 20
1.2 Background 20
1.2.1 Exposure and Health Concerns 21
1.2.2 EPA's SNAP Program and SPF Insulation 21
1.2.3 Research Needs 21
1.3 Objectives of the Study 21
1.4. Types of SPF Insulation 22
1.5 Consensus Method Development Process 23
1.5.1 Development of Draft ASTM Full-Scale Test Protocol 23
1.5.2 Need for an Integrated Consensus Full-Scale Emissions Test Protocol 25
1.5.3 EPA Support to Development of a Consensus Full-Scale Emissions Test Protocol 26
1.6 Overview of the Method 27
1.6.1 Phases of the Test Protocol 28
1.6.2 Data Analysis 29
1.7 Health and Safety 29
2.0 MATERIALS AND METHODS 31
2.1 Product Selection 31
2.1.1 Emissions to be Measured 31
2.2 Test Facility 32
2.2.1 Full-Scale Emissions Test Chamber 33
2.2.2 Substrate Frames for Application of SPF Insulation 34
SPFI Methods Development Page 4
-------
2.3 Application Procedure 34
2.3.1 Preparation for the Spray Events 34
2.3.2 Spray Event #1 35
2.3.3 Spray Event #2 35
2.4 Sampling Plan 35
2.4.1 Air Sampling Plan Objectives 35
2.4.2 Surface Sampling Plan Objectives 36
2.4.3 Material Sampling Plan Objectives 36
2.4.4 Supporting Information Sampling Plan Objectives 36
2.5 Air Sampling Systems 37
2.5.1 Air Sampling System for Collection of Isocyanates and Flame Retardants 38
2.5.2 Air Sampling System for Quantification of Blowing Agent HFC-134a 39
2.5.3 Air Sampling System for Collection of VOCs with Multi-bed Sorbent Traps and Flame Retardant for TD-
GC/MS Analysis 39
2.5.4 Air Sampling for TCPP and PMDETA with Modified OVS 40
2.5.5 Air Sampling for Aldehydes and Ketones 41
2.5.6 Air Sampling to Characterize Particle Size and Number 41
2.6 Surface Sampling Systems for TCPP 42
2.6.1 Deposition Samplers 42
2.6.2 Wipe Samples 44
2.7 HVAC Filter Samples to Quantify TCPP Removed from the Chamber Exhaust Flow 45
2.7.1 HVAC Filter Sample Collection and Extraction 46
2.8 Sampling to Quantify TCPP Deposited on PPE 46
2.9 Collection of SPF Insulation Samples for Determination of TCPP Concentrations 47
2.10 Measurement Systems 47
2.10.1 LC-MS/MS Quantification of MDI, p3-MDI, and p4-MDI 47
2.10.2 Photoacoustic Spectrophotometer (PAS) for Quantification of HFC 134a 48
2.10.3 TD-GC/MS Analysis System for Quantification of VOCs, PMDETA, and TCPP 49
2.10.4 GC/MS Analysis System for Analysis of TCPP and PMDETA from Extracts of Sampling Media 50
2.10.5 HPLC/DAD System for Analysis of DNPH Derivatives of Aldehydes and Ketones 50
2.10.6 Particle Analysis Systems 51
2.11 Measurement Systems for Collection of Supporting Information 51
2.11.1 Temperature Measurement Systems 51
2.11.2 Air Velocity Measurement System 52
2.12 Data Analysis 52
2.12.1 Parameters for Calculation of Emission Factors from Chamber Emission Tests 53
2.12.2 Data Evaluation to Identify Sources of Variability and Uncertainty 54
2.12.3 Rationale and Approaches for Addressing Key Questions 55
3.0 QUALITY CONTROL RESULTS FOR MEASUREMENT SYSTEMS 56
SPFI Methods Development Page 5
-------
3.1 Evaluation of Chamber Performance 57
3.1.1 Evaluation of Environmental Control System Performance 57
3.1.2 Agilent 6850 GC/ECD to Determine Chamber Air Exchange Rate 57
3.1.3 Evaluation of Mixing of Gases and Particles 58
3.1.4 Evaluation of Physical Properties of the SPF Insulation 58
3.1.5 Event Timing 59
3.1.6 Pre and Post-test Flow Rate Checks for Sampling Systems 59
3.2 Data Quality Indicator Goals for Analytical Instruments 60
3.3 Instrument Calibrations, lAPs, IDLs, DCCs 61
3.3.1 LC-MS/MS for Quantification of MDI, p3-MDI, and p4-MDI 61
3.3.2 Quantification of HFC 134a with the Photoacoustic Spectrophotometer (PAS) 66
3.3.3 Quantification of VOCs, Amine Catalyst, and TCPP by TD-GC/MS Analysis 68
3.3.4 GC/MS System for Analysis of Extracts from Sampling Media 73
3.3.5 HPLC-DAD Analysis System for Aldehydes 75
3.3.6 APS and ELPI Particle Instrumentation 79
4.0 RESULTS 80
4.1 Characterization of the SPF Insulation Emissions Source 80
4.1.1 Application Time and Rate 80
4.1.2 Amount of Foam Applied 80
4.1.3 Characteristics of the SPF Insulation 80
4.1.4 Temperature of the Foam During Curing 81
4.1.5 Amount and Variability of TCPP Concentrations in the SPF Insulation 81
4.2 Characterization of the Chamber Environmental Conditions 81
4.2.1 Air Temperature and Relative Humidity 81
4.2.2 Air Change Rates 82
4.2.3 Air Speed Near Surface of Sprayed Frames 83
4.2.4 Air Speeds Near Surface of Deposition Samplers 83
4.3 Characterization of Concentrations of Emissions 83
4.3.1 Phase I Concentrations of MDI, p3-MDI and p4-MDI 84
4.3.2 Phase I and II Chamber Air Concentrations of HFC-134a 86
4.3.3 Phase I, II, and III Air, Surface, and Material TCPP Concentrations 88
4.3.4 Phase I and II Concentrations of the Amine Catalyst PMDETA 98
4.3.5 Phase I and II Concentrations of VOCs Determined by TD-GC/MS 101
4.3.6 Phase I and II Concentrations of Aldehydes and Ketones Determined by HPLC-DA 107
4.3.7 Phase I Particle Size and Number Characterization 109
4.4 Discussion of Results Ill
4.4.1 Isocyanate Sampling and Analysis 112
4.4.2 Blowing Agent Sampling and Analysis 114
4.4.3 TCPP and PMDETA Sampling and Analysis 114
4.4.4 VOC Air Sampling and Analysis 116
4.4.5 Aldehyde Air Sampling and Analysis 117
4.4.6 Particle Size and Number Sampling and Analysis 118
SPFI Methods Development Page 6
-------
4.4.7 Surface and Material Sampling 119
4.4.8 Summary of Mass Determined forTCPP in Various Compartments 124
4.5 Summary of the Results 125
4.6 Lessons Learned 126
5.0 EMISSIONS RATES AND EMISSION FACTORS 127
5.1 Calculation of Emission Rates and Emission Factors 127
5.1.1 Phase I Calculation of Isocyanate Emission Factors 129
5.1.2 Emissions Modeling of HFC134a 132
5.1.3 TCPP Emissions Modeling 138
5.1.4 PMDETA Emissions Modeling 144
5.1.5 Emissions Modeling of 3-Chloropropene, 1,2-Dichloropropane, 1,4-Dioxane, 146
and Chlorobenzene 146
5.1.6 Emissions Modeling of Aldehydes and Ketones 155
5.2 Impact of Chamber Air Temperature Emission Factors 156
5.3 Emission Factor Predictions from 72-hour Test Data 156
6.0 SUMMARY AND CONCLUSIONS 157
6.1 Summary of the Experimental Process 157
6.1.1 Summary of the Chamber Environmental Conditions 158
6.1.2 Summary of the Emissions Source 158
6.1.3 Summary of Phase I Air Sampling to Characterize Emissions 159
6.1.4 Summary of Phase II Air Sampling to Characterize Emissions 159
6.1.5 Summary of Phase III Air Sampling to Characterize Emissions 159
6.2 Summary of the Results 159
6.2.1 Summary of Phase I Air Concentrations 159
6.2.2 Summary of Phase II Concentrations 160
6.2.3 Summary of Phase III TCPP Air Concentrations 162
6.2.4 Summary of Surface Concentrations of TCPP Determined with Deposition Samplers 162
6.2.5 Summary of Phase I Mass Emitted 162
6.2.6 Summary of Phase II Mass Emitted 163
6.2.7 Generation of Emission Rates from Chamber Concentration Data 164
6.3 Major Findings 166
6.4 Conclusions 166
7.0 RECOMMENDATIONS 167
8.0 REFERENCES 168
SPFI Methods Development Page 7
-------
ACKNOWLEDGEMENTS 171
APPENDICES 173
Appendix A Emissions Test Chamber Facility 173
A-l Emissions Test Chamber 173
A-2 Modifications to the Chamber for Characterization of SPFI Emissions 175
A-2.1 Air Lock Vestibule 176
A-2.2 Chamber Wall Covering 177
A-2.3 Air Supply for Supplied Air Respirators 177
A-2.4 Air Mixing Fans 178
A-2.5 System for Measurement of Chamber Air Change Rate 180
Appendix B 181
B-l Air Sampling Systems 181
B-l.l Isocyanate and Gas-Particle Phase TCPP Sampling System 181
184
B-l.2 Air Sampling System for High Concentrations of Blowing Agent 184
B-2 Deposition Samplers 186
APPENDIX C ANALYTICAL SYSTEMS 188
C-l. Thermal Desorption System for Analysis of VOCs Collected on Multi-bed Sorbent Traps 188
C-2 GC MS System for Analysis of Liquid Extracts from Various Media 190
C-3 Operating Parameters for the LC-MSMS 192
C-3.1 LC-MS/MS Analysis for Quantification of MDI, p3-MDI, and p4-MDI 192
APPENDIX D QUALITY ASSURANCE AND CONTROL RESULTS 194
D-l Evaluation of Physical Properties of the SPFI 194
APPENDIX E MODEL FILES 196
E-l Model Files for HFC-134a 196
E-l.l Model File for HFC-134a 197
E-l.2 Model Output File for HFC-134a 199
E-l.3 Statistics Report for HFC-134a Model fit 200
E-l.4 HFC-134a Model Files for IECCU 202
E-2 Model Files for 1,4-dioxane 205
E-2.1 Micromath Scientist Model File for 1,4-Dioxane 205
E-2.2 1,4-Dioxane Data Input File 207
E-2.3 1,4-Dioxane Plot 208
E-2.4 1,4-dioxane files for IECCU 208
SPFI Methods Development Page 8
-------
E-3 Model files for PMDETA 210
E-3.1 Micromath Scientist Model File for PMDETA 210
E-3.2 Data Input File 213
E-3.3 Plot of Model Output for PMDETA 213
SPFI Methods Development
Page 9
-------
List of Tables
Table 1-1 Summary of Measurements 30
Table 2-1 Target isocyanate, blowing agent, volatile and semi-volatile emissions 31
Table 2- 2 Aldehyde and ketone target compound list 32
Table 2- 3 Air sampling and analysis systems 37
Table 2- 4 Material and surface sampling and analysis 37
Table 3-1 Chamber control system performance summary 57
Table 3- 2 Data quality objectives for sampling systems flow controllers 59
Table 3- 3 Summary of sampling systems uncertainty 60
Table 3- 4 Data quality goals and QC checks for analytical instruments 61
Table 3- 5 Minimum detectable and quantifiable levels of isocyanates (|ig m 3) 62
Table 3- 6 Combined pre and post run linear calibrations results 63
Table 3- 7 Relative percent difference duplicate impinger samplers 65
Table 3- 8 Relative percent difference duplicate denuder-filter samplers 65
Table 3- 9 Recovery of isocyanate blind spikes 66
Table 3-10. Zero - span checks for the PAS 67
Table 3-11. Stock solutions for td-gcms standards 68
Table 3-12. Instrument calibration summary for the TD-GCMS system 69
Table 3-13. Minimum detectable levels for the TD-GCMS system 69
Table 3-14. Minimum instrument quantification concentrations for the TD-GCMS system 70
Table 3-15 Daily calibration control sample recovery for the TD-GCMS system 70
Table 3-16 Low-level multi-bed storage samples 71
Table 3-17 Mid-level multi-bed storage sample spikes 71
Table 3-18. Summary results for the spiked multi-bed storage samples 72
Table 3-19 GC/MS calibration range and r2 for the quadratic fit 73
Table 3- 20 Daily calibration check sample recovery for the GC/MS 74
Table 3- 21 Recovery check standard summary results 74
Table 3- 22 Summary of DNPH-aldehyde calibration results 75
Table 3- 23. Instrument detection limits and minimum detection limits for aldehydes 76
Table 3- 24. Daily calibration check summary results for aldehydes 77
Table 3- 25 Chamber/substrate background aldehyde concentrations 78
Table 3- 26. Duplicate pair results for aldehyde samples 78
Table 4-1 Summary comparison of impinger samplers to denuder-filter samplers for samples collected during
the application events 84
Table 4- 2 TCPP concentrations at deposition sampler locations 93
Table 4- 3 Concentrations of TCPP and PMDETA deposited on Tyvek® suits 94
Table 4- 4 TCPP collected on HVAC exhaust duct filters 95
SPFI Methods Development
Page 10
-------
Table 4- 5 TCPP recovered from wipe samples for the mixing fans 96
Table 4- 6 TCPP wipe sample results from chamber surfaces at the end of Phase III 97
Table 4-7 Phase I mass emitted determined from sampling directly from the chamber sampling port 116
Table 4- 8 Phase I mass emitted determined by sampling at the exhaust duct sampling port 116
Table 4- 9 Phase II emissions determined from the chamber concentrations 117
Table 4-10 Phase II emissions determined from exhaust duct concentrations 117
Table 4-11 Concentrations and estimated TCPP mass on surfaces of the chamber 120
Table 4-12 TCPP collected on HVAC exhaust duct filters 123
Table 4-13 Estimate of TCPP and PMDETA deposited on Tyvek® suits 123
Table 5-1 Inputs to IECCU for application phase sum MDI simulation 130
Table 5- 2 Model Input Parameters for HFC-134a Phase I Emissions 134
Table 5- 3 IECCU Phase I TCPP simulation inputs 139
Table 5- 4 Model parameters for fitting the dual first order equation to the application phase emissions 145
Table 5- 5 Model parameters for fitting the dual first order equation to the Phase I emissions 147
Table 5- 6. Comparison of average Phase II emission factors determined from chamber and exhaust duct port
sampling locations 154
Table 5- 7 Phase I Emission rates for selected aldehydes and ketones 155
Table 5- 8 Estimated impact of 3°C temperature rise on emission factors for selected compounds 156
Table 5- 9 Comparison of predicted emission factors calculated from power fit of Phase II emission factors
through 72 h to emission factors determined from the experimental data at those time periods 157
Table 6-1 Phase I air concentration summary 160
Table 6- 2 Summary of Phase I mass emitted for target compounds 163
Table 6- 3 Summary of mass emitted during phase I for TCPP and PMdeta 163
Table 6- 4 Summary of mass emitted during Phase II 163
Table C-1 markestd-100 thermal desorption sample introduction system 188
Table C- 2 Operating parameters for the agilent gems 189
Table C-3 Sources of standards for calibration of the TD-gc/ms system 190
Table C- 4 Hewlett-Packard 6890/5973 GC/MS system with a 6890 Series injector 191
Table C- 5 API 3200 LC/MS/MS System Operating Parameters for analysis of isocyanate-dibutylamine derivatives
193
Table C-l Chromatographic conditions and parameters 193
Table C- 63 Source/Gas Parameters 193
Table C- 7 Analyte specific parameters 194
Table D-l: Physical Properties of Foam Testing Method and Sampling Information 195
SPFI Methods Development
Page 11
-------
List of Figures
Figure 2-1 Plywood Substrates in foil-lined chamber 34
Figure 2-2 Deposition sampler array at the base of a sampling tree 42
Figure 2-3 Locations of arrays of deposition samplers 44
Figure 2-4 Wipe sampling template mounted in glass exhaust duct 45
Figure 2-5 HVAC filter placed over outlet duct opening 46
Figure 3-1 Mid-Level Linear PAS Calibration 66
Figure 3- 2 Low-Level Linear PAS Calibration 66
Figure 3-3 Extended Calibration of the PAS 67
Figure 4-1 Air temperatures at four elevations near the sprayed frames 82
Figure 4- 2 Sum of MDI, p3-MDI, and p4-MDI determined by sampling with impingers 85
Figure 4- 3 Sum MDI, p3-MDI, and p4-MDI determined by sampling with denuder-filter samplers 85
Figure 4- 4. HFC-134a Concentration-time profile during and following spray application* 87
Figure 4- 5: HFC134a Phase II concentrations 88
Figure 4- 6. TCPP chamber and exhaust duct air concentrations during Phase I determined by TD-GC/MS 89
Figure 4- 7 Phase I TCPP concentrations and % particle in the chamber determined with modified OVS 90
Figure 4- 8 Phase II TCPP Concentrations in the chamber and exhaust duct determined by TD-GC/MS 91
Figure 4- 9. Phase III TCPP concentrations in the chamber and exhaust duct air 92
Figure 4-10 Overshoot nodules collected from the floor of the chamber 98
Figure 4-11 Phase I Concentrations of PMDETA determined by sampling with multi-bed sorbent traps and TD-
GCMS analysis 99
Figure 4-12 Phase I PMDETA concentrations determined with multi-bed sorbent traps and modified OVS 100
Figure 4-13 Phase II PMDETA chamber and exhaust duct concentrations determined by TD-GCMS 101
Figure 4-14. Phase I concentrations of 1,2-dichloropropane 102
Figure 4-15. Phase I concentrations of 1, 4-dioxane 103
Figure 4-16. Phase I concentrations of chlorobenzene 103
Figure 4-17 Phase II concentrations of 3-chloropropene 105
Figure 4-18 Phase II concentrations of 1,2-dichloropropane 105
Figure 4-19. Phase II1,4-dioxane concentrations 106
Figure 4- 20. Phase II concentrations of chlorobenzene 106
Figure 4- 21. Phase I aldehyde concentrations 107
Figure 4- 22. Phase II aldehyde concentrations 108
Figure 4- 23. Phase II formaldehyde and hexaldehyde concentrations 109
Figure 4- 24. Particle concentrations during the first 1.5 h of Phase 1 110
Figure 4- 25. Phase I cumulative particle emissions 119
Figure 4- 26. Correlation between air speed 1 cm above the surface of the deposition samplers and log TCPP
concentration determined at the end of Phase 1 121
Figure 5-1. IECCU simulations of sum MDI Phase I concentrations 131
SPFI Methods Development Page 12
-------
Figure 5-2 IECCU simulation of sum MDI concentrations in the full-scale chamber during Phase I based upon
emission rated calculated directly from chamber concentrations 132
Figure 5-3. Observed and predicted Phase I concentrations for HFC-134a assuming Mi = 95% of Mo 134
Figure 5-4. Phase II emission factors showing impact of time and of the chamber air temperature excursion on
HFC-34a emission factor 136
Figure 5-5 Impact of chamber air temperature on HFC-134a emission factor 138
Figure 5-6 Phase I IECCU simulated TCPP chamber concentrations shown with chamber and exhaust duct air
concentrations 139
Figure 5-7 Power law Fit of Phase II TCPP emission factors determined with the direct calculation method 141
Figure 5-8 IECCU model output overlaid on chamber concentration data. Diffusion and partition coefficients
estimated by Bevington (2017) and estimated to fit this data set 143
Figure 5-9 Phase I predicted and observed chamber concentrations for PMDETA assuming 10% or 15% of
application phase emissions (Mo) are due to slow decay (k2) 145
Figure 5-10 Power fit of PMDETA emission factors determined by direct calculation from chamber concentration
data 146
Figure 5-11 Phase I IECCU simulation of 3-chloropropene concentrations assuming 70% of the emissions were
available for "rapid" emission (Mi = 70% of Mo) 149
Figure 5-12. Phase I IECCU simulation of 1,2-dichloropropane concentrations assuming 80% of the emissions
were available for "rapid" emission (Mi = 80% of Mo) 149
Figure 5-13 IECCU simulation of Phase 11,4-dioxane concentrations assuming 95% of the emissions were
available for "rapid" emission (Mi = 95% of Mo) 150
Figure 5-14 IECCU simulation of Phase I chlorobenzene concentrations 151
Figure 5-15 IECCU Chlorobenzene time - concentration simulation from an emission rate text file 151
Figure 5-16 Phase II emission factors calculated for 3-chloropropene 152
Figure 5-17 Phase II emission factors calculated for 1,2-dichloropropane 153
Figure 5-18 Phase II emission factors calculated for 1,4-dioxane 153
Figure 5-19 Phase II emission factors calculated for chlorobenzene 154
Figure A-1 Schematic of test chamber and clean air conditioning and delivery system 174
Figure A- 2 Large chamber control computer screen shot of air flow control 175
Figure A- 3 Air lock vestibule 176
Figure A- 4 Air supply lines for supplied air respirators 178
Figure A- 5 Mixing fan placement and directions 179
Figure A- 6 Location of air speed measurements on each frame 180
Figure A- 7 GC ECD with gas sampling valve 181
Figure B-1 Vacuum pump, gage, shutoff valve, outside of chamber with line into chamber 182
Figure B- 2 View of 4-port manifolds and sampling lines with critical a critical orifice in each line 182
Figure B- 3 Transport container for isocyanate samplers 183
Figure B- 4 Impingers and denuder-filter samplers positioned on a sampling stand 183
Figure B- 5 Modified OVS with PTFE filter placed ahead of the OVS glass fiber filter and XAD resin beds 183
Figure B- 6 OVS with end caps and PTFE filter material 184
SPFI Methods Development Page 13
-------
Figure B- 7 Air-tight box for collection of whole air samples in tedlar bags 185
Figure B- 8 Jig for cutting deposition samplers from aluminum foil 186
Figure B- 9 deposition sampler array at the base of a sampling tree 186
Figure B-10 Cutout jig to collect samples from the HVAC filter 187
Figure B-11 MERV 13 HVAC filter section prior to mounting over chamber outlet duct opening 187
Figure C-1 Markes tdlOO thermal desorption system and agilent Gems 188
Figure C- 2 GCMS SYSTEM FOR ANALYSIS OF EXTRACTS OF MEDIA 191
Figure C- 3 LCMSMS used for identification and quantification of isocyanates 192
Figure E-l. Scientist plot of fit of dual first order decay model to hrc-134a time - concentration data 200
Figure E-2 IECCU model file dual first order decay model 203
Figure E-3. Screen shot of IECCU application phase simulation parameter inputs for the dual first order decay
model 204
Figure E-4. Scientist plot of fit of dual first order decay model to 1,4-dioxane time - concentration data 208
Figure E-5 IECCU input page for 1,4-dioxane 209
Figure E-6 IECCU model parameter entry page for the dual first order decay model 210
Figure E-7. Scientist plot of fit of dual first order decay model to PMDETA time - concentration data 214
SPFI Methods Development
Page 14
-------
Acronyms and Abbreviations
ACC
American Chemistry Council
ACH
Air changes per hour
AEMD
Air and Energy Management Division
AMCD
Air Methods & Characterization Division
APS™
Aerodynamic Particle Sizer Spectrometer
ASTM
American Society for Testing and Materials
Denuder-filter
ASSET™ EZ4-NCO dry sampler for isocyanates
CAS
Chemical Abstract Service
CAA
Clean Air Act
CEMM
Center for Environmental Measurement and Modeling
CFM
Cubic feet per minute
CNS
Central nervous system
CPI
Center for Polyurethanes Industry
CPSC
Consumer Products Safety Commission
DBA
di-n-butylamine
DCC
Daily calibration check
DIY
Do it yourself
DNPH
2,4-Dinitrophenylhydrazine
DQI
Data quality indicator
DSBB
Distributed Source and Buildings Branch
ECD
Electron capture detector
ELPI®
Electrical low-pressure impactor
EPA
Environmental Protection Agency
FB
Field blank
FSETC
Full-Scale Emissions Test Chamber
GC/MS
Gas chromatography/mass spectrometry
GFF
Glass fiber filter
GSA
General Services Administration
GWP
Global warming potential
HDPE
High density polyethylene
HEPA
High efficiency particulate air
HFC
Hydrofluorocarbon
HFC-134a
1,1,1,2-Tetrafluoroethane
HFC-245fa
1,1,1,3,3 -Pentafluoropropane
HPLC
High performance liquid chromatography
HVAC
Heating ventilation and air conditioning
IAP
Independent audit program
IDL
Instrument Detection Level
ffiCCU
Indoor Environmental Concentrations in Buildings with Conditioned
and Unconditioned Zones
IS
Internal standard
LC-MS/MS
Liquid chromatography-mass spectrometry/mass spectrometry
SPFI Methods Development
Page 15
-------
LOQ
Limit of quantitation
MDI
Methylene diphenyl diisocyanate
MDL
Method detection limit
NIOSH
National Institute of Occupational Safety and Health
N/L
Air change to loading ratio
NRMRL
National Risk Management Research Laboratory
NTP
National Toxicology Program
ODS
Ozone depleting substance
OPPT
Office of Pollution Prevention and Toxics
ORD
Office of Research and Development
OSHA
Occupational Safety and Health Administration
OVS
OSHA Versatile Sampler
PAS
Photoacoustic spectrophotometer
PE
Polyethylene
PES
Performance evaluation sample
PMDETA
Pentam ethyl di ethyl enetri amine
PMDI
Polymeric MDI
PPE
Personal protective equipment
PTFE
Polytetrafluoroethylene
PUF
Polyurethane Foam
RCS
Recovery check standard
RH
Relative Humidity
RPD
Relative percent difference
RSD
Relative standard deviation
SDS
Safety Data Sheet
sf6
Sulfur hexafluoride
SFSB
Source and Fine Scale Branch
SNAP
Significant New Alternatives Policy
SPF
Spray polyurethane foam
SPFI
Spray polyurethane foam insulation
ss
Stainless steel
STV
Sample-time-volume
SVOC
Semivolatile organic compound
TCPP
Tris(l-chloro-2-propyl) phosphate
TD-GC/MS
Thermal desorption gas chromatography/mass spectrometry
QA
Quality assurance
QAPP
Quality assurance project plan
QC
Quality control
voc
Volatile organic compound
SPFI Methods Development
Page 16
-------
Executive Summary
A full-scale protocol was developed and demonstrated for characterization of gaseous and
particulate emissions from the application of spray polyurethane foam (SPF) insulation. Results
demonstrated the feasibility of conducting an integrated test that quantifies the emissions during
application as well as the longer-term emissions from cured foam insulation. Results will inform
the development of a consensus full-scale test protocol for use by stakeholders and provide
valuable data for calibration of an exposure model designed to predict emissions concentrations
in buildings with conditioned and unconditioned spaces.
SPF insulation is used extensively to insulate and seal buildings to improve energy performance.
Many of the chemicals used in production of SPF insulation have known health risks and
exposures to the complex mixture of emissions during and following application, are not well
characterized. The U.S. Environmental Protection Agency (EPA) needs reliable consensus
emissions test protocols and data to inform risk assessors and managers, chemical manufacturers,
product formulators, test laboratories, building managers, occupants, and consumers. The spray
foam industry needs a reliable consensus test method that can inform decisions regarding the
amount of time and ventilation needed for safe re-entry of workers and re-occupancy by
residents following application of SPF insulation.
To this end, a full-scale protocol was developed to characterize the gaseous and particulate
emissions from the application of SPF insulation. SPF insulation was sprayed onto 7 m2 of
plywood substrates inside a 30 m3 test chamber supplied with clean, conditioned air flowing
through the chamber. Air concentrations of the blowing agent, volatile organic compounds, an
amine catalyst, the flame retardant, isocyanates, aldehydes, ketones, and aerosols were quantified
during the application and initial curing phase of 3.3 hours (Phase I), and for four weeks
following the application (Phase II). The sprayed plywood substrates were then removed from
the chamber and air concentrations due to emissions of the flame retardant tris (l-chloro-2-
propyl) phosphate (TCPP) from chamber surfaces were monitored for twelve days (Phase III).
The amount of TCPP on chamber surfaces and on a heating, ventilation and air conditioning
filter covering the opening from the chamber to the exhaust duct were determined at the end of
each phase.
Source emission models were constructed for selected compounds for the initial application and
curing phase and for the longer-term emissions from the concentrations of emissions determined
in the air of the chamber. The results demonstrate that the full-scale emissions characterization
approach produces data that is suitable for construction of empirical source emissions models,
which may be used as primary inputs to building simulation models, key tools for risk
characterization and management. Test results quantify the mass of SPF insulation emissions per
unit of area, volume, or mass of SPF insulation applied during and following application.
Correlation coefficients (r2) between observed and model-predicted application phase
concentrations greater than 0.97 demonstrated the feasibility of this approach.
SPFI Methods Development
Page 17
-------
The TCPP chamber test data were also compared to the concentration predictions of a physical
parameter-based model available in the exposure simulation model Indoor Environmental
Concentrations in Buildings with Conditioned and Unconditioned Zones [IECCU]) developed
for EPA's Office of Pollution Prevention and Toxics (OPPT) that addresses emissions from
sources applied to conditioned and unconditioned spaces in buildings. Results demonstrated the
promise of the mass transfer modeling approach which utilizes the physical properties of
chemicals and materials and does not require full-scale emissions test data. However, the
observation that the average flame retardant emission rate from the chamber over a 290-hour
period following removal of the sprayed frames was 21% of the average emission rate over the
nearly 670-hour test period with the sprayed frames in the chamber demonstrated the need for
better understanding of emissions from flame retardant deposited on chamber surfaces.
Results demonstrated that emissions of gases and particles, including isocyanates, flame
retardants, amine catalysts, and air toxics such as 1,4-dioxane, and blowing agents can be orders
of magnitude higher during the application than during the long-term cured product phases.
Results demonstrated that emissions of some compounds such as flame retardants and amine
catalysts may persist for long periods of time. TCPP was deposited on all surfaces in the
chamber during application. Concentrations on the floor were about ten times those observed on
walls and ceiling. Concentrations decreased with time on all surfaces. The minimum efficiency
reporting value (MERV 13) HVAC filter over the exhaust duct opening was about 33% efficient
in removing the flame retardant TCPP from the air entering the exhaust duct during the
application/curing phase and less than 5% efficient during the post-application phase and
following removal of the sprayed frames from the chamber. The results also demonstrated a
greater than 50% increase in emission rate of flame retardant from cured foam and chamber
surfaces with a 3 °C increase in chamber air temperature.
Taken as a whole, the results demonstrate the feasibility of the approach to emissions
characterization from application and curing phases. The emissions data generated are suitable
for construction of source emissions models that may be used in building simulations that
evaluate control strategies that minimize transport of emissions to occupied spaces and to the
outdoors. Results demonstrate that gas phase emissions of TCPP persist for long periods of time
and air concentrations may not be impacted by common heating and air conditioning filters.
Stakeholders should continue development of a consensus test method that includes application
and post application emissions characterization, develop strategies to reduce emissions of flame
retardant and amine catalyst from spray foam products, and strategies to remove flame retardant
from indoor air. The results also demonstrate the potential of diffusion-based models to predict
emissions from spray foam products.
Limitations of the study
There are several limitations to the study, including: (1) there has been no replication of the
study, so though results inform further development of a consensus test protocol, precision of the
overall protocol cannot be determined from this experiment, (2) after January 1, 2021, the
blowing agent HFC-134a in the product used for method development can only be used in
SPFI Methods Development
Page 18
-------
limited aerospace and military applications and the approved hydrofluoroolefin (HFO)
replacement blowing agents may have very different emissions characteristics and require
different measurement techniques, (3) the study did not investigate the impact of varying the air
change rate to loading ratio on emission rates, thus, it is not known if emission rates were
impacted by concentrations in the air of the chamber, and (4) the study utilized a low-pressure
two-component SPF insulation kit whereas the draft consensus test protocol addresses emissions
characterization of low pressure kits and commercial scale high pressure SPF insulation systems
that are applied at a much higher pressure and application rate.
SPFI Methods Development
Page 19
-------
1.0 Introduction
Emissions from materials and products that impact concentrations of organic compounds in
indoor air has been an active area of research for more than 30 years [1], Research has focused
upon developing methods to identify emissions, characterizing concentrations in buildings, and
developing models that predict potential exposures from source emissions as the basis for
development of sound risk assessment and risk management practices [2], [3], [4], Typically,
emissions characterization isolates a product in the controlled environment of a test chamber
supplied with clean, conditioned air. Concentrations of emissions from the product are measured
in the air of the chamber and this information is utilized to develop source emissions models.
Spray polyurethane foam (SPF) insulation is a widely used product that presents challenges to
conventional source characterization methods. SPF insulation is manufactured on-site by spray
application of reactive chemicals in new and retrofit buildings. Some of chemicals in
formulations have known health hazards and risks and others are unknown or may be listed as
proprietary. When applied to building surfaces, the exothermic chemical reactions between
isocyanates and polyols form a rigid polymer with cellular structure. High ventilation rates
generated with portable exhaust fans are typically used to move aerosol and gas-phase emissions
from the area of application in the building environment to the outdoors during and for some
period-of-time after application. Following application, the effectiveness of SPF insulation in
sealing the building shell reduces natural air change due to building leakage. Therefore, a test
method that characterizes SPF insulation emissions must address the high and rapidly changing
emissions during the application and curing process as well as the longer-term emissions from
the fully cured product.
1.1 Nature of the Problem
SPF insulation is used extensively to insulate buildings. According to the American Chemistry
Council (ACC) Center for the Polyurethanes Industry (CPI) 2012 End-Use Market Survey,
approximately 54.4 million kilograms (kg) of two-component SPF insulation was used in U.S.
residential applications [5], Of that amount, approximately 36 million kg were open cell and
approximately 18 million kg were closed cell insulation. Many of the chemicals used in the
residential and commercial production of SPF insulation have known health risks and exposures
to the complex mixture of aerosol and gas-phase emissions during and following application are
not well characterized. Standardized consensus emissions test protocols, data, and models are
needed to inform risk assessors and managers, chemical manufacturers, product formulators, test
laboratories, building managers, occupants, and consumers.
1.2 Background
The U.S. Environmental Protection Agency (EPA) and Consumer Products Safety Commission
(CPSC) have received complaints from consumers that report odor or health concerns following
installation of SPF insulation. There have been infrequent reports that homeowners have become
sensitized to SPF insulation following application and were no longer able to enter their homes
[6], In addition, development of asthma has been linked to homeowners who reentered their
home four hours after application of SPF insulation [7],
SPFI Methods Development
Page 20
-------
1.2.1 Exposure and Health Concerns
The association between occupational exposures to some of the chemicals used in production of
polyurethane products and health risks have been known for some time. Isocyanate exposure is
the leading known attributable cause of workplace-related asthma. Krone for example, estimates
that 5 -13% of workplace-related asthma is attributable to isocyanate exposure [8], Methylene
diphenyl diisocyanate (MDI) irritates the eyes and irritates and sensitizes the respiratory tract and
skin [9] [10], MDI skin exposure has been shown to induce MDI-specific immune sensitivity and
promote respiratory tract inflammation responses in animal systems. Amine catalysts may also
cause respiratory tract, eye, and skin irritation and sensitization. Blowing agents may also irritate
the respiratory tract, eyes, and skin, and may have other central nervous system (CNS) effects.
See Sleasman et.al., for discussion of emissions, analytical approaches, regulatory exposure
limits, and health effects [11],
Concern over exposure to flame retardants used in polyurethane products has spurred research to
better understand sources, routes of exposures, and human and ecological impacts [12] [13] [14],
The National Toxicology Program (NTP) conducted a developmental toxicity study for tris (1-
chloro-2 propyl) phosphate (TCPP), which is the predominant flame-retardant compound used in
current formulations employed in SPF insulation products.
1.2.2 EPA's SNAP Program and SPF Insulation
Many propellants and blowing agents used in production of SPFs are listed for replacement
under the EPA's Significant New Alternatives Policy (SNAP) program, an implementation of
the amended Clean Air Act (CAA) of 1990, Section 612, to evaluate ozone depleting substance
(ODS) replacement compounds. The evaluation criteria encompass health, safety, and
environmental concerns, with a special emphasis on the global warming potential (GWP) which
is a measure of climate impact potential. Two blowing agents, HFC-134a and HFC-245fa,
associated with the research reported here are Class II ODS compounds that have been recently
phased-out except for limited military and space uses. The total ban for these refrigerant gases
takes effect as of January 1, 2025. See www.epa.gov/snap for program details [151.
1.2.3 Research Needs
Due to the increasing use of SPF insulation products, and reports of adverse health effects from
some homeowners following installation of SPF insulation, and uncertainties regarding potential
for longer-term emissions of semivolatile organic compounds (SVOCs) such as flame retardants,
there is a need for an improved understanding of the multi-pollutant emissions and potential
exposures from SPF during application, curing, and post-curing. In addition, there is a need for
improved understanding of the effectiveness of ventilation in removal of emissions during
application and curing, the impact of environmental variables on emissions and cure time, and
movement of emissions from application site to inhabited spaces over the lifecycle of the
material.
1.3 Objectives of the Study
This report summarizes results of research conducted by the EPA to develop a full-scale
emissions test method that characterizes the application and post-application emissions from SPF
SPFI Methods Development
Page 21
-------
insulation. The primary objectives of the research reported here are to support development of a
consensus full-scale test method for use by stakeholders to improve risk characterization and
management for SPF insulation and provide emissions test data generated in a well-controlled
environment for calibration of an exposure modeling tool developed for EPA's Office of
Pollution Prevention and Toxics (OPPT). The SPF insulation industry and other stakeholders
desire a consensus tool for comparing emissions due to formulation changes and a test protocol
that provides reliable data to inform re-entry and re-occupancy of buildings following application
of SFP insulation.
1.4. Types of SPF Insulation
SPF insulation is produced on-site by spray application of Side A and Side B reagent mixtures
using high or low-pressure systems. The Side A and Side B reagents mix in the nozzle of a spray
gun. Side A chemicals consist of isocyanates, and Side B chemicals consist of poly alcohols,
flame retardants, amine catalysts, surfactants, and blowing agents. The isocyanates, MDI and
poly MDI (3, 4, and 5 ring or higher MDI) react exothermically with polyols to form the cellular
structure of the SPF insulation. Amine and metal catalysts, propellants, blowing agents,
surfactants, flame retardants, and other compounds impact the speed of reaction and determine
the ratio of open and closed cells and the density of the insulating material. Open cell foam is
less dense than closed cell foam which seals and retards air movement as well as providing
insulation. The formation of the cellular structure occurs rapidly, often within a matter of
seconds, as the foam rises, and the cells expand [16], Depending upon the formulation, SPF
insulation is generally considered "tack-free" within a few minutes.
The reaction of isocyanates and polyols generates heat and the temperature of the foam increases
as the reagents polymerize to form the polyurethane structure. The low molecular weight
blowing agents and air are trapped within the cells of the foam and cause it to rise. Surfactants,
catalysts, and the heat generated by the polymerization processes impact the rate of reactions and
physical characteristics of the foam. The temperature of the foam increases rapidly as the foam
rises and forms and then returns to ambient temperature over a longer time frame. The time scale
for return to ambient temperature will depend upon many factors, including the formulation,
depth of application, characteristics of the substrate, and other environmental conditions
Low density or open cell foams form an excellent insulation material, air, and sound barrier.
Medium density SPF insulation or closed cell foam retards transport of water vapor. Low density
foams typically weigh about 8 kilograms per cubic meter (kg m"3), and most cells are open and
filled with air, whereas medium density foam weighs about 32 kg m"3 and most cells are closed
and initially filled with blowing agent. Low density SPF insulation has a thermal resistance R
value (°F ft2 sec/BTU) of about 4.3. Medium density SPF insulation retards transport of water
vapor and has a higher R value of about 6. Spray application enables effective application to
building surfaces that may be otherwise difficult to effectively insulate and seal and may also
strengthen the integrity of the building structure and improve stability and wind resistance of the
building shell.
SPFI Methods Development
Page 22
-------
High pressure SPF spray systems require sophisticated and expensive equipment that is used
exclusively by professionals whereas low pressure systems are used by professionals and by do-
it-yourself (DIY) homeowners. Reagent containers for two-component systems range in size
from 55-gallon drums used by commercial applicators to, e.g., one to five-gallon containers for
use in low pressure systems employed by professionals and available for DIY. One-component
systems used for sealing air pathways around plumbing fixtures or other small openings are
widely available in hardware and building product stores and typically contain 340 to 680 grams
of foam reagents in aerosol spray cans.
1.5 Consensus Method Development Process
The EPA (OPPT) formed an inter-governmental workgroup in 2009 to address concerns about
exposure to chemicals emitted from polyurethane products. The workgroup included staff from
EPA, National Institute of Occupational Safety and Health (NIOSH), Occupational Safety and
Health Administration (OSHA), the National Institute of Standards and Technology (NIST) and
CPSC. Representatives from EPA, NIST, and CPSC have worked with stakeholders on the
American Society for Testing and Materials (ASTM) Indoor Air Subcommittee D22.05 towards
development of consensus test methods and models to provide tools for characterizing and
managing emissions from SPF insulation.
This effort led to a 2015, ASTM D22.05 sponsored symposium titled "Developing Consensus
Standards for Measuring Emissions from Spray Polyurethane Foam (SPF) Insulation" [17], The
symposium brought together industry, government, testing laboratories, and other stakeholders to
present and discuss recent research and identify the most pressing research needs to move the
standards development path forward. Research needs identified by the symposium participants
that are addressed by this research include:
• Better characterize emission factors during the first 24 hours following application of
SPF and thereafter, develop full scale test methods to provide data to calibrate source
emission models and to improve ventilation guidance for a range of scenarios.
• Characterize interactions of flame retardants with surfaces.
• Improve sampling and analysis of reactive amines compounds that behave poorly in
thermal desorption systems.
• Improve mass transfer parameter estimation techniques and conduct experiments to
validate parameters and fate and transport mass transfer modeling results.
• Improve quality assurance procedures to ensure characterization of uncertainty of a test
protocol.
1.5.1 Development of Draft ASTM Full-Scale Test Protocol
Following the 2015 symposium, a draft ASTM test protocol was introduced in ASTM
Subcommittee on Indoor Air, D22.05, for characterization of post-application emissions in a
SPFI Methods Development
Page 23
-------
laminar flow spray booth. The draft protocol was based upon an SPF industry occupational
health approach to evaluation of exposure to product emissions [18], The approach was intended
to inform safe times for re-entry and occupancy of buildings following application of SPF
insulation by determining concentrations of emissions in the spray booth in what was termed as a
"worst-case" scenario. In this scenario, SPF insulation was sprayed on to approximately 11m2
cardboard substrates in a spray booth with a volume of 15 m3. The chamber ventilation rate was
150 m3 h"1 during the application. Concentrations of emissions in the spray booth were
monitored over a two-hour period starting one hour after completion of the spray event. The
ventilation rate was then decreased to 15 m3 h"1 and air samples were periodically collected until
24 h after the start of the test. The chamber concentration data were then evaluated by a
toxicologist for recommendations regarding safe times for re-entry of unprotected trade workers
and re-occupancy of the building.
1.5.1.1 Potential Shortcomings of the Draft Protocol
The initial draft protocol did not attempt to characterize the concentrations until an hour after the
spray event and it did not address other variables that are needed to utilize the chamber
concentration data to construct source emissions factors for use in exposure models. Also, the
low-flow test conditions were governed by air flow rate limitations of the test chamber. Actual
post-application air exchanges may be much lower than one per hour. For example, Huntsman
states that air changes per hour (ACH) rates of 0.1 to 0.2 h"1 should be considered for buildings
insulated with SPF when considering proper design of HVAC systems for spray foam homes.
See Proper Design of HVAC Systems for Spray Foam Homes , last accessed 8/14/2019).
1.5.1.2 CONSIDERATIONS FOR CHARACTERIZING SPF INSULATION EMISSIONS DURING APPLICATION
Utilizing concentrations of emissions in the air of a chamber or concentrations in the outlet air to
calculate mass emitted during the spray event assumes that the concentrations determined by
sampling are representative of the average concentrations in the air of the chamber. For a source
in a chamber, there is a time required to achieve well-mixed conditions which depends upon
characteristics of the emissions source and the other factors that impact mixing within the
chamber. With two mixing fans operating in EPA's chamber, a mixing level >85% occurs within
about 3 minutes. However, high pressure spray application may induce air currents and the heat
released by the exothermic reactions that form the SPF insulation may result in stratification.
Also, the position of the spraying relative to the chamber air outlet may have an impact on the
degree of mixing during the application process.
The high rate of application further complicates collection of representative samples. The
application rate for commercial high-pressure low density (open cell) systems is reported to be
>2 m2 min"1, thus application of high-pressure low density SPF insulation to 11 m2 substrate in a
15 m3 chamber typically takes from 3 to 5 minutes, or 0.05 to 0.08 h. Sampling during
application is further complicated by the high and rapidly changing concentrations of aerosols
and gases. Short duration and variable volume sampling may be needed to prevent overloading
of sampling media or analytical systems. However, the shorter the sampling time, the more
challenging it becomes to collect samples at multiple locations with the same start and stop
SPFI Methods Development
Page 24
-------
times. These challenges can be diminished by increasing the size of the chamber so that the time
of application is much greater than the time for mixing. Consider the following:
For a constant emission source in a well-mixed chamber supplied with clean air, the
concentrations of emissions can be calculated from Equation 1.
C(t) = (E/Q) * (l-exp(-k*t)) Equation (1)
Where:
C(t) = concentration (|ig m"3) at elapsed time t (h)
E = emission rate of the source (|ig h"1)
Q = air flow rate through the chamber (m3 h"1)
k = chamber air change rate (h"1)
t = elapsed time from start of source (h)
At an air change rate of 10 h"1, and a constant emission rate, the concentration of an emission
reaches 95% of the theoretical maximum concentration after 0.32 h. A chamber of approximately
50 m3 volume would provide the wall surface area needed to spray 45 m2 of SPF insulation. That
would provide time to collect multiple samples during the application and evaluate mixing with a
tracer compound. So, scaling up the size of the chamber and area sprayed would diminish some
of the potential challenges with characterizing application phase SPF insulation emissions. Also,
the application could be divided into a series of application events with time for sampling
between each application event.
In this experiment utilizing a low-pressure kit, the application was spilt into two spray events
separated by sample a collection event. This strategy provided a longer time for the application
and allowed collection of several samples to characterize the emissions.
1.5.2 Need for an Integrated Consensus Full-Scale Emissions Test Protocol
A modeling tool was developed for EPA's OPPT titled, IECCU. a Simulation Program for
Estimating Chemical Emissions from Sources and Related Changes to Indoor Environmental
Concentrations in Buildings with Conditioned and Unconditioned Zones (last accessed
8/14/2019) [19], The model accepts source terms that include application phase emissions factors
and empirical or mass transfer-based longer-term emissions source models. The model addresses
interactions between SVOC emissions and particles and surfaces as well as diffusion-controlled
emissions. There are few data available to calibrate the model for SPF insulation exposure
scenarios. There is a need for an integrated full-scale test protocol and for data that characterizes
the application and post-application phase emissions and provides insight into the impact of
overshoot, deposition, and re-emission of flame retardant from surfaces on indoor air
concentrations.
SPFI Methods Development
Page 25
-------
1.5.3 EPA Support to Development of a Consensus Full-Scale Emissions Test
Protocol
To inform the ASTM full-scale test method development process and address research needs
previously described, EPA's Center for Environmental Measurement and Modeling (CEMM),
Air Methods & Characterization Division (AMCD), Source & Fine Scale Branch (SFSB),
{formerly NRMRL/AEMD Distributed Source and Buildings Branch (DSBB)} conducted
experiments designed to characterize SPF insulation emissions during and following application
in an integrated test protocol. The objective was to develop and demonstrate methods and data
for a consensus test method based upon ASTM D6670, Standard Practice for Full-Scale
Chamber Determination of Volatile Organic Emissions fi'om Indoor Materials Products, that
informs potential exposures to workers and occupants, informs re-entry and re-occupancy of
buildings and informs ventilation guidelines [20],
1.5.3.1 Application-Phase Pilot Tests
Two pilot experiments were conducted prior to the experiment reported here. The objectives of
the pilot tests were to inform scaling of the experiment and provide insight into the magnitude
and duration of application and curing phase emissions. The pilot experiments were also
conducted to evaluate experimental procedures. Method feasibility questions addressed by the
pilot experiments included: (1) ability of the test chamber air handling system to respond to the
heat load generated by the curing foam and two people working in the chamber, (2) ability of the
chamber air control system response to the air released from the supplied air respirators of the
personnel working in the chamber, (3) ability of personnel working inside the chamber to
manage spray application of the foam and collection of isocyanate and flame retardant samplers
during and following application.
1.5.3.2 Considerations for an Integrated Application/Post-Application Emissions Test
The air exchange to loading ratio (N/L) is a parameter that describes the relationship between the
amount of ventilation air that removes emissions from an indoor environment with respect to the
amount of a source material in the environment. Air change rate (N) is the number of times in
one hour that room air is replaced with fresh air. The loading ratio (L) is the ratio of the surface
area of a source (S) to the volume of the room (V). Thus, the N/L ratio is often used to scale
emissions tests [21] [22], The utility of this scaling factor for a method that characterizes spray
application of a source has not been evaluated.
N = chamber air change rate (h"1)
L = S/V (m2 m"3)
and
S = source emissions surface area (m2),
V = chamber volume (m3)
SPFI Methods Development
Page 26
-------
1.5.3.3 Scaling the EPA Test to the Initial Proposed Full-Scale Consensus Test Protocol
The N/L ratio for the test conditions described in the initial draft for the consensus test method
was 15 during the application and 1.5 during the low flow period of the test. The EPA methods
development experiment was scaled to the initial draft method. However, an application area of
20 m2 would be required to achieve an N/L ratio of 15 in EPA's 30 m3 chamber at an air change
rate of 10 h"1. Spraying this much surface area in EPA's chamber would require substrates to be
mounted to all four walls. Due to concerns regarding potential trip hazards posed by supplied air
hoses and the 16 sampling lines running from the manifolds to the sampling support stands
holding impingers containing toluene, the area of application and air change rate were reduced to
minimize potential for spilling impingers in the chamber. Therefore, a target application area of
7.6 m2 at air change rate 3.8 h"1 during application and 0.38 h"1 during long-term testing, were
selected to achieve target N/L ratios of 15 and 1.5 during the application phase and long-term
periods, respectively. Also, based upon the observation in the pilot tests of TCPP emissions from
the chamber after removal of the sprayed substrates, a third phase was added to the protocol to
investigate secondary emissions of TCPP from the chamber surfaces.
The three-phase experimental design was developed to address the research needs identified in
the ASTM symposium that were not addressed in the initial draft consensus method proposal.
The experiment included; (1) a 3.3-hour application and curing phase referred to as Phase I,
followed by (2) a two-week long-term emissions characterization period referred to as Phase II,
and (3) a ten-day decay or re-emission phase to monitor concentrations of TCPP in the air of the
chamber due to emissions from TCPP on chamber surfaces, referred to as Phase III.
1.6 Overview of the Method
Two-component, low-pressure medium density SPF insulation was spray-applied to plywood
substrates in a 30 m3 emissions test chamber supplied with clean air. Concentrations of
isocyanates, blowing agent, flame retardant, VOCs, and aldehydes in the air of the chamber and
air leaving in the chamber exhaust duct were monitored utilizing multiple sampling and analysis
systems. Deposition of TCPP on non-target surfaces of the chamber were estimated from
deposition and wipe samples. Deposition of the flame retardant TCPP on PPE of the personnel
that conducted the spraying and sampling inside of the chamber was determined by extraction of
TCPP from fabric samples. The impact of TCPP deposited on chamber surfaces on airborne
concentrations of flame retardant was investigated by removing the sprayed substrates from the
chamber while continuing to determine concentrations of flame retardant in the air of the
chamber and air in the chamber exhaust duct. The ability of HVAC filters to remove TCPP from
the air leaving the chamber was determined by comparing the mass of TCPP collected on filters
placed over the opening to the exhaust duct with mass leaving the chamber system in the air of
the exhaust duct for each phase of the test.
The temperature of the foam and air temperatures in the chamber were continuously monitored.
ACH was periodically determined using a tracer decay method [23], Mass of target compounds
emitted during each phase was determined using the trapezoid method [24], Emission rates
during application were calculated by; (1) assuming instant and constant emission rate, and (2)
SPFI Methods Development
Page 27
-------
fitting a dual first-order decay model to the concentration data. Longer-term emission rates were
determined by the method of direct calculation from the time-concentration data [23] and for the
flame retardant TCPP, by employing a mass transfer emission model [25],
1.6.1 Phases of the Test Protocol
The test was conducted in three phases. The first phase was designed to characterize emissions of
isocyanates, flame retardant, amine catalyst, VOCs, blowing agent, and particle size and number
during and following spray application at the target air change rate of 3.8 h"1. The second phase
was designed to characterize VOC, amine catalyst, flame retardant, and blowing agent emissions
from the cured foam at a target air change rate of 0.38 h"1. The third phase was designed to
characterize flame retardant emissions from chamber surfaces after removal of the sprayed
frames from the chamber.
1.6.1.1 Phase I
Phase I began with the 20-minute period of spray application and ended when the
spraying/sampling personnel exited the chamber 3.3 h after the start of the spray event. The four
substrate frames were sprayed in two discrete spray events separated by a five-minute sample
collection event. The personnel in the chamber collected samples to quantify rapidly changing
concentrations of isocyanates and flame retardant at two locations in the chamber. Impinger and
denuder samplers were collected to quantify the concentrations of isocyanates and modified
OSHA vertical samplers (OVS) were collected to quantify gas and aerosol flame retardant
concentrations. At the end of the air sample collection period, the personnel in the chamber
harvested deposition samplers from chamber surfaces and collected and replaced the HVAC
filter that covered the opening to the chamber air exhaust duct. Personnel outside of the chamber
operated the instruments that provided near real-time concentrations of particles and the blowing
agent HFC-134a and collected multi-bed sorbent samplers through a sampling port in the side of
the chamber and from a port in the chamber exhaust duct to quantify VOCs and flame retardant
emissions by thermal desorption with gas chromatography and mass spectrometry (TD-GCMS).
1.6.1.2 Phase II
Phase II began at the end of Phase I with the factor of ten decrease in chamber ACH. The
planned 14-day post-application period for Phase II was extended to 28 days due to the failure of
the HVAC system for the room housing the chamber, which led to an unexpected increase in
chamber temperature and increase in concentrations of emissions. During Phase II, samples were
collected near and away from the SPF insulation by placing samplers into the chamber through
ceiling ports located at each end of the chamber to determine if air concentrations of flame
retardant were higher near the emissions source. Air samples were also collected from the
chamber side port and from the exhaust duct sampling port to characterize air concentrations of
VOCs and flame retardant and provide data to determine the efficiency of the HVAC filter in
removing gas-phase flame retardant. Samples were collected more frequently during the first 24
hours of Phase II to characterize changes in concentrations following the decrease in ACH. At
the end of Phase II, deposition samplers were collected from chamber surfaces, the HVAC filter
was replaced, and the sprayed frames were removed from the chamber. In addition, wipe
SPFI Methods Development
Page 28
-------
samples were collected to characterize flame retardant concentrations on the surfaces of the
chamber behind the frames that had not been exposed during Phases I and II.
1.6.1.3 Phase III
Phase III, which lasted 12 days, began with removal of the sprayed substrates from the chamber.
During Phase III, samples were again collected by air sampling through the chamber ceiling
ports at opposite ends of the chamber to evaluate uniformity of concentrations, and from the
chamber side port and exhaust duct sampling port to evaluate efficiency of the HVAC filter to
remove gas-phase TCPP. As with Phase II, samples were collected more frequently at the
beginning of the phase to track more rapidly changing concentrations. At the end of Phase III,
the HVAC filter and the final sets of deposition samplers and wipe samples were collected.
1.6.1.4 Post-Test Measurements
Additional supporting information to characterize the source emissions material included post-
test determination of density, adhesion, cohesion, and cell structure of the foam; estimation of
mass of overshoot deposited on the floor and walls near the spray application area; and
concentration of TCPP in the cured foam. The test environment was further characterized by
detailed measurements of air speed near the surfaces of the foam.
1.6.2 Data Analysis
The air, surface, and material concentration data were used to calculate mass emitted of target
chemicals and generate application and post-application phase emission factors for isocyanates,
blowing agent, flame retardants, amine catalyst, and other VOCs. The concentration changes due
to the unplanned temperature rise during Phase II were utilized to investigate the impact of
temperature on emission rates of selected compounds. A mass balance approach was employed
to account for the distribution of TCPP generated by the spray application process. Data from the
chamber tests were utilized along with least squares fitting routines to estimate decay constants
for empirical application phase source emissions models. The modified space-state diffusion
model was utilized to compare fundamental mass transfer model predictions of TCPP emissions
with experimental data. Table 1-1 provides a snapshot of the measurements that were conducted
to generate the experimental data.
1.7 Health and Safety
To address health and safety issues related to spraying foam in an environmental chamber we
added an air lock entryway to the chamber, developed and tested procedures for managing entry
and egress from the chamber, and established procedures for backup safety personnel stationed
outside of the chamber whenever personnel were working in the chamber. The operator of the
spray equipment completed the American Chemical Society on-line training for use of low-
pressure spray kits and had used the kits to produce SPF insulation samples many times. The
experiment was conducted with a health and safety protocol approved by the EPA's Office of
Health and Safety.
SPFI Methods Development
Page 29
-------
Table 1-1 Summary of Measurements
Method
Purpose
Location
Sample
Duration
Sample
Frequency
Pre-
Exp *
Phase I**
Phase II***
Phase III
****
Post
Exp
*****
Tracer Gas
Air change rate
Exhaust
3 min
Continuous
y
y
y
y
Thermocouples
Foam curing
Air mixing
Foam
Chamber air
1 s
Continuous
y
y
y
y
y
y
Impingers and
Denuders
Isocyanates,
compare
impingers and
denuders
Chamber
6 min
15 min
4 per hour
2 per hour
y
y
Thermal
Desorption
TCPP and
VOCs in air
Chamber &
Exhaust
1- 100 min
App: 5x
Post: Daily
y
y
y
y
Foil Extraction
TCPP on walls
Chamber
Phase length
Once per
phase
y
y
y
y
Wipe Extraction
TCPP on
surfaces
Exhaust and
Behind
Frames
958 h
670 h
Once
Twice
y
y
y
Tyvek Extraction
TCPP, PMDETA
on PPE
Applicator,
Helper
3.3 h
Once
y
y
Filter Extraction
TCPP, PMDETA
on HVAC filters
Chamber
exhaust
opening
3.3 h, 666 h,
290 h
Once per
phase
y
y
y
y
Modified OVS
TCPP on
particles
Chamber
9 - 900 min
y
y
y
y
Aerosol Particle
Analyzers
Particle
concentration
Chamber
5s
Continuous
y
y
DNPH
Cartridges
Aldehydes and
ketones in air
Chamber
15-115 min
4 per hour
3 per week
y
y
y
*Pre-Experiment: Prior to the spray event
**Phase I: Application through 3.3 h with chamber operating at 4 ACH
***Phase II: Following Phase I through 670 h with chamber operating at 0.4 ACH
****Phase III: Following removal of the sprayed frames from the chamber to monitor TCPP emissions from chamber surfaces for 290 hours
*****POSt-EXperiment: Wipe samples collected from inner exhaust duct surfaces after disassembling duct following completion of test
SPFI Methods Development
Page 30
-------
2.0 Materials and Methods
2.1 Product Selection
Chemical information was compiled from safety data sheets (SDS) for low-pressure SPF kits.
Three low-pressure two component SPF insulation kits were acquired for initial investigation.
One of the three products was selected for use in development of the full-scale test protocol. A
205-boardfoot low pressure SPF insulation kit was acquired from the General Services
Administration (GSA). It was estimated that 70% of the kit would be adequate for spraying 7.6
m2 substrate to the target depth of 5 cm.
2.1.1 Emissions to be Measured
Target compounds for emissions characterization were identified (1) from compounds listed in
the SDS, (2) from emissions identified in micro-scale chamber tests of freshly made SPF
following ASTM D8142 [26], and (3) from emissions determined in pre-test experiments in the
full-scale chamber designed to identify emissions from substrate materials and obtain product
emission range finding data to inform sampling strategies. Compounds selected for quantitation
are listed in Tables 2-1 and 2-2.
Table 2-1 Target isocyanate, blowing agent, volatile and semi-volatile emissions
Chemical Name
CAS#
Methylene diphenyl diisocyanate (MDI)
101-68-8
Polymethylene polyphenyl diisocyanate (pMDI)
9016-87-9
1,1,1,2-Tetrafluoroethane
811-97-2
3-Chloro-1-propene
107-05-01
1,2-Dichloropropane
78-87-5
1,4-Dioxane
123-91-1
2-Methyl-2-pentenal
623-36-9
Chlorobenzene
108-90-7
Pentamethyldiethylenetriamine
3030-47-5
Tris(1-chloro-2-propyl) phosphate
13674-84-5
Bis (2-chloro-1-methylethyl) 2-chloropropyl phosphate (TCPP isomer 1)
76025-08-6
Bis (2-chloropropyl) 2-chloro-1-methylethyl phosphate (TCPP isomer 2)
76649-15-5
The isocyanates listed in Table 2-1 include MDI and polymeric MDI (p-MDI). The latter
consists of several MDI isomers (4,4'-MDI, 2,4'MDI, 2,2'MDI) and oligomers of MDI with
varying numbers of aromatic rings. The standards utilized for quantitation contained di-n-
butylamine derivatives of 4,4'-MDI, which is the most prevalent isomer, and the three and four
ring oligomers referred to as p3-MDI, and p4-MDI.
The flame retardant TCPP is a mixture of up to five isomers. The bulk of the technical mixture is
tris (l-chloro-2-propyl) phosphate (50-85% w/w) and four main isomers that make up >97.9% of
technical grade TCPP. [20] The CAS ID 13674-84-5 refers to the technical grade mixture and
relative amounts of the isomers may vary from manufacturer to manufacturer. TCPP was
SPFI Methods Development
Page 31
-------
quantified as the sum of three main isomers by calibration of the GC/MS with a standard
containing the primary compound and the two most prevalent isomers, bis (2-chloro-l-
methylethyl) 2-chloropropyl phosphate, and bis (2-chloropropyl) (2-chloro-l-methylethyl)
phosphate.
Target aldehydes and ketones identified in micro-scale chamber headspace tests, full-scale
chamber pre-test background samples with the wood substrates in the chamber and identified
with SPF applied to the substrates, are listed in Table 2-2. The aldehydes were quantitated as the
dinitrophenylhydrazine (DNPH) derivatives.
Table 2- 2 Aldehyde and ketone target compound list
Compound
CAS ID #
Micro-scale
Headspace
Test
Full-scale Chamber
Background Test
with Wood
Substrates
Full-scale
Chamber Test
with SPF
Insulation
Formaldehyde
50-00-0
X
X
X
Acetaldehyde
75-07-0
X
X
X
Acetone
67-64-1
X
X
X
Propionaldehyde
123-28-6
X
X
Isovaleraldehyde
590-86-3
X
Valeraldehyde
110-62-3
X
X
Hexaldehyde
66-25-1
X
X
2,5-Dimethylbenzaldehyde
577-94-2
X
2.2 Test Facility
The facility utilized to conduct the experiment included:
• Room-sized emissions test chamber supplied with clean, conditioned air and equipped
with an airlock entry, air supply hoses for supplied air respirators, and mixing fans.
• Plywood substrates mounted against chamber walls on which to spray the SPF insulation.
• Two-component low pressure medium density SPF insulation kit.
• Gas and aerosol sampling systems, including sampling media for collection of gas and
aerosol emissions.
• Particle size and number analyzers.
• Photoacoustic spectrometer for online measurement of the blowing agent HFC-134a.
• GC equipped with electron capture detector (ECD) and automated gas sampling valve for
real-time measurement of the concentration of the tracer decay compound.
• A programmable automatic gas injection system to periodically dose the chamber air
supply with the tracer decay compound.
• Thermal desorption and chemical extraction systems for recovery of emissions from
sampling media.
• GC/MS and LC/MS for identification and quantification of emissions.
SPFI Methods Development
Page 32
-------
• Air velocity measurement instruments for characterizing air speeds near the surfaces of
the SPF insulation.
• Temperature sensors and data recording systems to characterize air temperatures in the
chamber and in the SPF insulation.
2.2.1 Full-Scale Emissions Test Chamber
The EPA's room-sized (30 m3) emissions test chamber, referred to as the Full-Scale Emissions
Test Chamber (FSETC), was modified to serve as an SPF insulation emissions characterization
method development test facility. See Appendix A for a schematic of the chamber and clean air
generation and conditioning system and see Howard et.al., (1995) for detailed description of the
chamber, clean air generation, conditioning and distribution system, and automated control
system [27],
Modifications to the chamber to facilitate spray application of potentially hazardous chemicals in
the enclosed space of the chamber included the addition of:
• Airlock entry room (vestibule) for ensuring that air escaping from the chamber during
entry and egress did not contaminate the laboratory space that houses the chamber.
• Air supply for supplied air respirators. Air compressors placed outside of the room
housing the chamber supplied breathing air for personnel working in the chamber and
vestibule. Two air supply lines penetrated the wall of the chamber through an airtight port
and three lines served the vestibule.
• Aluminum film (Grainger part #4UGG8, soft temper foil thickness of 0.127 millimeter,
30.5 m length, 1.2 m width) was applied to the stainless steel walls of the chamber to
provide a removable lining to minimize potential contamination of chamber surfaces with
flame retardant and SPF insulation.
• Two fans mounted on steel plates were installed near the ceiling to mix the air. Fan speed
and direction were manually adjusted to provide desired air speed of 0 to 0.25 m sec"1 at
0.01 m from the surfaces of the substrates.
• HVAC filters (Minimum Efficiency Reporting Value (MERV), 90 to 98% efficient at
capturing particles sized 3.0 to 10.0 microns), were fitted over the exhaust duct opening
of the chamber to evaluate filter efficiency in removing flame retardant from air leaving
the chamber during each phase of the experiment
• Additional details of modifications to the chamber are provided in Appendix A.
SPFI Methods Development
Page 33
-------
Figure 2-1 Plywood Substrates in foil-lined chamber.
2.2.2 Substrate Frames for Application of SPF Insulation
Four 1.9 m2 substrates shown in Figure 2-1 were produced inbouse from exterior grade 1.27 cm
(0.5") thick 1.22 by 1.83 m (4'X 6') plywood. The plywood substrates were framed and divided
into four quadrants with 3.8 by 3.8 cm (nominal 2" X 2") fir boards. The edges of the substrate
frames were covered with foil tape to reduce emissions from the wood frame material and the
frame edges were taped with foil tape to the chamber walls (not shown) to eliminate dead air
space. Substrate frames were stored in the high bay that houses the chamber for several weeks
prior to the experiment and were placed in the chamber six weeks prior to application of SPF.
2.3 Application Procedure
2.3.1 Preparation for the Spray Events
Prior to the application, the SPF kit was placed overnight in a chamber warmed to 25 °C.
Background samples were collected from the chamber at the low and high ACFI rates. Isocyanate
and OVS sampling media contained in air-tight transport boxes were placed in the chamber for
use during Phase I. The two sampling trees spaced approximately equal-distant along the
centerline of the chamber were populated with OVS and isocyanate samplers, with endcaps in
SPI I Methods Development
Page 34
-------
place. A and B side cylinders of the SPF kit were shaken per manufacturer's instructions just
prior to transport into the chamber.
2.3.2 Spray Event #1
The sprayer/sampler personnel entered the chamber with the SPF kit and a waste bucket
equipped with a carbon-filtered pressure relief opening, connected their respirators to the air
supply lines and prepared to spray the substrates. Sampler endcaps were removed from the OVS
and isocyanate samplers and the A and B side hoses and gun were primed by spraying SPF into
the waste bucket. Immediately after priming the hoses and gun, and closing the top on the waste
bucket, the operator began to spray the plywood frame to his left, then proceeded to the frame
next to it positioned on the back wall of the chamber. Priming the spray gun required less than
one minute. Application to the first two frames was completed in 7.4 minutes. Spraying was
paused for five minutes to harvest samples and install fresh samplers. Vaseline was placed over
the nozzle of the gun and around the trigger to prevent air intrusion and plugging of the gun
during the five-minute pause in spraying for sample collection.
2.3.3 Spray Event #2
Following placement of samplers for collection of samples during the second spray event and
preparation of the SPF kit by replacement of the nozzle, SPF was applied to the third and fourth
frames, beginning with the frame to the right on the back wall. During SPF application to the
third quadrant of the fourth frame, the applicator observed changes to the color and texture of the
SPF indicating that the A to B mixture was off-ratio and the application was terminated prior to
application to the fourth quadrant of Frame 4. Spraying of the second set of frames was
completed in 6.5 minutes.
2.4 Sampling Plan
The air and surface sampling plans were designed to quantify the concentrations and variability
of target emissions in the air as well as the concentrations of flame retardant on surfaces. These
measurements provide the information needed to calculate mass emitted and emission rates of
target compounds during each phase of the experiment. The material sampling plans were
designed to quantify the amount of flame retardant on personal protective equipment (PPE) of
the application/sampling personnel and on the HVAC filters that covered the exhaust duct
opening. Supplemental sampling systems were planned to characterize the test environment and
source of the emissions.
2.4.1 Air Sampling Plan Objectives
• Quantify rapidly changing concentrations of isocyanates and flame retardant over time in
Phase I at two locations in the chamber, close to and away from the area of application.
• Quantify concentrations of blowing agent, flame retardant, amine catalyst, VOCs, and
aldehyde emissions over time for each phase of the test in the chamber and exhaust duct.
• Quantify number and size distributions of particles over time in the chamber during
Phase I.
SPFI Methods Development
Page 35
-------
• Quantify the concentration of the tracer decay compound sulfur hexafluoride (SFr,) for
calculation of ACH rate throughout the test.
2.4.2 Surface Sampling Plan Objectives
• Quantify the amount and distribution of flame retardant on chamber surfaces at the end of
each phase by extraction of TCPP from deposition samplers and wipe samples.
• Quantify the amount of flame retardant on inner surfaces of the exhaust duct, surfaces of
chamber mixing fans, and sampling support stands at the completion of the experiment.
2.4.3 Material Sampling Plan Objectives
• Quantify the amount of flame retardant deposited on the PPE of the spraying/sampling
personnel at the end of Phase I.
• Quantify the amount of flame retardant on the fabric of the HVAC filter covering the
exhaust duct opening at the end of each phase.
• Quantify the concentration of flame retardant in the SPF insulation as applied to the
substrates.
2.4.4 Supporting Information Sampling Plan Objectives
• Monitor temperatures of the SPF insulation during Phase I to determine if temperatures
remained within recommended limits during curing.
• Monitor air temperatures at multiple locations and heights in the chamber to evaluate the
potential for temperature stratification due to heat released from curing foam.
• Measure air speed near the surface of the SPF insulation and deposition samplers to
characterize the variables that impact mass transfer and deposition.
• Characterize foam depth, density, adhesion, and cohesion to characterize the emissions
source.
Sampling and analysis systems used to identify and quantify the concentrations of emissions
from the SPF insulation in the air and of emissions on materials and surfaces are listed in Tables
2-3 and 2-4, respectively. The methods listed may not address the full spectrum of emissions
from SPF insulation. For example, surfactants, polyether polyols, and alkanolamine emissions
may not be detected by these sampling procedures.
SPFI Methods Development
Page 36
-------
Table 2- 3 Air sampling and analysis systems
Emissions Type
Analytical
Measurement System
Method Reference
Sampling Media
Isocyanates (MDI, p3-
MDI and p4-MDI)
Liquid chromatography-
mass
spectrometry/mass
spectrometry (LC-
MS/MS) analysis of
derivatized isocyanates
ISO-17734a [28]
Midget Impingers with outlet
filters, di-n-butylamine
(DBA) in Toluene
Denuder/filter samplers
(ASSET™ EZ4 samplers)
Blowing Agents
(HFC-134a)
Innova™ Model 1412
Photoacoustic
Spectrophotometer
Internal pump pulls air to
measurement cell and
Tedlar® bag whole air lung
sampler for dilution of high
concentrations
TCPP and PMDETA
Air Sampling
Thermal desorption with
GC/MS analysis (TD-
GC/MS)
ASTM D8142 [26]
Multi-bed sorbent traps
TCPP, gas and
aerosol phase
GC/MS analysis of liquid
extracts
NIOSH 5600 [29]
(organophosphorous
pesticides)
Modified OVS, Teflon™ filter
and XAD-2 resin
Volatile Organic
Compounds
Thermal desorption with
GC/MS analysis (TD-
GC/MS)
EPA TO-17 [30]
Multi-bed sorbent traps
Aldehydes
High Performance
Liquid Chromatograph
(HPLC)
EPATO-11 [31]
Collection on Silica Gel
dinitrophenylhydrazine
(DNPH) cartridges
Particle Number,
Mass and Size
Aerosol Particle Sizer
(APS)
Electrostatic Impactor
(ELPI)
TSI Manual
Dekati Manual
Internrnal pumps for each
instrument pull air from
chamber through a common
sampling line extending into
the chamber
Table 2- 4 Material and surface sampling and analysis
Emissions Type
Analytical
Measurement System
Method Reference
Sampling Media
Flame Retardant
Depostied on Exhaust
duct Filter
GC/MS analysis of
Liquid Extracts
NIOSH 5600
MOP 7472
Samples cut with 82.5 mm
diameter punch (3.25 Inch)
from MERV-13 HVAC filter
TCPP Deposition on
Interior Surfaces
GC/MS analysis of
Liquid Extracts
NIOSH 5600
64 mm diameter aluminum
depostion samplers, wipe
samplers
TCPP Deposited on
Fabric of
Spray/Sampling
Personnel
GC/MS analysis of
Liquid Extracts
NIOSH 5600
Samples cut from Tyvek®
suits with 82.5 mm diameter
punch (3.25 Inch)
2.5 Air Sampling Systems
Air sampling systems included (1) a sampling system utilizing critical orifices for control of
sampling flow rate for collection of isocyanate and OVS samplers mounted on sampling support
stands in the chamber, (2) sampling systems that employ vacuum pumps and mass flow
controllers for collection of OVS samplers, multi-bed sorbent traps and DNPH-treated silica gell
SPFI Methods Development
Page 37
-------
traps placed into the chamber through sampling ports in chamber walls, ceiling, and the exhaust
duct sampling port, and (3) online analysis systems with dedicated internal sampling pumps and
sampling lines extending into the chamber or exhaust duct through sampling ports.
2.5.1 Air Sampling System for Collection of Isocyanates and Flame Retardants
The air sampling system constructed for rapid collection of samples from inside the chamber by
the spraying/sampling personnel during Phase I consisted of:
• Vacuum pump, pressure gage, and master shut-off valve positioned outside of the
chamber in the vestibule.
• Two eight-port manifolds in the chamber connected to the vacuum pump through a
stainless-steel bulkhead fitting and a 9.5 mm (3/8') outside diameter (OD)
polytetrafluoroethylene (PTFE) line that penetrated the wall of the chamber.
• Sixteen 6 mm OD PTFE color-coded sampling lines each connecting to one port of a
manifold and fitted with a stainless-steel shutoff valve and a critical orifice.
• Eight lines run to one of two sampling media holding stands, referred to as Sampling
Tree 1 or Sampling Tree 2. Sampling trees were positioned along the centerline of the
chamber at 1.5 m and 3 m from the wall opposite the door.
• Each sampling tree consisted of an upright pole with two cross arms fitted with clamps to
hold sampling media at approximate breathing height and at waist height.
• The system collected impinger/filter isocyanate and OVS samplers at 1 L min"1, and
denuder-filter isocyanate samplers at 0.2 L min"1. See Appendix B-l. 1 for additional
details.
• The exhaust from the vacuum pump was routed through a carbon canister to remove
solvent and to the chamber exhaust duct upstream of the flow measurement station to
ensure that the air removed from the chamber by sampling did not result in increased
chamber supply air flow rate.
2.5.1.1 Verification of Flow Rates
Prior to use, the sampling flow rates for each sampling line were verified by measurement with a
certified electronic bubble flow meter. Flow rates were verified with maximum number of
samplers running to ensure that the vacuum pump had the capacity to maintain flows for the
entire system. Sampling rates for each sample line were checked at the end of the test. Sample
volumes, determined as the product of average flow rate (mL min"1) and sample time (min), were
standardized to barometric pressure of 760 mm Hg and 23 °C.
2.5.1.2 Sampling Media for Collection of Isocyanates and TCPP
Isocyanates were collected from the air using two sampling approaches described in ISO 17734a
which utilizes di-n-butylamine (DBA) to derivatize the highly reactive isocyanates to stable
SPFI Methods Development
Page 38
-------
compounds. Eighteen impinger and 24 denuder-filter samplers were collected during Phase I to
characterize isocyanate concentrations during and following the application.
Sampling media included:
• Midget impingers containing 0.1M DBA derivatizing agent in toluene with a backup filter in
the outlet to capture particles less than 2 |im in diameter that may pass through the solution.
• DBA impregnated denuder/filter samplers (Sigma-Aldrich, 5028U).
• OVS modified with addition of a PTFE filter in front of the glass fiber filter with sampling
flow rate of 1 L min for collection of aerosol and gas-phase TCPP.
The inlets to the impingers were oriented up and the inlets to the denuder-filter and OVS
samplers were oriented at approximately 90° and toward the area of spray application.
2.5.2 Air Sampling System for Quantification of Blowing Agent HFC-134a
A photoacoustic spectrophotometer (PAS) (Innova Model 1412 Photoacoustic Field Gas-
Monitor LumaSense, Technologies, Inc., Santa Clara, CA) was used to provide near real-time
measurements of HFC-134a concentration by sampling from the chamber exhaust duct sampling
port (DE-1) through a 3.2 mm (1/8") OD diameter PTFE sampling line that connected to the
sampling port of the instrument placed on a table located on top of the chamber. The internal
pump in the instrument pulled air from the exhaust duct port to the instrument at a flow rate of
30 mL min"1 to flush the sampling lines, and at 5 mL min"1 to flush the measurement cell. The
initial sampling frequency of one per minute was reduced to one sample per three minutes at 116
h.
It was expected that concentrations of the blowing agent HFC-134a would exceed the calibration
range of the Innova 1412 during the period of spray application. Eight whole air samples were
collected in Tedlar® bags with a lung sampler as described in EPA Method 18 during the first
two hours of Phase I. Air pulled with a vacuum pump from a rigid airtight box containing the
sampling bag for ten-minute periods was displaced by air pulled into the bag through a 6.4 mm
OD (1/4") PTFE sampling line connecting the inlet port on the lung sampler to the exhaust duct
sampling port DE-1. Sample flow rate of 100 mL min"1 was controlled by a mass flow controller
mounted on a sampling cart located on the top of the chamber. See Figure B-7 in Appendix B
for details of the lung sampler system.
2.5.3 Air Sampling System for Collection of VOCs with Multi-bed Sorbent Traps and
Flame Retardant for TD-GC/MS Analysis
VOC and TCPP emissions were collected for subsequent analysis by TD-GC/MS by pulling air
from the chamber or exhaust duct sampling ports through multi-bed sorbent traps (Markes
International, C3-AXXX-5304 containing Tenax® TA and Carbograph 5TD) at a rate of 100 mL
min"1 over sampling periods of 0.5 to 100 minutes using a vacuum pump and mass flow
controllers. Three 0.1 L volume samples were collected during the application using a gas-tight
syringe (Gastec 810-GV100). For samples collected from the chamber side port, the multi-bed
sorbent trap was placed at the end of 6 mm OD stainless steel tube of 1 m length that was placed
SPFI Methods Development
Page 39
-------
into the chamber through sampling port located mid height at the center of the left wall of the
chamber. This placed the inlet end of the multi-bed sorbent trap in a horizontal position about
1.3 m above the chamber floor and 1 m from the wall. The multi-bed sorbent traps collected
from the exhaust duct sampling location were collected by placing the multi-bed sorbent trap into
a bored-through 6 mm diameter stainless steel fitting located at the duct exhaust sampling port
(DE-1). This placed the inlet end of the sampling trap inside the duct and approximately 5 cm
from the interior wall.
Samples were collected rapidly from the chamber side port and exhaust duct sampling port
during the application then less frequently for the duration of Phase I. During the 3.3 h period of
Phase I, 11 samples were collected from the exhaust duct sampling port and 13 were collected
from the chamber side port. During the more than 665 h of Phase II, 13 samples were collected
from the exhaust duct port and 23 from the chamber side port. During the 290 h of Phase III, six
samples were collected from each sampling port location. Samples were collected more
frequently at the beginning of each phase when concentrations were expected to change at a
higher rate.
Sampling flow rates were checked with a calibrated electronic bubble flow meter prior to and
following collection of each sample. Start and stop time for each sample, temperature, and
barometric pressure were recorded in an electronic spreadsheet along with the sampling flow
rates determined as the average of pre- and post-sample flow checks. Sample volumes,
determined as the product of average flow rate (mL min"1) and sample time (min), were
standardized to barometric pressure of 760 mm Hg and 23 °C.
2.5.4 Air Sampling for TCPP and PMDETA with Modified OVS
Gas and aerosol phase TCPP were collected (1) from the air by pulling air through modified
OVS for analysis by injection of solvent extracted TCPP and PMDETA onto the column of a
GC/MSD. The OVS samplers were employed to provide insight into the gas-particle distribution
of TCPP during the application and to provide a backup to the thermal desorption samples where
mass of TCPP collected on multi-bed sorbent traps during the application phase may exceed the
calibration range of the TD-GC/MS. See Appendix B-2 for a description of modifications to the
OVS by addition of a membrane-backed Teflon filter (Whatman PTFE Membrane Filters, WTP
Type, Cat. #7590 004 1.0 |im pore size) in front of the glass fiber filter (GFF) of the OVS.
2.5.4.1 Extraction of Analytes from Modified OVS
The analytes collected on the filters, resin beds, and polyurethane foam (PUF) plugs were
recovered by solvent extraction with sonication. The Teflon™ filter, front XAD-2 resin bed and
the backup XAD-2 resin bed plus PUF separator were extracted separately in 2 mL acetone.
Teflon filters were spiked with 10 |iL of the recovery check standard (RCS) tripropyl phosphate-
d2i, prior to extraction and 10 |iL of the internal standard (IS) acenapthalene-dio prior to analysis.
XAD-2 resin beds were spiked with 20 |iL of IS and RCS. A portion of the extract was injected
into the GC/MS for identification and quantification. Liquid extracts were diluted for re-analysis
where the amount of an analyte fell outside of the calibration range.
SPFI Methods Development
Page 40
-------
2.5.4.2 Air Sampling Locations for Modified OVS
Air samples were collected from inside the chamber during Phase I by positioning modified OVS
on Sampling Trees 1 and 2 shown in Figure 2-3 and pulling air through the samplers at a rates of
1000 mL min"1 for periods ranging from 9 to 30 minutes. During Phase II, modified OVS were
collected by placing samplers into the chamber one meter below the level of the ceiling through
Ports CT2 (2.4 m from the sprayed frames) and CT4 (1.2 m from the sprayed frames) located on
centerline in the top of the chamber to compare TCPP concentrations in the chamber near and
away from the sprayed substrate frames. Eighteen OVS samples were collected from within the
chamber during Phase I and an additional 60 samples of 50 to 250 L were collected by placing
samplers into the chamber through the ceiling ports over the remainder of the experiment with
sampling flow rates of 250 and 500 mL min"1.
2.5.5 Air Sampling for Aldehydes and Ketones
Aldehydes were collected by pulling air through cartridges containing silica gel coated with
DNPH (Sep-Pak) using vacuum pumps and mass flow controllers. Samples were collected by
attaching the cartridges to a 6 mm OD stainless steel sampling tube which was placed into the
chamber through the chamber side port CS-11. Sampling flow rates of 1 L min"1 were used to
collect 60 L chamber background samples at the low and high chamber flow rates. Flow rates of
0.5 L min"1 were employed to collect samples ranging from 8 to 55 L during the test. Sampling
flow rates were recorded before and at the end of each sampling event. Sample volumes,
determined as the product of average flow rate (L min"1) and sample time (min), were
standardized to barometric pressure of 760 mm Hg and 23 °C. Five samples were collected
during Phase I and 14 samples were collected at 12 sampling periods between 4 h and 480 h.
DNPH samplers were not collected during the decay phase of the experiment.
2.5.6 Air Sampling to Characterize Particle Size and Number
Two real-time particle size and number characterization instruments, a dual laser time-of-flight
instrument (TSI Model 3321 Aerosol Particle Sizer, APS), and an electrostatic low pressure
impactor (ELPI, Dekati Model 97-2E) were utilized to measure aerosol concentrations and size
distributions during Phase I of the experiment. The instruments were positioned on the floor of
the mezzanine above the ceiling of the chamber. A 0.635 cm OD stainless steel common
sampling line penetrated through the chamber ceiling just off-center of the chamber, terminating
at a breathing zone height of-1.6 m from the chamber floor. The common sampling line
connected to a splitter above the ceiling of the chamber. The sample inlet line from each
instrument connected to one side of the splitter. The APS sampled at a rate of 5 L min"1 and the
ELPI sampled at 10 L min"1. The sample exhaust from each instrument was routed back into the
chamber system's exhaust ducting upstream of the flow rate measurement station to ensure that
the air withdrawn from the chamber by the particle counting instruments did not cause the
chamber flow control system to increase the supply air flow. A 20:1 aerosol dilutor was used as a
front end to the APS.
SPFI Methods Development
Page 41
-------
2.6 Surface Sampling Systems for TCPP
Deposition and wipe samples were utilized to characterize TCPP concentrations on surfaces in
the chamber.
2.6.1 Deposition Samplers
Foil coupons of 64 mm diameter were cut from the aluminum foil used to line the chamber using
a cutting jig fabricated by EPA's machine shop. Coupons were cleaned by sonication (Bransonic
2510) in methanol for 15 minutes, air dried, then baked overnight in a laboratory oven at 125 °C.
The coupons were removed from the oven and triple wrapped in aluminum foil until placement
in the chamber. The deposition coupons were mounted in the designated locations inside the
FSETC using photo mounting squares (MBI, a Division of MCS Industries, Inc.). The acid and
PVC-free mounting squares were extracted, analyzed, and determined to be free of interferences
related to the analytical methodology for the determination of flame retardants by direct injection
GC/MS.
Figure 2-2 Deposition sampler array at the base of a sampling tree.
2.6.1.1 Deposition Sampler Placement
Twenty-four hours prior to the start of the test, deposition samplers were placed in arrays of six
or seven samplers centrally located on three of the four walls, the floor, and the ceiling. No
SPFI Methods Development
Page 42
-------
deposition samplers were placed on the rear wall due to fact that most of the wall area was
occupied by Frames 2 and 3. Deposition samplers were placed on the foil-covered concrete-filled
buckets shown in Figure 2-2 that provided support for Sample Tree 1 and Sample Tree 2, on the
ceiling near sampling ports CT-2 and CT-4, on the door wall half-way between the door and left
wall, and on the right and left walls centrally located between the edge of the frame against that
wall and the edge meeting the door wall. Approximate locations (not to scale) for the arrays of
deposition samplers are shown as "LEFT, RIGHT, FRONT" for sampler arrays placed at mid
height on the walls, "CEILING 1 and CEILING 2", for arrays placed on the ceiling, and "Tree 1
and Tree 2" for arrays placed at the bases of the sampling trees, as shown in Figure 2-3.
2.6.1.2 Deposition Sampler Collection
• A sampler was collected from each of the seven locations just prior to the start of the test
to act as blanks.
• Two samplers were collected from each location by the spraying/sampling personnel
prior to exiting the chamber at 3.3 h.
• Two samples were collected from each of the seven locations at 669 h when personnel
entered the chamber to remove the frames.
• Two samples were collected from each sampling location at the end of the test.
• At collection, coupons were loosely coiled and placed into vials for storage and transport
to the laboratory.
The 40 mL amber glass VOA vials (Restek p/n 21797), pre-cleaned to EPA Protocol B
specifications, with an open top 24/400 cap and a 24 mm diameter PTFE faced 0.125" silicone
septa, were used for storage and extraction of the flame retardants deposited onto the aluminum
samplers.
2.6.1.3 Deposition Sampler Extraction
10 mL acetone was pipetted into each vial upon receipt in the laboratory. 100 |iL of RCS
(containing d2i-tripropyl phosphate and d27-tributyl phosphate at 25 ng uL"1) was added to the
vial prior to extraction. The vials were secured in a vial rack that was mounted horizontally on an
orbital shaker and shaken for 30 min. After the extraction, 100 |iL of IS (dio-Acenaphthene at 10
ng uL"1) was added to each vial. The vials were shaken by hand for 10 sec to distribute the IS
solution.
SPFI Methods Development
Page 43
-------
Frame 2
Frame 3
Figure 2-3 Locations of arrays of deposition samplers
2.6.2 Wipe Samples
Eighty-four wipe samples were collected at the end of Phase III to estimate deposition of flame
retardant on surfaces of the mixing fans, sampling trees, exhaust duct, walls, ceiling, and the area
behind the frames for comparison with deposition samplers. Repeat wipes were collected from
each area sampled to evaluate wipe efficiency. Three wipes (wipe plus two re-wipes) were
collected for exhaust duct (glass and stainless steel), wall (foil), sampling tree (stainless steel),
and behind the frames (foil). Two wipes (wipe plus one re-wipe) were collected from each of
five sample locations for each fan (plastic).
2.6.2.1 Wipe Sample Collection Procedure
A pre-cleaned 10 cm by 10 cm PTFE template was laid over the area to be wiped. A gauze
sponge folded and held using stainless steel forceps was wetted with 0.5 mL acetone. The area
SPFI Methods Development
Page 44
-------
within the template was wiped in the horizontal and vertical direction. The area was wiped a
second time with a new wipe. Each wipe was transferred to an amber 20 mL vial for storage and
extraction.
2.6.2.2 Wipe Sample Extraction
Prior to extraction, 100 uL of RCS was added to the wipe. Wipes were extracted in 10 mL
acetone for 30 minutes in an ultrasonic bath. Prior to analysis, 100 jiL of IS was added to the vial
which was capped and shaken prior to transfer of an aliquot to an autosampler vial for analysis
by GC/MS. Estimated minimum quantifiable level was 1.5 ng cm"2.
Figure 2-4 Wipe sampling template mounted in glass exhaust duct.
2.7 HVAC Filter Samples to Quantify TCPP Removed from the Chamber
Exhaust Flow
MERV 13 HVAC filters (90 to 98% efficient at capturing particles sized 3.0 to 10.0 microns)
were placed over the exhaust duct opening of the chamber as shown in Figure 2-5. The filter was
held in place with a rectangular steel plate with a cutout the size of the exhaust duct opening.
Edges of the plate were taped to the ceiling of the chamber to prevent leakage around the edges
of the filters. Filters were collected and replaced with clean filters at the end of Phase I and II.
The third filter was collected at the end of the test following 270+ h of monitoring TCPP
emissions from non-target surfaces in the chamber.
SPI I Methods Development
Page 45
-------
2,7.1 HVAC Filter Sample Collection and Extraction
To determine the amount of flame retardant on each filter, five 8.2 cm diameter discs were cut
from the filter with a punch. One sample was cut from the center and four others were cut from
the midpoints of diagonal lines running from corner to corner across the filter. Each filter
subsection was placed in a 40 mL vial and capped until extraction. Prior to extraction, 200 |aL
RCS and 20 mL acetone were added to the vial. Vials were sonicated for 30 min. 200 ,uL of IS
was added to each vial prior to analysis of an aliquot by GC/MS. The estimated minimum
quantifiable TCPP level based upon extraction of 53.5 cm2 filter sample in 20 mL solvent was 4
ng cm"2.
Figure 2-5 HVAC filter placed over outlet duct opening
2.8 Sampling to Quantify TCPP Deposited on PPE
Samples were collected from the PPE (Dupont Tyvek Coverall TY120SWHXL0025NF) worn by
the spray/sampling crew after exiting the chamber to estimate the amount of TCPP flame
retardant leaving the chamber system on the PPE. For exposed suits, two 53.5 cm2 samples were
cut from each sampling location on the suits, two locations on the torso front and one on the back
and one on each leg of the coverall.
The samples were placed into 40 mL amber VOA vials to which was added 200 jj.1 RCS and 20
mL acetone.
SPI I Methods Development
Page 46
-------
2.9 Collection of SPF Insulation Samples for Determination of TCPP
Concentrations
Fourteen samples of the cured foam insulation were cut from the substrates at the end of the
emissions test period using a 14.9 mm diameter coring device. Samples were stored in pre-
cleaned amber wide mouth jars with Teflon-lined caps. Each core was weighed on an analytical
balance and transferred to a Whatman cellulose extraction thimble. The extraction thimble was
placed into a Soxhlet extractor. 250 |iL of RCS was added to the surface of the foam core in the
thimble and the sample was extracted with methylene chloride for 18 h at a temperature setting
to yield six solvent cycles per h. At completion of extraction, the extract was concentrated to 10
mL and transferred to a 25 mL volumetric flask and brought to volume with methylene chloride.
100 |iL of extract was added to 20 |iL of IS and 1880 |iL of solvent in a 2.0 mL volumetric flask.
100 |iL was transferred to an autosampler vial for analysis by GC/MS.
2.10 Measurement Systems
Measurement systems described in this section include; (1) the LC-MS/MS for quantification of
isocyanates, (2) PAS for quantification of HFC-134a, (3) TD-GC/MS system for quantification
of VOCs, amine catalyst, and TCPP collected on multi-bed thermal desorption traps, (4) GC/MS
system for quantification of extracts from OVS, wipe samples, deposition samples, HVAC
filters, PPE samples, and SPF insulation cores, (5) HPLC for quantification of aldehydes and
ketones, (6) particle and number counting instruments for characterization of particle emissions,
(7) supplemental systems for measurement of air temperatures, foam temperatures, and air
velocity near the surfaces of the foam.
2.10.1 LC-MS/MS Quantification of MDI, p3-MDI, and p4-MDI
The target isocyanates were quantified by analysis of the DBA derivatives of the target analytes
using a Agilent 1100 HPLC coupled to an ABSciex 3200 Triple Quadrapole Mass Spectrometer.
The HPLC was equipped with an Ascentis Express C18 HPLC guard column (p/n 53501-U,
Supelco) and Ascentis Express C18 HPLC colulmn (p/n 53822, Supelco, Bellefonte, PA).
2.10.1.1 Sample Preparation
Samples collected in impingers or on dry samplers were worked up for analysis using the ISO
17734a protocol. Deuterated analogues of each isocyanate analyte were added to the toluene
extraction solution for the dry sampler or directly to the impinger solution as an IS. Following
sonication, the extraction solution was taken to dryness under a nitrogen purge and the sample
was reconstituted with acetonitrile. Calibration standards which were the deuterated analogues of
isocyanates followed the same protocol.
2.10.1.2 Impinger and Dry Sampler Analysis
Analysis of the derivatized isocyanates was accomplished by injection from an autosampler into
an Agilent 1100 HPLC coupled to an ABSciex API 3200 triple quadrupole mass spectromether
SPFI Methods Development
Page 47
-------
(LC-MS/MS). The response for each DBA derivative for each isocyanate in the calibration was
determined relative to the response of the corresponding deuterated analogue. A multipoint
calibration curve was generated at the beginning and at the end of each analysis sequence.
2.10.1.3 Calibration of the LC-MS/MS for Quantitation of Isocyanates
A certified stock solution (Supelco CRM40603 XA14439V) containing 4,4'-MDI, p3-MDI, and
p4-MDI at concentrations of 174.8, 90.1, and 33.8 |ig mL"1, was diluted 1:100 to produce
working stocks that were further diluted to generate calibration solutions at eight concentration
levels ranging from 1.7 to 437, 0.9 to 225, and 0.3 to 85 ng for MDI, p3-MDI, and p4-MDI,
respectively. An IS mix (Supelco CRM40604 XA18139V) with concentrations of 180.8, 94.2
and 35.6 |ag mL 1 was diluted 1:100 to create the IS working stock. The standards were DBA-
deuterated derivatives of the MDI parent compound, MDI-DBA-d9, p3-MDI-DBA-d9, and p-4-
MDI-DBA-d9.
2.10.2 Photoacoustic Spectrophotometer (PAS) for Quantification of HFC 134a
The Innova Model 1412 PAS Field Gas-Monitor (LumaSense, Technologies, Inc., Santa Clara,
CA) was employed to quantify the blowing agent HFC-134a. The use of the instrument for
quantification of application phase HFC-134a concentrations was considered exploratory due to
several factors; the HFC-134a concentrations were expected to exceed the linear range during the
application, and the instrument is subject to cross interference between compounds, including
water vapor.
2.10.2.1 Calibration of the PAS
The PAS was factory calibrated for HFC-134a prior to the experiment. The manufacturer
calibration provided internal compensation factors for water vapor and SF6 interactions with the
HFC-134a calibration. During evaluation of the instrument, it was noted that response to HFC-
245fa, another blowing agent commonly used in high pressure applications, was equivalent to
the HFC-134a response. This observation and availability of a 5000-ppm certified standard was
exploited to extend the calibration of the instrument above the 100-ppm factory calibration.
The PAS was field-calibrated by dilution of certified HFC 134a and HFC 245fa gas standards
with zero air over three ranges, 0 to 12 ppm (50 mg m"3), 0 to 200 ppm (835 mg m"3), and from 0
to 5000 ppm (20.9e03 mg m"3). Response was linear through 200 ppm and decreased
significantly above 3000 ppm. The lung sampler was employed to collect whole air samples for
off-line dilution to quantify HFC 134a where concentrations were expected to exceed the useful
calibration range of the instrument. Concentrations were above 2500 ppm for six samples
between 0.36 and 0.43 h. Those data were flagged and were not used in calculation of mass
emitted. The low-level calibration factor was used to process HFC-134a concentrations below 12
ppm because of the improved zero intercept for the reduced calibration range. The mid-level
linear calibration was used to process instrument responses between 12 and 200 ppm. The
minimum detectable level for the PAS was 0.04 ppm or 0.17 mg m"3.
SPFI Methods Development
Page 48
-------
2.10.3 TD-GC/MS Analysis System for Quantification of VOCs, PMDETA, and TCPP
An automated system (Markes TD-100 thermal desorption/concentration unit coupled to an
Agilent 7890A/5975C Inert GC/MS) was employed for analysis of the multi-bed sorbent traps.
The TD-100™ was operated using Markes International Thermal Desorption System Control
Program version 5.1.0 software, while the Agilent GC/MSD system was operated using the
Agilent Enhanced ChemStation E.02.02.1431 software package. The software for both Markes
and Agilent equipment was loaded on a Hewlett-Packard Compaq 8000 Elite personal computer.
The TD-100 was fitted with a Marks Material Emissions Cold Trap (U-T12ME-2S, Gwaun Elaj
Medi-Science Campus, Llantrisant RCT CF72 8XL, UK). The transfer line from the TD-100 to
the column of the GC (Rtx®-5 Amine, Restek Corporation, 110 Benner Circle, Bellefonte, PA,
16823) was through a Restek 0.25 mm ID fused silica transfer line. The transfer line from the
TD-100 was connected to the capillary column in the oven of the GC with an Agilent Ultimate
Union (G3182-61580, Agilent, Santa Clara, CA). Ultrapure nitrogen was used as the purge gas
for the TD-100 and ultrapure helium was the GC carrier gas. The standard 3 mm draw out plate
in the detector was replaced with a 6 mm plate. Operating parameters for the TD-100 and the
GC/MSD are found in Appendix C-l.
2.10.3.1 Internal Standards for the TD-GC/MS
Prior to analysis, 2 |iL of a solution containing 50 ng per |iL of l,4-dichlorobenzene-d4 and 50
ng per |iL of acenaphthene-dio was injected onto each trap using a 10 |iL gas tight syringe with a
Cheney adapter set to 2 |iL with a commercial trap loading device for use as IS for PMDETA
and TCPP, respectively. An automated system injected 1 mL of 43.8 ppm toluene-dsIS onto
each sorbent trap prior to thermal desorption as IS for VOCs.
2.10.3.2 Sample Analysis by TD-GC/MS
The analytes were transferred from the sample collection trap to a concentrator trap by heating
and purging with ultrapure inert gas in reverse direction of sample collection. Compounds
concentrated on the focusing trap, were flash desorbed to the column of the gas chromatograph
and identified and quantified by mass spectrometry in scan mode over the range of 35 to 450
atomic mass units (amu). VOC area count response was quantified relative to toluene-dg, TCPP
area count response was quantified relative to acenaphthene-dio, and PMDETA was quantified
relative to dichlorobenzene-d4.
2.10.3.3 Calibration of the TD-GCMS System
The TD-GC/MS system was calibrated for target compounds by loading clean sorbent traps with
standards prepared by serial dilution of reference standards with guaranteed purity of 95% or
better. The TCPP primary standard was supplied as a mixture of the primary compound, TCPP,
and two isomers in concentrations of 67.2, 26.2 and 4.1% respectively. The data station software
(Chemstation) stored the calibration data, and reported mass amounts for each analyte, corrected
by recovery of the IS. TCPP was reported as the sum of the three main isomers. See Appendix C,
Table C-3 for sources of primary calibration standards.
SPFI Methods Development
Page 49
-------
2.10.4 GC/MS Analysis System for Analysis of TCPP and PMDETA from Extracts of
Sampling Media
An Agilent 6890/+5973 GC/MS with 6890 series split/splitless injector running under
Chemstation software was used to quantify TCPP and PMDETA from extracts of various
sampling media. Analytes extracted from media as described below were injected onto an
RTX®5-Amine MS column (Model 12338, Retek Corporation, Bellefonte, PA). The system was
operated in selective ion monitoring (SIM) mode to maximize sensitivity. Operating parameters
for the system are presented in Appendix C-2.
2.10.4.1 Extraction of TCPP and PMDETA from Modified OVS
The analyte collected on the filters, resin beds, and PUF plugs of the OVS was recovered by
solvent extraction at 1000 rotations per minute (rpm) for 15 min. The Teflon™ filter, front XAD-
2 resin bed, and the backup XAD-2 resin bed plus PUF separator were extracted separately in 2
mL acetone. Teflon filters were spiked with 10 |iL of the RCS tripropylphosphate-d2i, prior to
extraction and 10 |iL of the IS acenapthalene-dio prior to analysis. XAD-2 resin beds were
spiked with 20 |iL of IS and RCS. A portion of the extract was injected into the GC/MS for
identification and quantification. Liquid extracts were diluted for reanalysis where the amount of
an analyte fell outside of the calibration range.
2.10.4.2 GC/MS Analysis of TCPP and PMDETA Extracted from Modified OVS
The GC/MS system was used to determine amounts of TCPP and PMDETA in extracts from the
OVS samplers. 9-point calibration curves for TCPP, PMDETA, and the RCS compounds were
prepared over the range of 10 ng mL"1 to 500 ng mL"1 for TCPP and 250 to 3000 ng mL"1 for
PMDETA. The IS solution recommended for SVOC analysis in SW-846 Method 8270D was
added to each sample and calibration standard to yield a concentration of 100 ng mL"1 in the
resulting solution/extract. The IS solution contained l,4-dichlorobenzene-d4; naphthalene-dx;
acenaphthene-dio; phenanthrene-dio; chrysene-di2; and perylene-di2. PMDETA was quantified
using l,4-dichlorobenzene-d4and TCPP was quantified using acenapthene-dio as the IS.
Tripropylphosphate-d2i and tributyl phosphate-d27 were evaluated as RCS. Both RCS
compounds were quantified using acenaphthene-dio as the IS. Tributyl phosphate was not used in
data evaluation because of interferences. Phenanthrene-dio and perylene-di2 were not integrated
or reported.
2.10.5 HPLC/DAD System for Analysis of DNPH Derivatives of Aldehydes and
Ketones
DNPH derivatives of aldehydes collected on DNPH cartridges were eluted with 5 mL
acetonitrile. Aldehyde -DNPH derivatives were identified and quantified by high performance
liquid chromatography per EPA TO-11A (Agilent 1200 HPLC with Diode Array Detector,
HPLC/DAD). The HPLC was calibrated with certified standards at six concentrations over the
range of 0.03 to 15 |ig mL"1 from serial dilutions of the 15 |ig mL"1 calibration stock solution
(CRM4M7285, lot number XA20836V, Supelco) that contains DNPH derivatives of a suite of 16
aldehydes and ketones. Daily calibration checks (DCCs) were run at the beginning of each
analytical run sequence. Independent audit program (IAP) samples are run whenever the
SPFI Methods Development
Page 50
-------
instalment must be recalibrated due to maintenance or repair. Reagent blanks and media blanks
were analyzed with each sample set. Instrument detection limits (DDLs) for each analyte are
presented in Table 3-23. DDLs, determined as |ig ml"1 are also presented as minimum detectable
concentrations assuming air sample volumes of 60 L and extraction of DNPH cartridges with 5
mL of acetonitrile and 10 jj.1 injection volume.
2.10.6 Particle Analysis Systems
Two particle size and number counting instruments were employed to characterize particle sizes
over the range of 0.04 to 20 |im. Number concentration (particles cm"3) was selected as the unit
of measure.
2.10.6.1 Dual Laser Time-of-Flight
The dual laser time-of-flight instrument (TSI Model 3321 Aerosol Particle Sizer, APS) has a
measurement range from 0.5 to 20 micrometers. Particle size data were collected in 54 size bins.
As a particle passes through a detection chamber, parallel lasers optically measure its velocity. A
velocity profile contained within the instrument's firmware converts the time-of-flight data to an
aerodynamic particle diameter.
2.10.6.2 Electrostatic Low-Pressure Impactor
The electrostatic low pressure impactor (ELPI, Dekati Model 97-2E) is a 12-stage electrical
impactor with a measurement range from 0.04 to 8.5 |im. As particles pass through and collect
on impactor plates, they impart an electric current measured by electrometers and this signal is
converted to particle number count for the size bin associated with each impactor stage.
2.11 Measurement Systems for Collection of Supporting Information
Measurement systems were employed to provide supplemental information about the
environmental conditions of the test environment. The systems included measurement systems to
provide detailed picture of air temperatures in the chamber and temperatures in the foam. An air
velocity measurement system was employed to provide insight into the air velocities above the
foam surfaces and above the deposition samplers at various locations in the chamber.
2.11.1 Temperature Measurement Systems
Two temperature measurement systems were deployed to measure and record temperature of the
foam that was applied to the substrates and to provide finer-scale record of air temperatures at
various locations in the chamber than afforded by the resistance thermocouple devices (RTDs) of
the chamber control system.
2.11.1.1 Foam Temperature Sensors
Tip welded thermocouples, manufactured inhouse and calibrated by the Metrology Laboratory
were attached to the substrate frames to record temperatures at the foam-plywood interface, the
SPFI Methods Development
Page 51
-------
expected half-depth of the applied foam, and in the air above the top of the substrate frames.
Thermocouple wires were run from the frames in the chamber through a sampling port in the rear
wall to a laptop PC with Pdaq (Personal Daq/50 Series USB Data Acquisition Modules, IOtech
Cleveland Ohio) data system positioned outside of the chamber. Initial sampling frequency of six
per minute was reset to one per minute after 24 h, then reduced to one per five min after six days.
2.11.1.2 Supplemental Air Temperature Sensors
Vertical arrays of four temperature sensors with data loggers (Onset HOBO series data loggers)
were deployed at adjacent corners of the chamber at elevations ranging from 0.3 to 2.2 m above
the floor. An array of three temperature sensors was attached to the sampling stand near the
center of the chamber at elevations ranging from 0.3 to 1.6 m.
2.11.2 Air Velocity Measurement System
An omnidirectional hot-wire air velocity sensor with temperature sensor for temperature
compensation (Clinomaster Model 6501 Series Multifunction Anemometer, Kanomax, Andover,
NJ) was used to measure air velocity across the surfaces of the substrate panels. The
measurement range is 0.01 to 5.00 m s"1, resolution is 0.01 m s"1, and accuracy is +/- 0.02 m s"1
for velocities from 0.01 to 0.99 m s"1. Air velocity measurements were conducted prior to the test
to set the speed and direction of the fans and following completion of the test to document air
velocities across the uneven surfaces of the SPF insulation.
2.12 Data Analysis
The data analysis plan was designed to (1) demonstrate the use of data generated by the
emissions test to calculate emission rates and construct source emissions models from the Phase I
(application) and Phase II (post application) emissions data and (2) identify potential sources of
variability and uncertainty for the testing approach.
For dynamic chamber tests, the relationship between the emission rate (R) and concentrations of
emissions in the chamber (C) is described by:
R(t) = + QC+SflJ Equation (2)
Where:
R = emission rate (|ig h"1) of the emissions source at time t (h).
V= chamber volume (m3).
dC/dt = the change in air concentration over the time period dt.
O = the flow rate of supply air to the chamber (m3 h"1).
C = the concentration of the emission in the air of the chamber (m3).
S(t) = rate of loss of mass to chamber surfaces, chemical reactions, or other process (|ig h"1)
SPFI Methods Development
Page 52
-------
The critical data are the concentrations of emissions, air flow rate through the chamber, and
volume of the chamber. If the chamber air is well-mixed, then concentrations of emissions in the
outlet air flow are assumed to be representative of average concentrations in the chamber. If the
chamber air is not well-mixed, then samples collected from multiple locations inside the chamber
may be needed to determine average concentrations. If emissions are lost to surfaces or chemical
reactions, then additional data is needed to account for those losses.
2.12.1 Parameters for Calculation of Emission Factors from Chamber Emission
Tests
Where there are no losses to chamber surfaces or chemical reactions that impact concentrations,
the parameters needed for calculating emission rates from dynamic chamber experiments follow
from the equation above, include:
Air concentrations of emissions (Ct) determined as the mass recovered from sampling (|ig)
media divided by the sample volume (m3) where sample volume is determined as the product of
sampling flow rate (m3 h"1) and sampling time (h) and t is the midpoint between start and stop
time of sample collection.
Net volume of the chamber (V) determined from measurements of chamber interior dimensions
less the volume occupied by emissions sources and other equipment.
Air flow rate (Q), determined as the product of the chamber volume (m3), and air change rate (h~
*) where air change rate is determined by tracer decay method or determined from measurement
of the supply air flow rate.
Emission factor (EF, |ig m"2 h"1) is determined by dividing the emission rate R (|ig h"1) by the
area of the source (m2).
Source emissions models that are built into IECCU were utilized to demonstrate the use of
chamber concentration data to construct source emissions models. The concentrations
determined from the chamber experiment are overlaid with the IECCU concentration-time
predictions for selected compounds in the chamber system.
2.12.1.1 Calculation of Emission Factors During SPF Application
Calculation of emission rates and EFs during the period of application requires additional
information, that includes,
• Start and stop times for the application process.
• Area sprayed, (in each spray event if spraying is discontinuous).
• Mass of compounds emitted during the application phase of the test determined as the
sum of:
o Mass leaving the chamber system in the air exhausted from the chamber plus
mass remaining in the air at the end of the integration period determined from
chamber air concentration and air exchange rate using the trapezoid technique.
SPFI Methods Development
Page 53
-------
o Mass deposited on chamber surfaces, determined as the product of the
concentration on the surfaces (|ig m"2) and the area of the surfaces (m2).
Two empirical approaches are presented in Section 5 for constructing empirical source emission
models for the application phase. One method assumes that emissions are instantaneous and
constant (|ig h"1). This approach apportions mass of emissions to the elapsed time of each spray
event. The second approach utilizes the dual first order decay model and requires that the user
employ a least square fitting routine to estimate decay constants. The dual first order decay
approach is more complex but yields a source emissions model that better predicts chamber
concentrations following cessation of the spray event. Equations for the source models are
presented in Section 5.
It is important to recognize that multiple processes generate the concentrations of emissions in
the air and on surfaces of the chamber during and following the spray event. If the mass of
emissions determined over the entire period of Phase I are defined as the emissions due to
application, then the application phase mass emitted is a function of the length of Phase I.
Application phase emissions could also be defined as the emissions that occur during the spray
event or e.g., for a fixed number of air changes after the termination of spraying.
2.12.1.2 Calculation of Emission Factors During Phase II
The method of direct calculation of emission rates from chamber concentration data described in
ASTM D5116 is utilized to calculate EFs from the chamber concentration data that are presented
in Section 4. The relationship between emission rates (and EFs) with time is further described by
fitting a power equation of the form EF=a*t-b to the EFs calculated from the chamber
concentrations. This model can then be employed as a source model. The utility of this approach
is evaluated by comparing long-term predictions (e.g., 600 h) based upon generating the source
model from the first 72 or 96 h of the experiment.
For TCPP, a diffusion-based emission model available in IECCU is employed to demonstrate an
approach to emission prediction that does not require chamber emissions testing. This approach
utilizes the concentration of the compound in the material (Co), the material-air partition
coefficient (Kma), the diffusion rate of the chemical in the material (Dm), and the mass transfer
coefficient (/?). For purposes of demonstration, the average concentration of TCPP in the SPF
was utilized as Co, values suggested by Bevington et. al, [25] were used for Dm and K,„a, and h
was computed by the method of Sparks, the average of the air velocity measurements at the
surface of the foam as air velocity input.
2.12.2 Data Evaluation to Identify Sources of Variability and Uncertainty
The process of spraying SPF insulation generates high concentrations of reactive chemicals in
gas and particle phase in the local area of the application and the exothermic reaction between
the isocyanates and polyols releases heat into the chamber. Therefore, thermal stratification and
poor mixing may occur if fans in the chamber do not adequately mix the air. Also, emissions
may be lost to chamber surfaces due to aerosol deposition or through chemical reactions. The
SPFI Methods Development
Page 54
-------
experimental design and data analysis attempt to address key questions that may inform
variability and uncertainty associated with the test protocol.
The questions relate to potential for stratification in the chamber (1, 2) sampling issues (3, 4),
losses of flame retardant to surfaces (5), the impact of environmental conditions on emission
rates (6), and potential impacts of substrate emissions (7). Key questions include:
1. Is there evidence of thermal stratification in the chamber?
2. Is there evidence of concentration gradients in the chamber?
3. How do concentrations of isocyanates determined with denuder/filter samplers compare
to concentrations determined with impinger samplers?
4. How do concentrations of TCPP determined by sampling with modified OVS compare to
concentrations determined by sampling with multi-bed sorbent traps?
5. How much TCPP is deposited on surfaces in the chamber?
a. How variable is the distribution and how do concentrations change with time?
b. How does air velocity near the chamber surfaces impact deposition of TCPP?
c. How much TCPP is deposited on the PPE of the spraying/sampling personnel?
d. How much TCPP is removed by a residential grade HVAC filter?
6. How does chamber air temperature impact concentrations and emission rates?
7. How does the use of plywood substrate materials impact interpretation of the emissions?
2.12.3 Rationale and Approaches for Addressing Key Questions
Question 1 and 2 are important because if the chamber is not well-mixed then sampling at
multiple locations is needed to determine average concentrations of emissions. The question is
addressed by comparison of temperature - time plots at multiple elevations and locations in the
chamber, by comparison of concentrations of compounds determined for the same time periods
near and away from the area of application and comparison of concentrations of samples
collected from chamber air with concentrations determined by sampling from the air in the
exhaust duct.
Question 3 addresses concerns that denuder/filter samplers may underestimate isocyanate
concentrations during application. The question is important because impingers, considered to be
the gold standard for sampling isocyanate emissions during application of SPF, pose a risk of
solvent spillage in the chamber. The question is addressed by comparing concentrations
determined with denuder-filter samplers with concentrations determined by sampling with
impingers during spray application.
Question 4 addresses concerns that multi-bed sorbent traps may become overloaded with TCPP
and amine catalyst when sampling during spray application. The question is addressed by
SPFI Methods Development
Page 55
-------
comparing concentrations determined by sampling with OVS with concentrations determined by
sampling with multi-bed sorbent traps during spray application.
Question 5 address the significance of flame retardant deposition on characterization of
emissions during and following spray application. The relative importance of deposition of
TCPP on surfaces is determined as the percent of TCPP mass leaving the chamber system in the
air exhausted from the chamber (plus mass collected on the exhaust duct filter) compared to the
sum of TCPP mass emitted, excluding the mass of TCPP in the SPF insulation. The question is
further addressed by monitoring TCPP emissions from the chamber system following removal of
the sprayed frames from the chamber.
Question 6 addresses the impact of temperature on EFs. The exothermic reaction of isocyanates
and polyols to form polyurethane raises temperature in the chamber during the spraying process.
Also, SPF insulation installed in attics and exterior walls where air temperatures are impacted by
environmental factors. The impact of temperature on EF is determined for HFC-134a by linear
least squares fit of chamber air temperature and EFs over an unplanned three-degree temperature
rise and fall. For other compounds which lack continuous concentration data, the relationship
between temperature and EF is determined by comparison of the EF determined at the maximum
temperature of 26.8 °C with an EF predicted by power law fit of EFs over time where chamber
temperature was at 23.8 °C.
Question 7 addresses the need to select a substrate material that is relevant to normal use of the
product but does not have confounding emissions. SPF insulation is often applied to plywood
roof deck underlayment however, plywood is a known source of aldehyde emissions. The
potential for confounding emissions is evaluated by comparison of aldehydes and ketone
emissions determined with plywood substrates in the chamber at the low and high air flow rate
conditions with concentrations determined after application of the SPF.
3.0 Quality Control Results for Measurement Systems
Quality assurance (QA) and quality control (QC) procedures were implemented by following
procedures detailed in the approved Quality Assurance Project Plan (QAPP): Category B
Characterizing Emissions Daring and Following Spray Application of SPF Insulation in the
Fall-Scale Emissions Test Chamber. Quality metrics were established for sampling and analysis
systems to ensure that the data quality is adequate to meet project objectives, including;
collection of sufficient samples to determine the time-concentration profile for the emissions,
and to ensure that the concentrations of emissions in various sampling media met quality goals. It
is important to recognize that this experiment is a step in development of a consensus test
protocol for characterization of emissions during and following spray application of a reactive
mixture of chemicals. The information gained informs the next steps in the process of
development of a reliable consensus test protocol.
Information needed to characterize the test environment includes volume of the test chamber, air
change rate, temperature, relative humidity (RH), and concentrations of emissions in the air and
SPFI Methods Development
Page 56
-------
on surfaces during the experiment. Information needed to characterize the source of emissions
include the chemical makeup of the source and the physical properties of the product produced
by the spray event. Information needed to establish the relationship between the spray event and
concentrations of emissions in the test environment include time of each event, sample collection
start and stop times, and flow rates of sampling systems.
3.1 Evaluation of Chamber Performance
Prior to conducting the pilot tests, several experimental tests were conducted to verify that the
chamber system could supply 10 ACH of conditioned air (300 m3/h), and to control flow at a
minimum flow rate of 0.33 h"1 (9.9 m3 h"1). Tests were conducted to investigate optimum settings
and placement of the fan in the chamber to achieve desired air speeds near the surface of the
substrates to be sprayed. Tracer gases and particles were released near the base and between the
substrates to investigate mixing of gases and particles with the chamber mixing fan on and off.
The net volume of the chamber was estimated by estimating the volume of substrates,
equipment, and people that occupy the chamber during application. The net volume of the
chamber was also determined by the static dilution method described in the Appendix to ASTM
E741.
3.1.1 Evaluation of Environmental Control System Performance
The FSETC environmental control system is designed to control chamber air temperature, RH,
air flow, and pressure relative to the room housing the chamber. The outputs from the
temperature, RH, flow, and pressure sensors are stored as one-minute averages by the control
system computer. Table 3-1 below summarizes control system performance for the three phases
of the experiment. Prior to the test, flow control setpoints were determined by linear regression
of air change rate determined by ASTM E741 versus flow setpoint. It was determined that flow
setpoints of 62.36 (106 m3 h"1) and 6.06 (10.3 m3 h"1) cubic foot per minute corresponded to
chamber air change rates of 4 and 0.4 h"1. Observed exhaust flows as reported by the control
system in Table 3-1 averaged within 1% of target setpoints.
Table 3-1 Chamber control system performance summary
Test Phase
Temperature (°C)
Relative Humidity (%)
Exhaust Flow (m3 h_1)
Average
Standard
Deviation
Average
Standard
Deviation
Average
Standard
Deviation
Phase 1
25.3
0.5
40.4
1.9
106.0
7.1
Phase II
24.2
0.7
39.0
2.0
10.2
0.2
Phase III
23.7
0.10
39.7
6.7
10.2
0.3
3.1.2 Agilent 6850 GC/ECD to Determine Chamber Air Exchange Rate
The Agilent 6850 GC/ECD was operated to measure SFeused as a tracer gas to determine the air
exchange rate of the FSETC. The GC/ECD was calibrated from 0.067 ppm to 1.0 ppm SF6 by
injecting known volumes of SF6 into the sealed full-scale chamber. The QA/QC for the GC/ECD
SPFI Methods Development
Page 57
-------
consisted of ensuring the instrument calibration was current and using a single point span check.
An 0.800 ppm (800 ppb) SF6 calibration gas standard was used to perform single point span
checks over the course of the experiment to confirm that instrument response was within ±10 %
of the SF6 calibration standard.
The single point span check was performed on ten occasions throughout the experiment: the first
prior to the start of the experiment with the remaining spread throughout the experiment run
time. Each span check result was the average of six measurements. All span checks met the
acceptable range of ±10 %, therefore the data met the QA acceptance criteria.
3.1.3 Evaluation of Mixing of Gases and Particles
Experiments were conducted to investigate the impact of the chamber mixing fan on particle
residence time in the chamber at an air change rate of 4 h"1.
The particle mixing tests were conducted with the mixing fan on and off. In addition, gas mixing
tests were conducted alongside the particle testing. Prior to or following the smoke generated
particle injection, HFC-245fa was injected into the chamber through the same Teflon® tubing
used to inject the particles. The HFC-245fa was injected at a rate of 200 mL min"1 from 1 to 4
min (varied by test date). The gas was measured by the Innova, Model 1412 PAS Field Gas-
Monitor. The gas mixing tests were performed in part to determine time to reach concentration
equilibrium.
Results from the mixing tests indicated both particle and gas reach a well-mixed chamber
concentration within a 3-min time period with the mixing fans operational. With the fans non-
operational (OFF), the mixing period extended out towards 6 min or more.
The gas mixing tests also validated the chamber ventilation rate of 4 ACH with the fans both
operational and non-operational. This was apparent by the linear fit constants to the natural log
(In) of the gas concentration (ppm).
3.1.4 Evaluation of Physical Properties of the SPF Insulation
Physical properties are characteristics that determine the overall usefulness of the insulation
produced. They effect the ability of the foam to produce the air barrier and barriers to heat
transfer. They also can help determine if the foam produced was mixed in the correct ratios.
Physical property tests referenced in CAN/ULC-S705.2-05 were employed to evaluate depth,
density, adhesion, cohesion, cell structure, and voids for the foam [32], The details for use of the
referenced method to evaluate quality of the SPF insulation created as the emissions source are
presented in Appendix D-l.
The application proceeded normally through application to the first three substrate frames.
During application to the third quadrant of Frame 4, the spray gun operator noticed that the foam
was off-color, and the application was stopped prior to spraying the fourth quadrant. The
physical properties examination revealed that the cells of the second quadrant were small, so it is
SPFI Methods Development
Page 58
-------
likely that the application started to go off-ratio after the application to the first quadrant of
Frame 4. The foam applied to all quadrants passed the adhesion and cohesion tests. Some voids
between substrate and foam were apparent from some areas of Frame 4 where samples were
removed for cohesion tests. Density and depth were within the target specifications for the first
three substrate frames.
3.1.5 Event Timing
Timing of events was ensured by synchronizing all timing devices employed by personnel and
measurement systems collecting samples to one timing device linked to the atomic clock prior to
the start of the spray event. All start and stop times were entered in laboratory notebooks and on
sampling media labels, for transfer to the master Sample-Time-Volume (STV) spreadsheet
maintained on the common L drive. The STV spreadsheet linked all events and sample analysis
results to form a record of the experiment. The STV spreadsheet contained the start and stop
times for sampling systems, flow verifications, temperature, RH, and barometric pressure during
each sampling event. Corrections of sample volume to standard conditions were made in the
STV spreadsheet.
3.1.6 Pre and Post-test Flow Rate Checks for Sampling Systems
Air sampling to collect emissions on various sampling media and in Tedlar bags was conducted
by pulling air through sampling media using vacuum pumps and mass flow controllers or critical
orifices to obtain the desired flow rate. Data quality objectives for flow controllers are
summarized in Table 3-2. Flow rate was checked before and after each sample collected with
mass flow controllers and before and after the experiment for the critical orifice sampling
system. The percent relative standard deviation (RSD) between pre and post sample collections,
adjusted to standard conditions, is summarized in Table 3-3. In two instances the pre - post
sample volume %RSD was greater than 5%. DNPH sample 7188 had a difference of 14.8% and
OVS sample 1130 had a 6.1 %RSD. The 90% completion target was met for all sampling
systems and the two samples that did not meet <5% RSD for pre- and post-flow verification have
no impact on intended use of the data.
Table 3- 2 Data quality objectives for sampling systems flow controllers
Analytical
Instrument
Parameter
Accuracy
Precision
Completeness
Mass Flow
Controlled
Samplers
Repeatability of flow rate (pre- post-
test)/average
±5%
±5%
90%
Critical
Orifice
Samplers
Repeatability of flow rate (pre- post-
test)/average
±5%
±5%
90%
SPFI Methods Development
Page 59
-------
Table 3- 3 Summary of sampling systems uncertainty
Sampling Media
Controller
Number of
Average
Number > 5%
%
Type
Samples
%RSD
RSD
Completioness
C3
MFC*
90
0.42
0
100
OVS
MFC
55
0.76
1
98
OVS
COFC**
20
1.4
0
100
Lung Sampler
MFC
11
0.91
0
100
Tenax
MFC
4
0.16
0
100
denuder-filter™
MFC
5
0.21
0
100
denuder-filter™
COFC
25
2.5
0
100
Impinger
COFC
19
3.0
0
100
DNPH
MFC
24
0.80
1
96
*Mass Flow Controller
"Critical Orifice Flow Controller
Emissions Test Protocol QC samples consisted of background samples collected prior to the test,
field blanks (FB), spiked field controls, and duplicates. DCC samples were analyzed on each
instrument on each day of analysis. RCS samples were analyzed with each batch of OVS
desorptions/extractions, deposition samplers, and wipe sample extractions. Results of QA/QC
activities are described in the remainder of this section.
3.2 Data Quality Indicator Goals for Analytical Instruments
Data quality indicator (DQI) goals and QC checks for analytical instrumentation are summarized
in Table 3-4.
SPFI Methods Development
Page 60
-------
Table 3- 4 Data quality goals and QC checks for analytical instruments
Analytical
Instrument
Parameter
Accuracy
Precision
Completeness
TD-GC/MS a
Repeatability of Internal Standard (IS) response
NA
± 25%
90%
Calibration - relative standard deviation (RSD)
triplicate analyses
na
<25%
RSD
90%
Calibration - linear regression, R2
NA
<0.99
90%
DCC - all target compounds
± 25%
NA
90%
Independent Audit Program (IAP)
± 25%
± 25%
90%
DI-GC/MS b
Calibration - relative standard deviation (RSD)
triplicate analyses
NA
<25%
RSD
90%
Calibration - linear regression, R2
NA
<0.99
90%
DCC - all target compounds
± 25%
NA
90%
Recovery Check Standard (RCS) recovery
± 50%
50%
90%
Independent Audit Program (IAP)
± 25%
± 25%
90%
LC-MS/MS
Calibration correlation for target compounds
NA
R2> 0.95
90%
Pre- and post-calibration slope factor difference
NA
< 20%
90%
Performance Evaluation Samples (PES),
low/medium/high concentrations
< 30% Er
d
NA
90%
HPLC/DAD e
Calibration - RSD triplicate analyses
NA
<20%
RSD
90%
DCC - all target compounds
± 15%
NA
90%
PES
NA
<25%
RSD
90%
IAP
± 25%
± 25%
90%
a Thermal Desorption
b Direct Injection of Liquid Sample Preparations (OVS tubes, wipe samples,
deposition samples, PPE extracts, HVAC extracts)
c Percent difference between pre-and post-calibration [(pre-post)/pre] must be <
0.2.
d Percent Relative Error
e Diode-array detector
3.3 Instrument Calibrations, lAPs, IDLs, DCCs
The results for the calibrations, IAP comparison samples, calculated IDL, and DCCs of the
GC/MS systems utilized for this project are grouped together by instrument to allow a more
direct comparison of the performance of the instrument.
3.3.1 LC-MS/MS for Quantification of MDI, p3-MDI, and p4-MDI
The IDL and instrument quantification limits (IQLs) of the LC-MS/MS system were determined
prior to the experiment per Appendix B to Part 136 Code of Federal Regulations, Title 40,
Volume 24, Revision 1.11. The IDL and IQLs (ng) shown in Table 3-5 have been divided by the
minimum and maximum air sample volumes for the denuder-filter samplers and the impinger
samplers to provide estimates of minimum method detection limits (MDLs) and method
SPFI Methods Development
Page 61
-------
quantification limits (MQLs) for the samples collected during the application and curing phase of
the experiment with each sampling media. Note that these detection and quantification thresholds
represent the sensitivity of the instrument and do not include the uncertainties introduced by
handling and processing sampling media in the environment of the test chamber during spray
application.
Table 3- 5 Minimum detectable and quantifiable levels of isocyanates (hg m 3)
Sampler Type
Denuder-filter (D-F)
Impinger
D-F & Impinger
Sample Volume (Liters)
1.9
1.9
9.5
9.5
19.7
19.7
IDL
IQL
MDL
MQL
MDL
MQL
MDL
MQL
ng
ng
(|jg m"3)
(|jg m"3)
(|jg m"3)
(|jg m"3)
(|jg m"3)
(|jg m"3)
MDI
1.5
4.7
0.8
2.5
0.2
0.5
0.08
0.24
p3-MDI
0.9
2.7
0.5
1.4
0.1
0.3
0.04
0.14
p4-MDI
0.3
0.9
0.2
0.5
0.03
0.1
0.02
0.05
3.3.1.1 Calibration
The LC-MS/MS was calibrated per the procedure specified in ISO 17734a. A certified stock
solution (Supelco CRM40603 XA14439V) containing 4,4'-MDI, p3-MDI, and p4-MDI at
concentrations of 174.8, 90.1, and 33.8 |ig mL"1, was diluted 1:100 to produce working stocks
that were further diluted to generate calibration solutions at eight concentration levels ranging
from 1.7 to 437, 0.9 to 225, and 0.3 to 85 ng for MDI, p3-MDI, and p4-MDI, respectively. An
IS mix (Supelco CRM40604 XA18139V) with concentrations of 180.8, 94.2 and 35.6 |ag mL 1
was diluted 1:100 to create the IS working stock. The standards are DBA-deuterated derivatives
of the MDI parent compound, MDI-DBA-d9, p3-MDI-d9, and p4-MDI-d9.
Quality control procedures for calibration require that slope of the pre- and post-calibration runs
agree within 20% and the R2 for the fits of the relative response factors (RRFs) must be 0.95 or
better. If these criteria are met, the calibration data are combined into one calibration file for data
reduction. Calibrations were run for analysis of (1) impinger samplers and for (2) diluted
impinger samplers where analyte mass was above the calibration range, (3) chamber background
(Bkg) samples where denuder and filter were extracted together, (4) denuders, and (5) filters
where denuders and filters from each sampler were extracted and analyzed separately.
As seen in Table 3-6, all calibrations met the criteria that pre- and post-slope factors be within
20%, and correlation of the slope factor, R2>0.95.
SPFI Methods Development
Page 62
-------
Table 3- 6 Combined pre and post run linear calibrations results
Impinger Samplers
Combined Pre and Post Run Calibrations
Compound
Slope
Intercept
R2
(Pre/Post)
p3-MDI
101.44
-5.07
0.9955
92%
p4-MDI
19.59
-0.42
0.9977
101%
MDI
110.61
-13.34
0.9973
97%
Impinger-dilution*
Combir
ed Pre and
Post Run
Calibrations
Compound
Slope
Intercept
R2
(Pre/Post)
p3-MDI
89.95
1.06
0.9959
96%
p4-MDI
14.65
0.52
0.9959
107%
MDI
139.14
-15.90
0.9971
99%
Denuder + Filter Bkg**
Combined Pre and Post Run Calibrations
Compound
Slope
Intercept
R2
(Pre/Post)
p3-MDI
88.63
0.80
0.9987
99%
p4-MDI
17.76
-0.86
0.9961
111%
MDI
132.55
-12.22
0.9980
100%
Denuder***
Combined Pre and Post Run Calibrations
Compound
Slope
Intercept
R2
(Pre/Post)
p3-MDI
53.69
-4.66
0.9964
94%
p4-MDI
7.91
0.28
0.9916
94%
MDI
67.95
-17.61
0.9978
103%
Filter***
Combined Pre and Post Run Calibrations
Compound
Slope
Intercept
R2
(Pre/Post)
p3-MDI
49.15
-3.31
0.9983
100%
p4-MDI
7.88
-0.13
0.99723
98%
MDI
65.46
-14.94
0.9978
101%
*Calibration factors for impinger samples with that were diluted and re-analyzed.
**Calibration for chamber background samples where denuder and filter were extracted
together.
***Denotes calibrations where denuder and filter were analyzed separately.
Other QC procedures include analyses of reagent blanks, field blanks, duplicate samples, and
analysis of denuder-filter samplers and impinger samplers spiked by an independent analyst.
3.3.1.2 Field Blanks
Two denuder-filter sampler field blanks and one impinger field blank were collected. A denuder-
filter sampler field blank was collected when the chamber/substrate background samples were
collected at the high and low air flow rates. One denuder-filter and one impinger sampler were
placed on Sample Tree 2 prior to the spray event and collected 1.5 h later to investigate potential
SPFI Methods Development
Page 63
-------
for contamination when samplers are handled in the chamber by the spraying/sampling
personnel.
Results were quite variable. The denuder-filter sampler collected with the chamber background
samples was non detect for MDI and 4-MDI but had 4 ng of p3-MDI. The impinger sampler
placed in the chamber on Sample Tree 2 during the spray event had no detectable MDI but had
2.9 and 4.9 ng p3- and p4-MDI, respectively. The denuder-filter sampler attached to Sample Tree
2 during the spray event appeared to be seriously contaminated on the filter and denuder
sections. The filter section had 33, 12 and 3.3 ng, and the denuder section had 36, 14, and 4.6 ng
MDI, p3-MDI, and p4-MDI, respectively. It is not possible to draw conclusions from a single
field blank; however, the results indicate a need for further investigation and suggest that the
denuder-filter samplers should be managed from outside of the chamber by personnel that are not
directly exposed to the product emissions during application.
3.3.1.3 Chamber Background Samples
Duplicate six-liter samples were collected with denuder-filter samplers at the low and high air
change rates prior to the start of the test. Samplers were placed into the chamber through Port
CT-2 located in the ceiling of the chamber. A field trip blank was also collected. The results
were essentially non-detects except for p3-MDI on the chamber background and field blank. For
this compound, the amount on the denuder and filter of the duplicate pairs (1.78±0.14 ng) was
very similar to the amount determined on the field blank (1.4 ng). Therefore, there is no evidence
of MDI or pMDI in the chamber background air with the substrates in the chamber.
3.3.1.4 Duplicate Sample Results
Five sets of duplicate impinger samples and four sets of duplicate denuder-filter samplers were
collected with quantifiable concentrations of MDI, p3-MDI, and p4-MDI. The relative percent
difference (RPD) was calculated for each MDI compound for each pair. Results are summarized
in Tables 3-7 and 3-8. As can be seen in Table 3-7, the RPD between the MDI compounds for
four of the five duplicate pairs was within the ±25% quality assurance acceptance criteria for the
samples collected with impingers. RPD for two of the three compounds, did not meet the quality
assurance target of ±25%, however sums of MDI compound concentrations for the duplicate pair
were 5.3 and 3.0 |ig m"3, so the increased variability is consistent with lower precision at low
concentrations. Overall data capture rate for the impinger samplers was 85%.
It is clear from the duplicate sampler results presented in Table 3-9, that the denuder-filter
sampler results do not meet quality assurance target of ±25% RPD for duplicate samples. Seven
of sixteen results met the acceptance criteria. There may be multiple reasons for the failure to
meet QA criteria. The high level of contamination on the denuder-filter sampler field blank
provides evidence that the sample handling process failed to prevent contamination of the
samplers. This variability was not reflected in the recovery of the mid and high-level denuder-
filter sampler blind spikes. Therefore, it is likely that the variability is due to sampler handling
procedures in the chamber during the application/curing phase.
SPFI Methods Development
Page 64
-------
Table 3- 7 Relative percent difference duplicate impinger samplers
MDI
p3-MDI
p4-MDI
Sum MDI
(hours)
Location
RSD (%)
RSD (%)
RSD (%)
RSD (%)
0.08
Tree 1
12%
19%
20%
15%
0.08
Tree 2
1%
8%
16%
4%
0.31
Tree 1
6%
4%
4%
3%
0.31
Tree 2
12%
3%
2%
9%
0.95
Tree 1
39%
129%
21%
56%
Average
14%
33%
12%
18%
Table 3- 8 Relative percent difference duplicate denuder-filter samplers
(hours)
Location
MDI
p3-MDI
p4-MDI
Sum
MDI
0.07
Tree 1
3%
25%
25%
2%
0.07
Tree 2
14%
26%
29%
19%
0.31
Tree 1
41%
45%
48%
43%
0.31
Tree 2
58%
37%
52%
45%
Average
29%
33%
38%
27%
3.3.1.5 Challenge Sample Results
Challenge samples at high, mid, and low concentrations were prepared by spiking impingers and
denuder-filter samplers. The LC-MS/MS instrument operator was not informed of the amount
spiked to the sampling media. The summary results are provided in Table 3-10. Twelve of 18, or
67% of the analyses met the acceptance criteria of ±30% of true value. Two of the low-level
denuder samplers failed (MDI, and p3-MDI) and all three of the p4-MDI impingers failed. There
does not seem to be consistency in the analyses of the challenge samples except that 50% of the
six failures were at the low-level spike, two were at the mid-level and one was at the high level.
P4-MDI failed at all three levels in the impingers and passed for all three of the denuder
samplers. It is possible that a spiking error occurred with the mid-level impinger spikes as the
MDI result was more than twice the expected amount and the mid-level p4-MDI analysis did not
have a response. In contrast to the instrument detection and quantification limits determined prior
to the experiment, the results of the blind spike samples indicate that the analyses were not
consistently reliable for samples containing MDI compounds in range of <7 ng. The implication
of this is that we are not able to quantify MDI concentrations less than 1 |ig m"3 and this limits
our ability to determine how long emissions persist after spray application. Changes to the
sample handling protocols during sample collection to minimize contamination of sampling
media and possibly improvements to sample handling techniques in the laboratory will be
necessary to reduce uncertainties at low concentration levels.
SPFI Methods Development
Page 65
-------
Table 3- 9 Recovery of isocyanate blind spikes
Impinger Spikes
Denuder-filter Spikes
MDI
p3MDI
p4MDI
MDI
p3MDI
p4MDI
Spike Level 1 (ng)
5.7
2.9
1.1
6.6
3.4
1.3
Recovery (ng)
5.3
2.9
0.1
ND
ND
1.1
%Recovery
102%
97%
9%
0%
0%
82%
MDI
p3MDI
p4MDI
MDI
p3MDI
p4MDI
Spike Level 2 (ng)
27.8
14.8
5.5
32.0
16.5
6.2
Recovery (ng)
80.6
15.3
ND
30.4
14.1
6.0
%Recovery
290%
103%
0%
95%
85%
95%
MDI
p3MDI
p4MDI
MDI
p3MDI
p4MDI
Spike Level 3 (ng)
317
163
61.3
247
129
48.5
Recovery (ng)
317
147
36.1
226
116
40.5
%Recovery
100%
90%
59%
91%
90%
83%
3.3.2 Quantification of HFC 134a with the Photoacoustic Spectrophotometer (PAS)
The PAS was calibrated by dilution of certified gas standards with zero air. As seen in Figures 3-
1,3-2, and 3-3, response was linear through 200 ppm and decreased dramatically between 2000
and 5000 ppm.
_ 12
i.10 y = 0.9166X + 0.0366 L''''
-S 8 R! = 0.9999
-------
3000
_ 2000
E
Q.
Q.
8 1500
c
o
o_
^ 1000
v = -O.OOOlx2 + 1.0229X
R2 = 0.9996
4.9988
.V
o ••
0 1000 2000 3000 4000 5000 6000
Calibration Gas Concentration (ppm)
Figure 3-3 Extended Calibration of the PAS
3.3.2.1 Calibration of the Innova Model 1314 Photoacoustic Spectrophotometer (PAS)
The response becomes flat as the instrument response approaches 2500 ppm. The instrument
reported concentrations above the 2500 ppm for six samples between 0.36 and 0.43 h. Those data
were flagged and were not used in calculation of mass emitted. The low-level linear calibration
factor was used to process HFC-134a concentrations below 12 ppm because of the improved
zero intercept for the reduced calibration range. The mid4evel linear calibration was used to
process instrument responses between 12 and 200 ppm and the quadratic calibration was used to
reduce instrument responses between 200 and 2500 ppm. Clearly, additional standards were
needed to better characterize instrument response to standard concentrations between 2000 and
3000 ppm. A diluter that brought the HFC-134a concentrations below 2000 ppm would also have
greatly improved confidence in the determination of emissions concentrations during the
application. Data capture rate was 121/127, or 95% completeness.
3.3.2.2 Zero/Span Checks for the PAS
Quality control for the Innova included periodic zero checks and span checks with an 11.8 ±5%
ppm certified HFC 134a standard and 99.9 ±2% ppm certified HFC 245fa standard. Summary
results are shown in Table 3-10. Average response of the calibration check samples was within
uncertainty range for each of the calibration check standard gases.
Table 3-10. Zero - span checks for the PAS
Calibration Gas
ppm
Accuracy (%)
Lower Limit
Upper Limit
Average
Standard
Deviation
% Recovery
Bias (%)
Number of
Measurements
Zero Air
-0.03
0.04
22
HFC 134a
11.11
0.32
94.1
5.9
32
HFC 245fa
96.56
2.83
96.7
3.3
16
SPFI Methods Development
Page 67
-------
3.3.2.3 QC Results for the Bag Dilution System
A dilution bag was filled with 11.77 ppm calibration gas. The bag was analyzed on each day that
other bags were diluted and analyzed. The average ratio of response to true value was 0.92 with
RSD of 1.8% over a 24-hour period. Up to eight replicate measurements of HFC-134a were
collected from bags on two or three successive days. Percent RSD for multiple analyses from bag
dilutions averaged 1.8% and ranged from 0.3 to 4.8%. One bag sample was run multiple times
with 10:1 dilution and straight from the bag. The ratio of straight from the bag to 10:1 dilution
was 10.05. These measurements indicate that the HFC-134a was stable in the bags and that the
dilutions were accurate.
3.3.3 Quantification ofVOCs, Amine Catalyst, and TCPP by TD-GC/MS Analysis
A Markes International TD-100/ Agilent 7890A/5975C GC/MS was used for thermal desorption
and quantification of analytes collected on the multi-bed sorbent traps. QA/QC procedures for
the TD-GC/MS system include calibration of target compounds with serial dilutions made from
certified standards over a minimum of six calibration levels, calibration acceptance criteria for
replicate analyses at each calibration level, calibration verification by meeting the acceptance
criteria for analysis of an independently prepared standard made from a second certified source
(IAP), acceptance criteria for DCC samples analyzed with each set of samples, acceptance
criteria for duplicate samples collected simultaneously from the chamber or exhaust, and
acceptance criteria for cleaned sorbent sampling traps. IDLs were determined for each target
analyte by seven replicate analyses of a low-level standard.
3.3.3.1 Calibration
Dilutions from the stock solutions were created from stock solutions listed in Table 3-11 and the
instrument was calibrated by analysis of thermal desorption traps loaded by injecting 2 ul of
standard onto traps using a commercial loading system to produce six concentration levels across
the calibration range shown in Table 3-12.
Table 3-11. Stock solutions for td-gcms standards
Individual Stock Solutions:
Analytes
Formula
CAS#
Concentration (|jg
mL1)
tris(1 -Chlor-2-propyl)
phosphate
C9H18CI3PO4
13674-84-5
9,742
TCPP Isomer 1
3,798
TCPP Isomer 2
594
N, N, N', N", N"-Pentamethyl-
diethylenetriamine
C9H23N3
3030-47-5
50,024
Chlorobenzene
CeHsCI
108-90-7
10,090
1,2-Dichloropropane
CsHeCb
78-87-5
10,071
1,4-Dioxane
C4H8O2
123-91-1
10,010
3-Chloro-1-propene (Allyl
chloride)
C3H5CI
107-05-1
10,050
Triethyl phosphate
C6H12O4P
78-40-0
10,003
2-methyl-2-pentenal
C6H10O
623-36-9
10,010
SPFI Methods Development
Page 68
-------
3.3.3.2 Minimum Instrument Detection Levels for the TD-GCMS System
The DDLs and instrument quantitation limits (IQLs) are presented in Table 3-12. DDLs and IQLs
were determined as three standard deviations of the response from analysis of seven low-level
standards for VOCs and TCPP, TCPP isomer 1, and TCPP isomer 2 respectively.
Table 3-12. Instrument calibration summary for the TD-GCMS system
Instrument Calibration Summary
Calibration Range
Fit
rA2
Compound ID
Minimum
Maximum
IDL*
IQL**
(ng)
(ng)
(ng)
(ng)
3-Chloro-1-propene
10
1000
Q
0.999
2.4
7.9
12-DCP
10
1007
Q
0.9998
0.8
2.8
1,4-Dioxane
10
1000
Q
0.9998
1.3
4.4
2-Methyl-2-pentenal
10
1000
P
0.9941
2.7
8.9
Chlorobenzene
10
1000
Q
0.9999
1.3
4.2
PMDTA
250
3001
Q
0.9994
73.3
244
TCPP
10.1
974
Q
0.9986
1.2
4.1
TCPP-11
3.9
380
Q
0.9995
0.7
2.5
TCPP-I2
0.62
59.4
Q
0.9994
1.5
4.9
*IDL is the instrument detection level determined as 3 times the standard deviation of seven
replicate analyses of a low-level standard.
**IQL is the estimated instrument quantitation level determined as 10 times the standard
deviation of seven replicate analyses of a low-level standard.
3.3.3.2 Method Detection and Quantitation Limits for the TD-GCMS System
Minimum detectable levels determined as three standard deviations of the response from analysis
of seven low-level standards for VOCs and TCPP, TCPP isomer 1, and TCPP isomer 2
respectively are shown in Table 3-13. MDLs and quantifiable concentrations for sample volumes
of 0.1 through 10 L are provided in Table 3-14.
Table 3-13. Minimum detectable levels for the TD-GCMS system
Chemical Name
Method Detection Limit (|jg m~3)
Sample Volume (L)
0.1
0.5
1
5
10
3-Chloro-1-propene
24
12
2.4
0.12
0.24
1,2-Dichloropropane
8
4
0.8
0.16
0.08
14-Dioxane
13
6.5
1.3
0.26
0.13
2-methyl-2-pentenal
27
13.5
2.7
0.54
0.27
Chlorobenzene
13
6.5
1.3
0.26
0.13
N,N,N',N",N"-
Pentamethyldiethylenetriamine
733
147
73.3
14.6
7.3
tris(1-Chlor-2-propyl) phosphate
12
6
1.2
0.24
0.12
TCPP Isomer 1
7
3.5
0.7
0.14
0.07
TCPP Isomer 2
15
3
1.5
0
0.15
SPFI Methods Development
Page 69
-------
Table 3-14. Minimum instrument quantification concentrations for the TD-GCMS system
Chemical Name
Method Quantification Limit (|jc
rrr3)
Sample Volume (L)
0.1
0.5
1
5
10
3-Chloro-1-propene
79
16
7.9
1.6
0.8
1,2-Dichloropropane
28
5.5
2.8
1.6
0.3
1,4-Dioxane
44
8.8
4.4
0.9
0.4
2-methyl-2-pentenal
89
18
8.9
1.8
0.9
Chlorobenzene
42
8.5
4.2
0.8
0.4
N,N,N',N",N"-
Pentamethyldiethylenetriamine
2440
490
240
49
24
Tris(1-chloro-2-propyl) phosphate
41
8.3
4.1
0.8
0.4
TCPP isomer 1
25
4.9
2.5
0.5
0.2
TCPP isomer 2
49
9.9
4.9
1.0
0.5
3.3.3.3 Daily Calibration Check Samples
A total of 26 DCC samples were analyzed. Recoveries of all compounds met the 90%
completeness goal except for 2-methyl-2-pentenal, PMDETA, and TCPP-2. 2-methyl-2-pentenal
failed on 16 occasions, PMDETA failed on four occasions and TCPP-2 failed on three occasions.
Phase II and Phase III multi-bed sorbent samples were collected as duplicates. Where one of the
duplicates was analyzed on a day with DCC failure, and the result was compared to the other
sample run on a day with passing DCCs, the average of the ratios across nine reported
compounds was 1,03±0.10. There is no evidence that the DCC failure for PMDETA and TCPP-2
represented a problem with the analysis system. The PMDETA and TCPP-2 data are considered
reliable and the 2-methyl-2-pentenal data were flagged as unreliable and not reported. DCC
results are summarized in Table 3-15.
Table 3-15 Daily calibration control sample recovery for the TD-GCMS system
Average
Recovery
Number
Pass
Number
Fail
Completeness
(%)
%RSD
(%)
3-Chloro-1-propene
88.9%
11%
25
1
96
12-DCP
97.4%
7%
26
0
100
1,4-Dioxane
97.6%
6%
26
0
100
2-Methyl-2-pentenal
76.3%
20%
10
16
38
Chlorobenzene
97.7%
6%
26
0
100
PMDTA
91.1%
13%
22
4
85
TCPP
99.6%
6%
26
0
100
TCPP-11
98.3%
6%
26
0
100
TCPP-I2
84.3%
32%
23
3
88
3.3.3.4 Internal Standard Response
The IS was automatically injected onto every sorbent trap prior to the thermal desorption step.
One mL of d8-toluene certified National Specialty Gases at 43.8 ppm (163.6 ng) was injected.
Repeatability criteria was ±25%. The IS variability, determined as the %RSD averaged 5.6%,
11%, and 0.9%, for samples run for samples analyzed over a four-day period.
SPFI Methods Development
Page 70
-------
3.3.3.5 Calibration Level 3 Sample Storage Test
Spiked storage samples were prepared when it became clear that there would be a delay in
analysis of the study samples due to malfunction of the TD-GC/MS. Nine samples were
prepared: three blanks, three spiked at a low level, and three at a mid-range level. Multi-bed
sorbent traps were spiked at levels of 50 and 400 ng per trap except for PMDETA, which was
spiked at 250 and 1000 ng. The samples were capped and placed in Tedlar® bags and stored in a
freezer at -20 °C along with the study samples. Samples were removed from the freezer with
study samples and brought to room temperature prior to analysis. Recoveries of the spikes were
acceptable for all compounds. Therefore, it was determined that storage of the samples during
the 28-day period had no apparent impact on the data quality. The details for spiking the multi-
bed sorbent traps at low and mid-levels are presented in Tables 3-16 and 3-17.
Table 3-16 Low-level multi-bed storage samples
Calibration Level 2A:
CAL02A
Prepared in 10 mL of Methanol
Analyte
Cal Concentration
(ng/ML)
Mass on Tube (ng) 2 pL
spiked
PMDETA
125.1
250.1
Calibration Level 2B:
CAL02B
Prepared in 10 mL of Methanol
Analyte
Cal Concentration
(ng/pL)
Mass on Tube (ng) 2 pL
spiked
Chlorobenzene
25.2
50.4
1,2-Dicholoropropane
25.2
50.4
1,4-Dioxane
25.0
50.0
3-Chloro-1-propene
25.0
50.0
2-Methyl-2-pentenal
25.0
50.1
TCPP
30.20
60.40
TCPP iso 1
11.77
23.55
TCPP iso 2
1.84
3.69
Table 3-17 Mid-level multi-bed storage sample spikes
Calibration Level 4A:
CAL04A
Prepared in 10 mL of Methanol
Analyte
Cal Concentration (ng/pL)
Mass on Tube (ng)
2 pL spiked
PMDETA
500.2
1,000.5
Calibration Level 4B:
CAL04B
Prepared in 10 mL of Methanol
Analyte
Volume (pL)
Cal Concentration (ng/pL)
Mass on Tube (ng)
2 pL spiked
Chlorobenzene
2
202
404
1-2-Dichloropropene
2
201
402
14-Dioxane
2
200
400
3-Chloro-1-propene
2
200
400
2-Methyl-2-pentenal
2
200
400
TCPP
2
195
390
TCPP iso 1
2
76
152
TCPP iso 2
2
12
24
SPFI Methods Development
Page 71
-------
The results from analysis of the spiked storage samples are summarized in Table 3-18. The
average low-level spike for TCPP isomer 2 was 84% , which was within acceptable range
although the spiking level of 3.7 ng was actually below the quantitation level of 4.9 ng. Recovery
ranged from 57 to 118% for the three analyses of the low-level TCPP-2 spiked storage traps. The
PMDETA recovery at the low-spike level ranged from 120 to 138% recovery for three samples,
indicating bias high. The results are consistent with the increased uncertainty of PMDETA at the
lower end of the calibration range.
Table 3-18. Summary results for the spiked multi-bed storage samples
C3 Spiked Sorbent Storage
Study
Low Concentration Level
High Concentration Level
Average
Standard
Deviation
n
Average
Standard
Deviation
n
3-Chloro-1-propene
110%
9%
3
101%
9%
3
1,2-Dichloropropane
114%
4%
3
97%
5%
3
1,4-Dioxane
110%
3%
3
96%
5%
3
2-Methyl-2-pentenal
80%
3%
3
118%
22%
3
Chlorobenzene
109%
4%
3
98%
5%
3
PMDETA
129%
9%
3
103%
7%
3
TCPP
97%
14%
3
98%
6%
3
TCPP isomer 1
93%
14%
3
97%
6%
3
TCPP isomer 2
*BQL (84%)
31%
90%
5%
3
*Spike level of 3.7 ng was below the minimum quantitation level of 4.9 ng.
3.3.3.6 TD-GC/MS System Background
Carryover was observed in system blanks that were run immediately after each sample.
However, no carryover was observed when a second system blank was run so it was determined
that running one system blank after each analysis prevents sample to sample carryover. A total of
166 system blanks were run over the course of the six days of instrument operation.
3.3.3.7 Field Blank Summary
In 13 field blanks, TCPP was detected but below the quantification limit (10 ng) in four samples,
detected at the quantification limit on two samples, and was not detected on other samples. 1,4-
Dioxane was detected below the limit of quantification (10 ng) in two field blanks and was not
detected in any others. No corrections to the data were made based upon the field blanks.
3.3.3.8 Chamber Blank Summary
Chamber background samples were collected twice at high and low flow conditions utilized in
the test protocol. In each case, the TCPP amounts on the chamber background samples were
below the quantification level and similar or less than the amounts observed on the field blanks.
No corrections were made to the data based upon the chamber background samples.
SPFI Methods Development
Page 72
-------
3.3.3.9 Duplicate Thermal Desorption Sample Summary
Six duplicate pair of multi-bed sorbent traps were collected from the exhaust duct sampling port
and 16 sets were collected from the chamber side port. Acceptance criteria for duplicate samples
of ±25% RPD was exceeded by two compounds (3-chloro-l-propene and TCPP-2) in one
exhaust duct sample. RPD ranged from 0.3 to 31.6% for compounds with concentrations above
the minimum quantitation level. Averages for individual compounds ranged from 0.6 to 10.2%
and averaged 3.9% across the compounds quantified by TD-GC/MS. The RPD for the 16
duplicate pairs collected from the side port of the chamber ranged from 0.8 to 10.4% and
averaged 3.1% for valid samples. One duplicate pair was disqualified from the analyses due to
apparent failure of the thermal desorption system during analysis. Overall data capture rate for
duplicate pairs, based upon the number of individual compound results at concentrations above
the minimum quantitation level meeting the acceptance criteria was 99% or 92% if one includes
the sample that failed during analysis.
3.3.4 GC/MS System for Analysis of Extracts from Sampling Media
A Hewlett-Packard 6890 Plus/5973 GC/MS with a HP 6890 Series injector was used for liquid
injection of sample desorption, extractions, and dilutions. QA/QC procedures for the GC/MS
system included calibration of target compounds with serial dilutions made from certified
standards over a minimum of six calibration levels, calibration acceptance criteria for replicate
analyses at each calibration level, calibration verification by meeting the acceptance criteria for
analysis of an independently prepared standard made from a second certified source (IAP),
acceptance criteria for DCC samples analyzed with each set of samples, acceptance criteria for
duplicate samples collected simultaneously from the chamber, and acceptance criteria for
recovery of the RCS. Instrument detection limits were determined for each target analyte by
seven replicate analyses of a low-level standard.
3.3.4.1 Calibrations
The instrument was calibrated with dilutions made from certified standards at nine levels from
10 to 507 ng mL"1 for the flame retardant TCPP and at nine levels from 125 to 500 ng mL"1 for
the RCS d2i-tripropyl phosphate. The instrument was also calibrated for a second RCS, d27-
tributyl phosphate, however it was not used in the evaluation of the data quality due to an
interferent in some of the extracts. The standards were run in triplicate at each level. The RSD at
each level averaged 5.1% and 3.1% for TCPP and d2i-TPP, respectively with r2 of 0.9975 and
0.9987 for the quadratic fit of the data. DCC samples were run each day of analyses. Calibration
ranges and r2 for the quadratic fit for each calibration standard are shown in Table 3-19.
Table 3-19 GC/MS calibration range and r2 for the quadratic fit
Analvtes
Calibration Ranae
ng mL"1
£
Tripropylphosphate-chi (RCS1)
12.5-500
0.999
Tributylphosphate-d27 (RCS2)
12.4-496
0.997
PMDETA
250 - 3000
0.997
TCPP (total)
14.7-735
0.999
SPFI Methods Development
Page 73
-------
3.3.4.2 Daily Calibration Check Samples
Summary results for analysis of DCCs are presented in Table 3-21. Some issues were observed
with analysis of the DCCs. PMDETA, TBP-L, and TCPP did not meet acceptance criteria on all
samples. Deposition, XAD, and wipe extracts analyzed during the period of DCC failures were
re-analyzed utilizing the new DCC solution. The data are suitable for the intended purpose.
Table 3- 20 Daily calibration check sample recovery for the GC/MS
Analyte
Concentration
in DCC
Number
DCCs
% Pass
Notes
PMDETA (ng/mL):
1,000.5
36
69
All pass with new DCC
standard
TPP-L (RCS) (ng/mL):
250.0
36
100
TBP-L (RCS) (ng/mL):
248.2
36
75
Not suitable as RCS*
TCPP Concentration (ng/mL):
294.0
36
83
All pass with new DCC
standard
* Seven TBP-L fail after remake of DCC standard solution.
3.3.4.3 Recovery Check Sample Analyses
RCS was added to each sample prior to workup to validate recovery through extraction,
concentration, and dilution steps. Although the spiking solution contained (RCS1) tripropyl
phosphate-d2i, and (RCS2) tributyl phophase-d27, only recovery of RCS1 was used in data
validation due to analytical issues with RCS2. The acceptance criteria for the RCS was set to
100±50% due to the exploratory nature of the analyses. For each media sample type, the RCS
was analyzed to verify stability over the course of the analysis period. Results are presented in
Table 3-21.
Table 3- 21 Recovery check standard summary results
Tripropyl Phosphate-chi (RCS1)
Sample Type
Average RCS
% Recovery
% RSD
Number
Pass
% Completion
Deposition
122%
16%
60
100
OVS
125%
16%
213
96
Tyvek®
126%
16%
14
88
Wipes
114%
21%
112
96
HVAC Filters
121%
18%
22
100*
*Five samples from the HVAC filter in-place during the application period required 100-fold
dilution and this diluted the RCS below the level of quantitation (12.5 ng mL"1).
Of 421 analyses, 12 failed to meet the 100±50% criteria for RCS1. Most of these samples had
RCS1 concentrations at or below the minimum quantifiable level due to sample dilution. The
overall data capture rate was 97%. The data set met the criteria for recovery of the RCS.
SPFI Methods Development
Page 74
-------
3.3.5 HPLC-DAD Analysis System for Aldehydes
The HPLC was calibrated with certified standards. QC procedures included analysis of system
blanks, DCCs, chamber/substrate and clean air background samples, duplicate samples, and field
blanks. Results for the QC procedures are presented in the following subsections.
3.3.5.1 Calibrations
The HPLC was calibrated at six concentrations over the range of 0.03 to 15 |ig mL"1 from serial
dilutions of the 15 |ig mL"1 calibration stock solution CRM4M7285, lot number XA20836V
(Supelco) that contains DNPH derivatives of a suite of 16 aldehydes and ketones. Summary
results of the instrument calibration showing average and standard deviation of the retention
time, slope factor, and correlation coefficient for each compound are provided in Table 3-22.
Table 3- 22 Summary of DNPH-aldehyde calibration results
Summary of Response Factors
Retention Time
Aldehyde or Ketone
Minutes
Standard
Deviation
Slope
R2
Formaldehyde-DNPH
3.37
0.003
0.002791
0.9998
Acetaldehyde-DNPH
4.28
0.003
0.003587
0.9999
Acrolein-DNPH
5.49
0.006
0.004222
0.9999
Acetone-DNPH
5.65
0.006
0.004643
0.9999
Propionaldehyde-DNPH
6.20
0.006
0.004682
0.9998
Crotonaldehyde-DNPH
7.57
0.008
0.005591
0.9998
n-Butyraldehyde-DNPH
8.91
0.013
0.005851
0.9998
Benzaldehyde-DNPH
10.04
0.014
0.008134
0.9998
Isovaleraldehyde-DNPH
11.66
0.011
0.007952
0.9999
Valeraldehyde-DNPH
12.09
0.011
0.007155
0.9999
o-Tolualdehyde-DNPH
12.56
0.012
0.009737
0.9999
m&p-Tolualdehyde-DNPH
12.84
0.010
0.009586
0.9999
Hexaldehyde-DNPH
14.54
0.082
0.008179
0.9998
2,5-Dimethylbenzaldehyde-DNPH
14.63
0.096
0.011266
0.9999
3.3.5.2 HPLC Instrument Detection Levels
The DDL established by analysis of seven replicate low-level standards per 40 CFRPart 136 are
presented in Table 3-23. MDLs have been calculated for sample volumes collected in the course
of the experiment.
SPFI Methods Development
Page 75
-------
Table 3- 23. Instrument detection limits and minimum detection limits for aldehydes
Compound
IDL
MDL
Sample Volume (L)
50
30
15
7.5
M9 mL"1
M9 m"3
M9 m"3
M9 m"3
M9 m"3
Formaldehyde
0.007
0.7
1.2
2.3
4.7
Acetaldehyde
0.01
1
1.7
3.3
6.7
Acrolein
0.01
1
1.7
3.3
6.7
Acetone
0.011
1.1
1.8
3.7
7.3
Propionaldehyde
0.011
1.1
1.8
3.7
7.3
Crotonaldehyde
0.02
2
3.3
6.7
13.3
n-Butyraldehyde
0.019
1.9
3.2
6.3
12.7
Benzaldehyde
0.047
4.7
7.8
15.7
31.3
Isovaleraldehyde
0.024
2.4
4.0
8.0
16.0
Valeraldehyde
0.029
2.9
4.8
9.7
19.3
o-Tolualdehyde
0.02
2
3.3
6.7
13.3
m&p-Tolualdehyde
0.012
1.2
2.0
4.0
8.0
Hexaldehyde
0.023
2.3
3.8
7.7
15.3
2,5-Dimethylbenzaldehyde
0.032
3.2
5.3
10.7
21.3
3.3.5.3 System Blanks
No peaks were detected for any of the analytes for two of three system blanks. Propionaldehyde
(0.02 |ig mL"1), n-butyraldehyde (0.04 mL"1), benzaldehyde (0.01 |ig mL"1), and hexaldehyde
(0.03 |ig mL"1) were reported at levels near the DDL for the third system blank. Average
concentrations for the three analyses are therefore below the DDL and no corrections were made
to the data.
3.3.5.4 Daily Calibration Check Sample Results
Summary results for the aldehyde and ketone DCCs are presented in Table 3-24. All compounds
met the 100±25% acceptance criteria so percent completion for analysis of the DCCs is 100%.
SPFI Methods Development
Page 76
-------
Table 3- 24. Daily calibration check summary results for aldehydes
Concentration (|jg ml_~1)
Percent Recovery
Compound
-
DCC-
1
DCC-2
-
Formaldehyde-DNPH
1.53
102%
102%
Acetaldehyde-DNPH
1.53
102%
102%
Acrolein-DNPH
1.53
102%
102%
Acetone-DNPH
1.52
101%
101%
Propionaldehyde-DNPH
1.53
101%
102%
Crotonaldehyde-DNPH
1.53
100%
102%
n-Butyraldehyde-DNPH
1.54
105%
103%
Benzaldehyde-DNPH
1.52
101%
101%
Isovaleraldehyde-DNPH
1.52
101%
101%
Valeraldehyde-DNPH
1.51
101%
101%
o-Tolualdehyde-DNPH
1.50
101%
100%
m&p-Tolualdehyde-
DNPH
3.00
101%
100%
Hexaldehyde-DNPH
1.55
103%
103%
2,5-
Dimethylbenzaldehyde-
DNPH
1.50
99%
100%
3.3.5.5 Field Blanks
A field blank was collected with low and high air change rate background samples collected at -
1054 hours and a second was collected at -21.5 hours. The field blanks contained 0.2 and 0.3
|ig/sample acetaldehyde, respectively, and 0.4 and 0.1 |ig/sample acetone, respectively. No other
DNPH field blanks were collected.
3.3.5.6 Chamber Background Samples
Chamber background samples were collected in duplicate at the high and low flow conditions
with substrates in the chamber. The chamber background samples clearly indicate that the
substrates and/or foil lining are a source of several aldehydes and a ketone, as shown in Table 3-
25. It is not clear that all the emissions have the same source. Concentrations of formaldehyde
and hexaldehyde appear to scale directly with the 10:1 change in chamber flow whereas
concentrations of the other compounds appear to scale at2:lto5:l.A121L sample was
collected from the clean air supply at a port upstream of the chamber. Formaldehyde,
acetaldehyde, and acetone were detected at concentrations of 0.7, 3.2, and 0.9 |ig m"3
respectively. It is not clear from this one sample that the clean air system is a source of these
compounds since acetaldehyde and acetone were detected in the field blanks. The data indicate
that further investigation is warranted to identify the sources of the emissions. It may indicate
that the permanganate filter that removes formaldehyde from the air supply is due for
replacement.
SPFI Methods Development
Page 77
-------
Table 3- 25 Chamber/substrate background aldehyde concentrations
Compound
*MDL (|jg mL"1)
MDL (|jg m-3)
Concentration (pg rrr3) ±%**RSD
60 L Sample
Volume
@0.38 ACH
@3.8 ACH
Formaldehyde
0.007
0.58
9.8±18%
<0.58
Acetaldehyde
0.010
0.83
1.0±29%
***nd
Acetone
0.011
0.92
12.7±15%
ND
Propionaldehyde
0.011
0.92
3.2±2%
1.3±141%
Isovaleraldehyde
0.024
2.00
<2.0
ND
Valeraldehyde
0.029
2.42
13.5±15%
<2.4
Hexaldehyde
0.023
1.92
54.4±16%
5.1 ±43%
2,5-Dimethylbenzaldehyde
0.032
2.67
ND
ND
* Minimum Detectable Level (MDL) for 60 Liter air samples. Extracted in 5 mL of acetonitrile.
** %RSD = relative standard deviation expressed as % of average of duplicate samples
*** ND = Not Detected
3.3.5.7 Duplicate Emissions Test Sample Results
Samples collected at elapsed times of 4.9 and 24 hours were collected in duplicate. Results are
shown in Table 3-26. Formaldehyde did not meet the ±25% RSD criteria for one of the two
samples and 2,5-dimethylbenzaldehyde did not meet the criteria for either duplicate sample.
However, 2,5-dimethylbenzaldehyde concentration was below the MDL for three of the four
samples. For compounds above the MDL, one of 11 determinations failed to meet the criteria,
therefore the overall data capture rate for duplicate samples is 91%, which meets the 90% target.
Table 3- 26. Duplicate pair results for aldehyde samples
Duplicate Sample ID
7178 & 7179
7181 & 7182
Elapsed Time (h)
4.9
24
Sample Volume (Liters)
30
50
MDL
Average
RPD
MDL
Average
RPD
Compound
pg nr3
pg nr3
(%)
pg nr3
pg nr3
(%)
Formaldehyde
1.2
8.7
55%
0.6
7.2
2%
Acetaldehyde
1.7
27.2
6%
0.8
17.3
7%
Acetone
1.8
22.6
9%
0.9
9
3%
Propionaldehyde
1.8
9.8
2%
0.9
5.9
3%
Benzaldehyde
7.8
ND
NA
3.9
ND
NA
Isovaleraldehyde
4.0
ND
NA
2.0
8.1
15%
Valeraldehyde
4.8
ND
NA
2.4
D (2.0)
13%
Hexaldehyde
3.8
11.5
12%
1.9
10.3
15%
2,5-Dimethylbenzaldehyde
5.3
D (2.2)
48%
2.7
5.9
140%
SPFI Methods Development
Page 78
-------
3.3.6 APS and ELPI Particle Instrumentation
The quality assurance checks discussed below are recommended by the instrument manufacturer.
3.3.6.1 ELPI QA Checks
The pre- and post-sampling HEPA filter samples and the ambient air samples serve as instrument
performance checks. The HEPA filtered samples confirm there are no leaks in the sample train
and that a "zero" instrument measurement is attainable. With a HEPA filter installed on the APS
sample inlet, the total particle count must be < 1 particle/cm3. The ELPI must have a current
value (noise level) within ± 25 fA in the 400,000 fA operating range with a HEPA filtered
sample inlet. If an ELPI channel (stage) continues to have current values outside of the allowable
range after multiple zeroing attempts, that channel will not be considered acceptable for use and
its data will not be included in analysis. The pre- and post-ambient samples should be
comparable in size distribution and concentration for each instrument given no significant
change has occurred to the environment between these two sample times. If any significant
deviation is observed, this could indicate a baseline shift or response failure of the instrument.
The data will require review to determine what, if any, corrective actions may be necessary.
Zeroing of the ELPI was performed to adjust the bias current of the instruments 12 electrometers
prior to the experiment's high and low air exchange rate test parameters. Post zeroing, the ELPI
met the acceptable current noise level of ± 25 fA in the 400,000 fA operating range with a HEPA
filtered sample inlet. The pre- and post-ambient sample comparison for size distribution and
concentration was performed by sampling the ambient background air within the Butler
Building. The background samples were averaged and compared for their size distribution and
concentration. No significant difference was discerned from the ambient samples.
3.3.6.2 APS QA Checks
The APS instrument was coupled to a 20:1 aerosol diluter during the application phase of the
experiment. Previous SPF experiments indicated the upper concentration range of the APS was
exceeded due to the high particle concentration anticipated during the application phase of the
experiment, a 20:1 diluter was used with the APS during the spray event due to high particle
concentrations generated during the application phase of the experiment. With the HEPA filter
attached to the diluter inlet, there was no measurable particle concentration throughout the
instrument's measurement range. With the HEPA filter removed, sporadic particle measurements
were detected from the ambient background of the room housing the full-scale. The summed
total particle count for the APS remained well below the required limit of < 1 particle cm"3.
Follow-on measurements from the room housing the full-scale chamber indicated a particle
distribution consisting of particles mostly <3 |j,m with a total concentration in the low single
digits.
SPFI Methods Development
Page 79
-------
4.0 Results
The presentation of results is organized by (1) characterization of the SPF insulation emissions
source, (2) summary of environmental conditions during each phase, (3) concentrations of target
emissions in the air and on surfaces during each phase. The results provide the information
needed to generate emissions source terms, presented in Section 5, for isocyanates, the blowing
agent, TCPP, and VOCs. Source emissions terms generated for isoycanates do not account for
losses due to chemical reactions in the air or losses to surfaces due to depostion.
The results for the flame retardant TCPP include mass deposited on surfaces, concentration of
TCPP in the SPF insulation, and emissions from the chamber system following removal of the
sprayed frames. The results for depostion and emission of TCPP from surfaces are presented to
inform understanding of the impact of secondary emissions on TCPP concentrations in the air.
Results also include air velocity information needed to estimate a mass transfer coefficient in
support of demonstration of diffusion-based emissions modeling for TCPP.
4.1 Characterization of the SPF Insulation Emissions Source
SPF insulation was sprayed on to four plywood substrate frames. Each 1.9 m2 frame was divided
into four equal quadrants creating 16 quadrants. The application was terminated after application
to 15 quandrants resulting in an application area of 7.1 m2 . The average foam depth, determined
from four measurements per quadrnat, was 4.6±0.4 cm (average±standard deviation) for the
foam applied to the first three frames. The average depth for the fourth frame where the
application appeared to go off-ratio during application to the second quadrant, was 2.8±1.0 cm.
4.1.1 Application Time and Rate
The first two frames were sprayed in 0.125 hours and the second two frames were sprayed in
0.110 hours. Application rate averaged 30.5±1.9 m2 h"1 for the four frames. The time to spray 7.1
m2 SPF insulation to the four panels, not including the time spent collecting air samples between
spray events, was 0.235 h.
4.1.2 Amount of Foam Applied
The total amount of foam reagent sprayed from the two-component low pressure kit, determined
by difference between initial and final weight of the kit was 12.1 kg including the foam sprayed
into the waste bucket at the start of the experiment to prime the transfer lines. The total amount
of foam applied to the substrates, estimated as the sum of the mass applied to the four frames
was 10.6 kg.
4.1.3 Characteristics of the SPF Insulation
Uniform cells with a few larger cells were observed for all quadrants of Frames 1, 2, and 3. Cells
in the second quadrant of Frame 4 were not uniform with multiple larger cells and inclusions and
the foam was not attached to the substrate at the location of the cohesion test in quadrant 2. The
SPFI Methods Development
Page 80
-------
cells of quadrant 3 were uniform with small and very small cells with no inclusions. The physical
properties examination indicates that the foam application began to go off-ratio during
application to the second quadrant of Frame 4.
4.1.4 Temperature of the Foam During Curing
Maximum temperatures determined as the average of three mid-depth temperature probes in each
frame ranged from 96 to 110 °C for Frames 1, 2, and 3 whereas the average maximum
temperature for Frame 4 was 73 °C. Mid-depth temperature of the foam in all four frames
averaged less than 40 °C within one hour after initiation of spraying.
4.1.5 Amount and Variability of TCPP Concentrations in the SPF Insulation
The mean weight percent TCPP for the eleven cores taken from all locations except Frame 4,
quadrant 3 was 12.7±1.8% (mean ± standard deviation) whereas the mean percent TCPP for
three samples collected from Frame 4, quadrant 3 was 6.1±2.7%. The efficiency of the extraction
is not known because the 1:5000 dilutions needed to bring the concentrations in the extracts
within the calibration range of the GC/MS diluted the RCS below the level of detection. The
total amount of TCPP in the foam applied to the substrates is estimated to be 1.33 kg.
4.2 Characterization of the Chamber Environmental Conditions
Averages and variability of air temperature, RH, ACH, and air velocity near the surfaces of the
SPF insulation are presented for Phases I and II. The air velocity measurements were made at the
end of the test following placement of the sprayed frames back into the chamber at their original
positions.
4.2.1 Air Temperature and Relative Humidity
The temperature and RH results reported in Sections 4.2.1.1 and 4.2.1.2 are those recorded by
the OPTO 22 chamber control system. The temperatures reported in Section 4.2.1.3 were
recorded by temperature sensors hung at either end of the chamber and recorded by the HOBO
data loggers.
4.2.1.1 Phase I Chamber Air Temperature and Relative Humidity
During Phase I, air temperature in the chamber reported by the chamber operating system during
application increased rapidly from 23.8 to 30.5 °C and decreased to less than 26 °C within one
hour. The air temperature decreased to 25 °C when the personnel exited the chamber at the end
of Phase I. RH averaged 40.6±1.6% during Phase I.
4.2.1.2 Phase II and III Chamber Air Temperature and Relative Humidity
Chamber air temperature averaged 24.2±0.6 °C (n=2E+04) and 23.7±0.1 °C (n=9E+03)
(average±standard deviation) for Phase II and III, respectively. RH averaged 39±2% and
39.7±6.7% for Phases II and III, respectively.
SPFI Methods Development
Page 81
-------
4.2.1.3 Spatial Variability of Air Temperatures During Phase I
During the spray event, air temperature in the chamber varied by elevation and distance from the
sprayed frames by as much as 2 °C. As seen in Figure 4-1, temperature differences by elevation
were less than 0.5 °C within one hour near the sprayed frames.
Figure 4-1 Air temperatures at four elevations near the sprayed frames.
Air temperature differences by elevation were smaller at the opposite end of the chamber,
reaching a maximum of less than 27 °C at an elevation of 210 cm above the floor. Near the
center of the chamber, air temperatures at the lowest elevation appeared to run about one degree
lower than seen at the lowest elevation in Figure 4-1.
4.2.2 Air Change Rates
Chamber ACH determined by tracer decay method (ASTM E741) are presented in the following
two sections.
4.2.2.1 Phase I Air Change Rate
The chamber air change rate determined by tracer decay during Phase I averaged 4.1±0.06 h"1
(average ± standard deviation, n=4). The concentration of the tracer compound decreased from
7E+03 to 5E+03 ug m"3 when the personnel exited the chamber at the end of Phase I, indicating
loss of up to 30% of the tracer compound from the chamber during the 4-minute period of egress
when the door was not fully closed.
SPFI Methods Development
Page 82
-------
4.2.2.2 Phase II and III Air Change Rates
ACH determined by tracer decay averaged 0.40±0.005 h"1 (n=27) and 0.40±0.006 h"1 (n=15)
(average ± standard deviation) for Phase II and III, respectively.
4.2.3 Air Speed Near Surface of Sprayed Frames
Air velocity and turbulence near the surface of an emissions source impact the emissions
characteristics of the source, particularly for "wet" sources where high concentrations of
compounds in the boundary layer may retard mass transfer from the source through the boundary
layer to the bulk air of the room. The air velocity and turbulence near the boundary layer impacts
the mass transfer coefficient (/?), a key variable in physically-based models such as the diffusion-
based source emissions models (Equation 13, Section 5.1.3.2.2). Pre and post-test air speed
measurements were collected 1 cm above the surfaces of the substrates to document the
variability of air movements near the source and provide data for calculation of the mass transfer
coefficient (/?) for use in Equation 13.
4.2.3.1 Surface Air Speed Characterization of Sprayed Frames for Phase I
Post-test average air speeds determined 1 cm above the surface of SPF insulation for quadrants 2
and 3 for each frame ranged from 0.07±0.03 m sec"1 to 0.22±0.03 m sec"1 with 21 to 28
measurements in each quadrant. The measurements were conducted with the chamber air change
rate of 4.1 h"1 and mixing fans on. The average air speed determined as the average of the
measurements for each of the eight quadrants was 0.12±0.05 m sec"1.
4.2.3.2 Surface Air Speed Characterization of Sprayed Panels for Phase II
Post-test air speed measurements taken with chamber operating at 0.4 h"1 at central locations of
quadrants 2 and 3 of each frame ranged from 0.04±0.02 to 0.25±0.06 m sec"1. The average air
speed determined as the average of the measurements for each of the eight quadrants was
0.14±0.08 m sec"1.
4.2.4 Air Speeds Near Surface of Deposition Samplers
Air speed was measured 1 cm above the arrays of the deposition samplers at the seven locations
shown in Figure 2-3 at the completion of the test with chamber air change rate of 4.1 h"1.
Averages of 11 to 16 measurements per location, ranged from 0.14 to 0.89 m sec"1 with overall
average of 0.38±0.28 m sec"1. The highest speed readings occurred at the right and left wall
positions and at the door wall, consistent with the positioning of the mixing fans that directed air
flow towards the back wall and diagonally across the chamber toward the frames.
4.3 Characterization of Concentrations of Emissions
The results are organized by class of compound in the order of (1) isocyanates, (2) blowing
agent, (3) flame retardant, (4) amine catalyst, (5) VOCs, (6) aldehydes and ketones, and (7)
particle size and number. The results for each class of compound are further ordered by phase.
SPFI Methods Development
Page 83
-------
The emissions concentration characterization includes: (1) measurements of concentrations of
emissions determined in the air of the chamber, (2) exhaust duct, and (3) concentrations of the
flame retardant determined on various surfaces during each phase. Concentrations of the amine
catalyst, PMDETA, are also reported for extractions from the HVAC filters and PPE.
4.3.1 Phase I Concentrations of MDI, p3-MDI and p4-MDI
Impinger and denuder-filter samples were collected frequently over the first two hours to capture
the rapidly changing isocyanate concentrations during and following the spray event. Duplicate
samples were collected for each media at each of two sampling locations in the chamber for the
first two sampling events. For the samples collected during the spray application, RPD between
duplicate impinger samplers was 14.7, 4.4, 3.3, and 9.4 % for the four impinger pairs of samplers
compared to 8, 19, 43, and 45% for the denuder/filter sampler pairs. The large RPD between
duplicates for two of the sets of denuder/filter samplers may be indicative of the contamination
observed on the field blank (sum MDI104 ng) and may reflect the potential for sample
contamination where the spray/sampling personnel handle samplers inside the chamber during
and immediately following spray application. Average concentrations for samples collected
during the spray events are presented for each sampler type in Table 4-1 for the sum of MDI, p3-
MDI, and p4-MDI.
Table 4-1 Summary comparison of impinger samplers to denuder-filter samplers for samples collected
DURING THE APPLICATION EVENTS.
Sampler
Type
Impinger/Filter*
Denuder/Filter
Notes
Elapsed time
(hours)
Average
ftjg nr3)
Standard
Deviation
(n=4)
Average
(pg nr3)
Standard
Deviation
(n=4)
*lmpinger samples were diluted and
rerun due to mass amounts above the
highest calibration level
0.07/0.08
532
42
536
175
Denuder/Filter: Impinger ratio: 1.1
0.31
416
93
486
133
Denuder/Filter: Impinger ratio: 1.3
As can be seen in Table 4-1, the concentrations, averaged for the four samplers of each type at
the two locations in the chamber were similar, however the variability was larger for the
denuder-filter samplers, as seen in Figures 4-2 and 4-3. The impinger sampling results suggest
that the concentrations of isocyanates were higher at Sampling Tree 1 nearer the area of
application than at the location of Sampling Tree 2, further from the area of application. The
denuder/filter sampler results are much more variable and suggest higher concentrations during
application at the more distant sampling location. The error bars in Figures 4-2 and 4-3 represent
the average of the RPD between duplicate samples collected from Tree 1 and Tree 2. The RSD
(average ± standard deviation) was 18±22% for 5 pairs of impingers samplers and 33±16% for 6
pairs of denuder-filter samplers. The RPD for the impinger duplicate pair collected at 0.95 h was
SPFI Methods Development
Page 84
-------
56%. The large RPD for this pair is attributed to the greater variability at low concentrations
(5.3, 3.0 ng m"3).
• Sampling Tree 1 A Sampling Tree 2
700
ST~ 600
£
5® 500
1 400
2 300
4->
§ 200
S 100
0
0.0 0.5 1.0 1.5
Elapsed Time (hours)
Figure 4- 2 Sum of MDI, P3-MDI, and P4-MDI determined by sampling with impingers.
~ Sampling Tree 1 o Sampling Tree 2
1000
\ 800
DO
f 600
o
2 400 <~
£Z
CD
= 200
o
o
0
0.0 0.5 1.0 1.5 2.0
Elapsed Time (hours)
Figure 4- 3 Sum MDI, P3-MDI, and P4-MDI determined by sampling with denuder-filter samplers.
11 1 1
1—~—II—•—1
A A
SPFI Methods Development
Page 85
-------
The experiment was designed to inform the question of how long MDI emissions persist after
application by collection of several samples over a two-hour period following the application.
Although the denuder-filter™ samplers appear to indicate quantifiable emissions for more than
an hour following the end of the application, a sampler that was placed in the chamber on
Sampling Tree 1 for 1.5 h as a field blank with end caps in place was contaminated with 104 ng
of sum MDI (Filter: 33, 12, and 3 ng, Denuder: 37, 14, 5 ng MDI, p3-MDI, p4-MDI
respectively). Therefore, the denuder/filter sampler results are not considered reliable beyond the
application phase. The same issue was not observed with the impinger sampler utilized as a field
blank. It appears that concentrations of sum MDI were quantifiable for at least one half hour
following the completion of spraying in the chamber.
For the denuder/filter samplers, the percent of isocyanates collected on the filter to total collected
by the sampler was 55±15, 54±18, and 53±19%, indicating that approximately half of the MDI
and p-MDI mass was in the particle phase during application. Note that the denuder/filter
samplers were placed in horizontal position with the inlet facing the direction of the area of
application for use as an area sampler. The normal position of the denuder-filter sampler would
be vertical with the inlet down if attached to the lapel of a worker.
4.3.2 Phase I and II Chamber Air Concentrations of HFC-134a
The PAS provided near real-time concentrations of the blowing agent, HFC-134a. Exhaust duct
air concentrations were quantified during Phase I and II and fell below quantifiable range after
the frames were removed from the chamber at the start of Phase III.
4.3.2.1 Phase I Exhaust Duct Air Concentrations of HFC-134a
Use of the PAS instrument for characterization of emissions during application was considered
exploratory due to the large HFC-134a concentration range expected during the application
process. To address this limitation (1) whole air samples were periodically collected in Tedlar®
bags from the exhaust duct during the period where high concentrations were expected for
subsequent dilution and (2) the instrument was calibrated beyond the factory calibration range as
described in Section 3.3.2.
The six highest HFC-134a concentration values collected over a five-minute period were above
the extended calibration range of the PAS and are not plotted in Figure 4-4. Dilutions of air
samples collected in Tedlar® bag samples plotted in the figure demonstrate agreement between
the diluted bag samples and the output of the PAS. However, it is clear more bag samples were
needed during the period of spraying to provide reliable concentrations in the region where
concentrations exceeded the calibration. A key sample collected during the application was lost
due to failure of the sample bag valve. Therefore, the data did not capture the peak
concentrations of the HFC-134a emissions during the spray application process which exceeded
18E+06 |ig m"3. Concentrations of HFC -134a decreased to about 60E+03 |ig m"3 within 3 h
following completion of spraying.
SPFI Methods Development
Page 86
-------
2.0E+07
1.8E+07
ST 1.6E+07
1.4E+07
DC
— 1.2E+07
£Z
~ 1.0E+07
fU
¦£ 8.0E+06
£ 6.0E+06
Q 4.0E+06
2.0E+06
O.OE+OO
Figure 4- 4. HFC-134a Concentration-time profile during and following spray application*
* Concentrations between 0.33 and 0.46 h are not shown as they were above the extended calibration range of the
photoacoustic gas monitor. Uncertainty of concentrations above 1.2E+07 |ig m 3 is increased due to the nonlinear
calibration.
4.3.2.2 Phase II Exhaust Duct Air Concentrations of HFC-134a
Phase II exhaust duct concentrations are presented in Figure 4-5. The HFC-134a exhaust duct air
concentration peaked at 7.4 h which is about 4 h after the chamber air change rate was changed
from 4 to 0.4 h"1. Although not apparent in Figure 4-5 due to the long time period of the elapsed
time axis, concentrations remained at near steady-state conditions between 6.3 and 8.4 h,
averaging 1.4E+05±1.2E+03 |ig m"3 (average ± standard deviation). The concentration decreased
by half in about 12 h, and then decreased from 1.8E+04 to 4E+03 |ig m"3 over the period of 100
to 670 h. As shown on the secondary Y axis in Figure 4-5, HFC-134a concentration increased
from 5.5E+03 to 8.1E+03 |ig m"3 when the chamber air temperature increased from 23.8 to 26.8
°C, starting at 440 h.
• •
• •
Olnnova
Lsi
i v
i
® Cp ft ~ Ratr 1
eaky valve
o "
§ *
o Increased Calibration
Uncertainty
V
o
3^
.0 0.5 1.0 1.5 2.0
Elapsed Time (hours)
SPFI Methods Development
Page 87
-------
1.6E+05
sr 1.4E+05
E 1.2E+05
DO
— 1.0E+05
~ 8.0E+04
£ 6.0E+04
§ 4.0E+04
° 2.0E+04
O.OE+OO
Elapsed Time (hours)
Figure 4- 5; HFC134A Phase II concentrations
4.3.3 Phase I, II, and III Air, Surface, and Material TCPP Concentrations
Data are presented that characterize TCPP air concentrations determined in the chamber and
exhaust duct during and following the application at the high flow rate, during the extended
emissions characterization period at the low flow rate and following removal of the treated
frames from the chamber. Data are presented that characterize TCPP concentrations on the
exhaust duct filters, deposition on walls, floor, and ceiling, mixing fans, and sampling support
stands determined by deposition and wipe samplers. Deposition of TCPP on the PPE of the
spraying/sampling personnel was determined from extractions of fabric samples collected from
the PPE worn by the applicator and helper. The amount of TCPP deposited as overshoot was
estimated from the mass of SPF nodules collected from the chamber floor at the end of the test.
The data paint a detailed picture of the TCPP emissions.
4.3.3.1 TCPP Concentrations Determined in the Air of the Chamber and Exhaust Duct
Concentrations of TCPP in the air were determined by sampling on multi-bed sorbent traps for
analysis by TD-GC/MS and by sampling with modified OVS for determination of the gas-
particle distribution of TCPP. The multi-bed sorbent traps were collected by pulling air through
sampling cartridges placed into the chamber through the side port of the chamber and into the
exhaust duct port. The modified OVS samplers were collected in Phase I with samplers placed
on the sampling trees and during Phase II by placing samplers into the chamber through ceiling
ports CT-4 and CT-2, located at opposite ends of the chamber.
HFC 134a
' HFC34a {response to temperature change)
y
A
j
M
H
4
J
I
~
V
V
\
\
i
t
9.E+03
o
sr
8.E+03
o
£
i
DO
a
03
_£Z
6.E+03
O
CO
4.E+03
cz
a;
cv
CL
u
c
£
3.E+03
o
u
O)
4—»
100 200 300 400 500 600 700
SPFI Methods Development
Page 88
-------
4.3.3.1.1 PHASE I TCPP CONCENTRATIONS IN THE CHAMBER AND EXHAUST DUCT AIR DETERMINED WITH
MULTI-BED SORBENT TRAPS AND TD-GC/MS.
TCPP concentrations in the air of the chamber and chamber exhaust duct determined by TD-
GC/MS of multi-bed sorbent traps are presented in Figure 4-6. Maximum observed
concentrations in the air of the chamber peaked at about 3000 |ig m"3 and decreased rapidly
following cessation of spraying. Concentrations in the air of the exhaust duct appeared to peak at
about 800 |ig m"3 and appeared to be higher than the concentrations in the chamber after about
one hour. For the Phase I samples collected between 1 and 3.2 h, concentrations determined by
sampling through the side port of the chamber averaged 130±30 |ig m"3, whereas concentrations
determined by sampling from the exhaust duct port averaged 276±87 |ig m"3.
O Chamber A Exhaust Duct 3 Chamber (100 mL grab samples)
4000
3500
3000
m
£
oo
=L
2500
c
o
4->
03
2000
4—1
£Z
Q)
U
c
1500
o
u
1000
500
0
<
)
T <
>
\ j
>—
i %
* 4)
2
* i i
© ©
© *
Q
A
0.0
0.5
1.0
2.5
3.0
3.5
1.5 2.0
Elapsed Time (hours)
Figure 4- 6. TCPP chamber and exhaust duct air concentrations during Phase I determined by TD-GC/MS.
4.3.3.1.2 PHASE I TCPP CONCENTRATIONS IN THE CHAMBER DETERMINED WITH MODIFIED OVS
OVS samples were collected to provide insight into potential concentration gradients in the
chamber and insight into the gas-particle distribution of TCPP. Nine sets of OVS samples were
collected during Phase I, including three sets of duplicate pairs and eight sets of samples where
OVS samplers located near the area of application (Sample Tree 1) and away from area of
application (Sample Tree 2) were collected simultaneously.
The OVS results were characterized by variability. The variability was observed in the
distribution of TCPP mass collected on the Teflon filters, XAD-2 resin bed, and breakthrough to
the backup XAD resin portion of many of the samplers. The RPD of TCPP (sum of the filter,
primary and backup XAD-2 resin beds) between duplicate samples averaged 38±18% for three
SPFI Methods Development
Page 89
-------
duplicates collected from the same sampling location during Phase I. The OVS sample extracts
for samples collected during Phase I required dilutions ranging from 5 to 25 times to bring
analyte concentration within the calibration range of the GC/MS. Many samples required
reanalysis due to failure to meet the 100±35% recovery criteria for the d-27 tributyl phosphate
RCS and all except two of the samples failed to meet the RCS acceptance criteria for analysis of
all three portions of the sampler. Therefore, the Phase I OVS sample results failed to meet
quality assurance acceptance criteria. The concentrations and % mass collected on the Teflon
filter presented in Figure 4-7 illustrate the variability and suggest that 30 to 50% or more of the
emissions are in aerosol during the spray event.
3500
3000
O
E 2500
2000
1500
1000
500
0
OTreel ATree2 0%Particle
w>
3
c
o
c
(L>
u
c
o
(J
0.0
<
c
>
)
<
S
\
r
.
1
©
t
§
&
1 $
0.5
1.0 1.5 2.0
Elapsed Time (hours)
2.5
70%
60%
50% „
tA
-------
|ig m"3 and then decreased to 65 |ig m"3 following repair of the high bay facility air handler and
return of chamber air temperature to setpoint conditions. The increased TCPP air concentrations
were due to increased emissions from the foam and potentially from chamber and exhaust duct
surfaces.
m
E
CUD
c
_o
4—'
(0
4—'
c
(L)
o
c
o
U
O
160
140
120
100
80
60
40
20
0
A TCPP Exhaust Duct Chamber Air Temperature
27.5
27.0
200 300 400 500 600 700
Elapsed Time (hours)
TCPP Chamber
Figure 4- 8 Phase II TCPP Concentrations in the chamber and exhaust duct determined by TD-GC/MS.
4.3.3.1.4 PHASE II TCPP CONCENTRATIONS DETERMINED IN THE CHAMBER AIR WITH MODIFIED OVS
Over the 28-day period of Phase II, 24 modified OVS were collected by placing samplers one
meter into the chamber through ceiling ports CT-4 (near the source) and CT-2 (away). Five sets
of duplicate samplers were collected by sampling on two OVS from the same port. Nine sets of
samples were collected by sampling simultaneously from CT-4 and CT-2. Dilutions of 2 to 10
times were required for some of the samples and some samples required reanalysis due to failure
to meet the RCS 100±35% acceptance criteria.
For the Phase II modified OVS, TCPP was mostly observed on the primary XAD-2 resin bed,
with <1% on the filter and <0.5% on the backup XAD-2 resin bed. From 26 h until the end of
Phase II, TCPP concentrations determined by sampling near the source declined from 88 to 38
|ig m"3. The ratio of concentrations determined near the source to concentrations determined
away from the source ranged from 0.96 to 1.25 for four sets of samples where all samplers met
RCS acceptance criteria. The concentration ratio for three of the four sets of valid samples
averaged 1.24±0.01, indicating that concentrations may have been higher close to the sprayed
frames, however five sets of samplers did not meet acceptance criteria, so the data are not
SPFI Methods Development
Page 91
-------
considered sufficiently reliable for drawing conclusions regarding potential TCPP concentration
gradients in the chamber during Phase II.
4.3.3.1.5 PHASE III TCPP CONCENTRATIONS IN THE CHAMBER AND EXHAUST DUCT AIR DETERMINED
WITH MULTI-BED SORBENT TRAPS AND TD-GC/MS ANALYSIS
Phase III began with removal of the sprayed frames from the chamber and replacement of the
exhaust duct HVAC filter at 670 h. Chamber and exhaust duct air concentrations were monitored
through 958 h. As seen in Figure 4-9, TCPP concentrations determined by TD-GC/MS declined
over that time from 61 to 11 |ig m"3 in the chamber air and from 59 to 9.6 |ig m"3 in the exhaust
duct, respectively. Five hours after the sprayed frames were removed from the chamber the
concentration in the chamber was almost twice the concentration in the exhaust duct. The
differences in concentration between the chamber air and exhaust duct at 675 h were due in part
to uptake of TCPP by the new HVAC filter that was placed over the exhaust duct opening after
removal of the sprayed frames from the chamber. Concentrations of TCPP determined in the
chamber air and exhaust duct air were 17.0 and 16.5 |ig m"3 at 790 h, respectively indicating that
the filter was not effective in reducing TCPP air concentrations within 120 h of replacement.
OTCPP - Chamber A TCPP - Exhaust Duct
80
70
E
60
ClO
3
50
c
o
-------
ratio of concentrations of samples meeting the RCS acceptance criteria determined at CT-4
relative to CT-2 was 1.02 with a range of 0.9 to 1.26. TCPP was not detected on the Teflon filter
or backup XAD-2 resin bed on any of the samplers. TCPP concentrations determined with the
OVS at 675 h were 21.5 compared to 42 |ig m"3 for the thermal desorption samples. TCPP
concentrations determined with modified OVS appeared to change little from 10 |ig m"3 after 793
h. As with the Phase II OVS samples, the results indicate a need to further evaluate their use for
characterizing TCPP emissions from SPF insulation.
4.3.3.2 TCPP and PMDETA Deposited on Non-target Surfaces
Concentrations of TCPP were determined with the deposition samplers placed on walls, ceiling,
and floor of the chamber at the end of each phase, and with wipe samples for sampling support
stands, mixing fans, and surfaces in the chamber behind the sprayed frames at the end of Phase
III. Concentrations of TCPP and PMDETA were determined for the PPE worn by the applicator
and helper at the end of Phase I and at the end of each phase for the HVAC filters that covered
the opening between the chamber and exhaust duct.
4.3.3.2.1 CONCENTRATIONS OF TCPP DETERMINED WITH DEPOSITION SAMPLERS
Results for the deposition samplers are presented in Table 4-2. Concentrations for floor and
ceiling represent the average of two samples collected at the end of each phase whereas the wall
averages are the averaged results for deposition samplers placed on the left wall, right wall, and
door wall. The Floor 1 and 2 samples were collected from the samplers placed at the bases of the
sampling support stands (see Figure 2-2). Floor 1 was positioned just behind the sprayer on the
centerline of the floor and Floor 2 was positioned further away, placing the samplers at roughly
one-third and two-thirds of the length of the floor. Ceiling 1 and 2 were also positioned along the
centerline of the chamber, however, Ceiling 2 was the location closer to the spray event. When
the frames were removed from the chamber at the end of Phase II, deposition samplers were
placed behind the area that had been covered by the frames.
Table 4- 2 TCPP concentrations at deposition sampler locations
Location
Floor 1
Floor 2
Ceiling 1
Ceiling 2
*Wall (L, R, D)
***Behind
Frames
M9 m"2
M9 m"2
MS m 2
M9 m 2
M9 m 2
M9 m"2
Phase I
16.4E+03
6.6E+03
0.33E+03
0.64E+03
0.88E+03
Phase II
12.2E+03
5.4E+03
0.64E+03
0.70E+03
0.80E+03
Phase III
10.6E+03
5.9E+03
0.33E+03
0.33E+03
0.43E+03
0.42E+03
*Wall (L, R, D) average concentration for deposition samplers from left, right, and door wall samplers.
"""Average relative percent difference for 26 duplicate samples 24±14%
***Average of concentrations determined from depostion samplers placed behind the four frames when the frames
were removed from the chamber at the end of Phase II.
SPFI Methods Development
Page 93
-------
As can be seen by inspection of the data in Table 4-2, (1) TCPP concentrations were much
higher on the floor relative to the ceiling and walls, (2) floor concentrations were higher near the
area of application, and (3) floor concentrations near the area of application decreased during
each phase. It also appears that TCPP concentration increased in the area behind the frames after
they were removed from the chamber.
When the frames were removed at 670 h, two unexposed deposition samplers were placed on the
wall area that was previously covered by each frame. The samplers were collected at the end of
the test to provide insight into sorption by previously unexposed chamber surfaces. The
concentrations averaged 204±16 |ig m"2for the six samplers deployed at positions behind Frames
2, 3, and 4. Only one valid Phase III sample for the Frame 1 location was collected and the
concentration determined from that sample was 856 |ig m"2 Since this value is more than 10
standard deviations above the six samples collected from the areas occupied by Frames 2, 3, and
4, it may indicate contamination upon collection. In any case, the deposition samplers placed at
the locations previously occupied by the frames indicate an increase in TCPP concentration of a
minimum of 204 |ig m"2 during the 290-h Phase III.
4.3.3.2.2 TCPP AND PMDETA DEPOSITED ON PPE OF THE SPRAYING/SAMPLING TEAM
Concentrations of TCPP at various locations on the suits ranged from 8E+03 to 27E+03 and
from 1.7E+03 to 3.4E+03 |ig m"2 for the sprayer and helper, respectively. Concentrations of
PMDETA at various locations on the suits ranged from 3.8E+03 to 22E+03 and from 0.15E+03
to 4.6E+03 |ig m"2 for the sprayer and helper, respectively. Surface area of the Tyvek® suits was
estimated to be 2.9 and 2.5 m2 for the suits worn by the applicator and helper, respectively.
Sampling locations included two locations on the torso, one on each leg and one on the back of
each suit. No samples were collected from gloves, booties, or hoods. Unexposed Tyvek® was
extracted for TCPP background as described in Section 2.11 and no TCPP or PMDETA was
detected in the extracts. Results are summarized in Table 4-3.
Table 4- 3 Concentrations of TCPP and PMDETA deposited on Tyvek® suits
PPE Tyvek®
Suit
Sampling
location
Sprayer
Average TCPP*
(M9 m-2)
Standard
Deviation
Average PMDETA
(M9 m-2)
Standard
Deviation
Front (n=4)
2.11E04
7.3E+03
1.23E+04
6.94E+03
Back (n=1)
8.14E04
2.20E+04
Helper
Average TCPP*
(M9 m-2)
Standard
Deviation
Average PMDETA
(MS m-2)
Standard
Deviation
Front (n=4)
2.73E+03
9.96E+02
1.75E03
2.02E03
Back (n=1)
1.67E+03
1.53E02
*TCPP Limit of Quantitation = 27 |ig m"2
SPFI Methods Development
Page 94
-------
Not surprisingly, concentrations of TCPP and PMDETA on the PPE were much higher for the
sprayer than the helper.
4.3.3.2.3 TCPP COLLECTED BY HVAC FILTERS PLACED OVER THE CHAMBER EXHAUST DUCT
As described in Section 2.10, HVAC filters were fitted over the opening of the exhaust duct in
the ceiling of the chamber. A new filter was installed at the start of the test, and the end of Phase
I and Phase II. The filter installed after Phase II was collected at the end of the test. The amount
of TCPP collected by each filter was estimated from the average of the five subsamples from
each filter. Dilutions factors of 100, 10, and 5 were required for Filters 2, 3, and 4 respectively,
where Filter 2 designates the filter used during Phase I, Filter 3, Phase II, and Filter 4, Phase III.
Summary results for the extractions of TCPP from the filters are presented in Table 4-3.
Tripropylphosphate-d2i RCS was added to each filter prior to extraction by sonication in acetone.
The Filter 2 extractions with 100-fold dilution were not adjusted by the RCS because at 100-fold
dilution, the RCS was below the level of quantitation. Recovery of the RCS for Filters 3 and 4
averaged 102 and 108%. The RSD of TCPP concentration for the five samples collected from
Filter 2 TCPP is similar to the percent RSD observed for Filters 3 and 4.
Table 4- 4 TCPP collected on HVAC exhaust duct filters
Filter
ID
Average TCPP
Concentration
(Mg m-2)
Standard
Deviation
(n=5)
RSD
(%)
Mass per
filter (mg)
Notes
1
ND*
Blank Filter (blank)
2
204E+03
20.2E+03
9.9
66±6.6
Phase I
(0 - 3.3 hours)
3
23E+03
2.3E+03
9.8
7.6±0.7
Phase II
(3.3 - 670 hours)
4
6.2E+03
0.8E+03
12.7
2.0±0.3
Phase III
(670 - 960 hours)
*Non detect
TCPP Limit of quantitation = 55 |ig m"2
PMDETA was quantitatively recovered from the Phase I filter at the 5-fold dilution level. The
concentration determined on the filter for five subsamples was 15.7 ±3.1E+03 |ig m"2 (average ±
standard deviation) and the estimated PMDETA mass on the filter was 5.1± 1.0 mg. PMDETA
was not detected on an unused filter or the Phase II or Phase III filters. At the 5-fold dilution
level, recovery of the tripropylphosphate-d2i RCS averaged 110±15% for the analyses of the five
subsamples.
4.3.3.2.4 TCPP DETERMINED BY WIPE SAMPLING
Wipe samples were collected at the end of Phase III to estimate deposition of flame retardant on
surfaces of the mixing fans, sampling trees, exhaust duct, walls, ceiling, and the area behind the
frames for comparison with deposition samplers. Repeat wipes were collected from each area
sampled to evaluate wipe efficiency.
SPFI Methods Development
Page 95
-------
• Three wipes (wipe plus two re-wipes) were collected for exhaust duct (glass and
stainless steel), wall (foil), sampling tree (stainless steel), and behind the frames
(foil).
• Two wipes (wipe plus one re-wipe) were collected from each of five sample
locations for each fan (plastic). It was observed that two wipes with acetone began
to soften the plastic.
Wipe Sampling Results for Fans
Wipe samples (wipe plus re-wipe) were collected from the vanes covering the front and back of
the fans, fan blades, base plates, and support arms. TCPP concentrations, determined as the sum
of the first and second wipes, ranged from 0.18E+03 to >13E+03 |ig m"2 with highest
concentrations observed on the front vanes, blades, and back vanes. Although the front vanes
appear to have the highest concentrations, the measurements were not reliable due to failure of
theRCS.
Table 4- 5 TCPP recovered from wipe samples for the mixing fans
Wipe Location
Average Concentration
Surface Area
Mass TCPP
RSD**
MS m 2
m2
M9
%
Fan Blades
13.1±0.9E+03
0.12
1600
9
Front Vanes*
22.7±1.4E+03*
0.04
840*
9
Back Vanes
6.8±0.9E+03
0.03
170
20
Base Plate
0.18±0.04E+03
0.04
7
30
Support Arms
0.57±0.14E+03
0.03
20
34
*RCS out of QC limits, (25 fold dilutions, RCS 0.22, 0.31 for Fan 1 and Fan 2, respectively.)
**% RSD determined as difference (fanl - fan2)/average(fan 1, fan2)
Wipe Sampling Results for the Exhaust Duct, Behind Frames, and Adjacent to Deposition
Samplers on Walls and the Ceiling
Wipe samples were collected at the end of the test from two locations in the glass exhaust duct:
(1) from the 90° elbow above the ceiling of the chamber and (2) at the end of the horizontal run
as shown in Figure 2-4. Wipes were collected from the four locations adjacent to the deposition
samplers placed behind the frame locations, adjacent to the deposition samplers on the walls, and
on the ceiling. The TCPP wipe results are presented in Table 4-6. Each surface was wiped three
times. The re-wipe percent of total indicates that the initial wipe recovered about half of the mass
recovered with three wipes at the sampling location. No wipes were collected from the floor.
SPFI Methods Development
Page 96
-------
Table 4- 6 TCPP wipe sample results from chamber surfaces atthe end of Phase III
Wipe Location
Concentration
Percent of Total Mass on
Re-wipes
M9 m"2
%
Exhaust Elbow
0.39E+03
35
Exhaust Duct
0.23E+03
23
Behind Frame 1
0.57E+03
59
Behind Frame 2
0.45E+03
58
Behind Frame 3
0.40E+03
67
Behind Frame 4
1.1E+03
65
Door Wall
0.34E+03
36
Left Wall
0.11E+03
45
Right Wall
0.64E+03
44
Ceiling 1 (CT4)
0.08E+03
64
Ceiling 2 (CT2)
0.09E+03
69
4.3.3.2.5 SPF NODULES RECOVERED FROM THE CHAMBER AT THE END OF PHASE III
Inspection of the chamber at the completion of the test revealed pieces of foam on the floor that
had apparently fallen from the edges of the frames when they were removed from the chamber
and nodules of foam that may have dripped from the nozzle of the gun during application. The
foam pieces were collected and placed into a tared bag and weighed. The amount of foam
insulation recovered from the floor was 31.1 grams. The nodules collected from the floor of the
chamber are shown in Figure 4-10. The projected planar surface area is estimated to be 600 cm2,
or approximately 0.06 m2
SPFI Methods Development
Page 97
-------
It4 . /m c
J> 2> 74 » » ¦
Figure 4-10 Overshoot nodules collected from the floor of the chamber
4,3.4 Phase I and II Concentrations of the Amine Catalyst PMDETA
Concentrations of PMDETA determi ned by extraction from PPE and the HVAC covering the
opening to the chamber exhaust duct at the end of Phase I were presented in Sections
4.3.3.2.2and 4.3.3.2.3. Concentrations of PMDETA determined in the air of the chamber and
exhaust duct by sampling with multi-bed sorbent traps with TD-GC/MS analysis and modified
OVS with solvent extraction and direct injection to a GC/MS are presented in the following
sections.
4.3.4.1 Phase I Concentrations of PMDETA Determined by TD-GC/MS
Twelve multi-bed sorbent samples were collected from the chamber through the side port and 12
from the exhaust duct during Phase I. Sample volumes ranged from 0.1 to 5 L. Three 0.1 L
samplers were collected from the chamber side port sampling location by pulling the air through
the multi-bed sorbent traps with a gas sampling syringe whereas all other samples were collected
at a sampling rate of 0.1 L min"1 using mass flow controllers to control sampling flow rate for
periods of five to 50 minutes. The 0.1L sample collected at 0.04 h was below the detection level
SPI I Methods Development
Page 98
-------
whereas the 0.5 L pumped sample collected at 0.07 h was within the calibration range. The
masses determined for the 0.1 L samples collected at 0.25 and 0.42 h were within the calibration
range of the TD-GC/MS system whereas the mass of PMDETA collected on the 0.5 L samples
0.26 and 0.44 h were above the calibration range. Maximum concentration observed was over
13,000 |ig m"3 determined from the 0.1 L samples collected at the chamber side port. As seen in
Figure 4-11, concentrations declined rapidly after the completion of spraying to <300 |ig m"3
within the Phase I sampling period.
O Chamber A Exhaust Duct O Chamber (100 mL grab sample)
18000
16000
^ 14000
^ 12000
§ 10000
E 8000
5 6000
u
o 4000
2000
0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
Elapsed Time (hours)
Figure 4-11 Phase I Concentrations of PMDETA determined by sampling with multi-bed sorbent traps
AND TD-GCMS ANALYSIS
4.3.4.2 Phase I Air Concentrations of PMDETA Determined by OVS with GC/MS Analysis of
Extracts
Twenty modified OVS were collected by the spraying/sampling personnel from inside the
chamber during Phase I at ten time periods. Sample volumes ranged from 9.3 to 30 L with a
sampling rate of 1 L min"1. PMDETA was not quantified in extracts from the filter or backup
XAD-2 resin bed (IQL 12 - 38 |ig m"3 for sample volumes of 9 to 30L). Concentrations
determined in the chamber ranged from 10.9E+03±6% (RPD) to
-------
Tree 2 yielded valid results so the data are not useful for evaluating potential PMDETA
concentration gradients in the chamber during Phase I.
When the PMDETA time - concentration thermal desorption and OVS extraction data are
plotted in log format, it is apparent that the concentrations determined from the OVS decay more
quickly following completion of the spray event than the concentrations determined by thermal
desorption of multi-bed sorbent traps. See Figure 4-12, below. The data suggest systematic bias
with decreasing concentrations for the OVS compared to the concentrations determined by
thermal desorption of multi-bed sorbent samplers. The results demonstrate the advantages of use
of non-thermal desorption techniques when sampling environments with high and rapidly
changing concentrations. However, the results indicate need for additional work to improve
quantification of PMDETA collected on XAD-2.
O Modified OVS o Chamber -multi-bed sorbent A Exhaust Duct - multi-bed sorbent
10.00
~ 9.00
fO
E 8.00
Jl 7.00
§ 6.00
'H 5.00
c 4.00
-------
The concentrations determined after 500 h are near the MQL and concentrations dropped below
IQL as soon as the frames were removed from the chamber.
O PMDETA Chamber A PMDETA Exhaust Duct
® ^ a © at
0 100 200 300 400 500 600 700
Elapsed Time (hours)
Figure 4-13 Phase II PMDETA chamber and exhaust duct concentrations determined by TD-GCMS
4.3.4.4 Phase II Air Concentrations of PMDETA Determined by OVS with DI-GC/MS
Modified OVS were collected at 12 sampling periods during Phase II by placing samplers into
the chamber through ceiling ports CT-2 and CT-4. Sample volumes were increased from 70 L at
17 h to 250 L at 323 h. PMDETA concentrations determined with the modified OVS were
consistently lower than those determined by TD-GC/MS of multi-bed sorbent traps and the ratio
of concentrations of OVS to multi-bed sorbent samplers decreased from 0.23 to 0.07 over that
time period. A rationale for the apparent bias has not been identified. The OVS PMDETA are not
considered reliable and they are not presented.
4.3.5 Phase I and II Concentrations of VOCs Determined by TD-GC/MS
Emissions of 3-chloropropene, 1,2-dichloropropane, 1,4-dioxane, 2-methyl-2-pentenal, and
chlorobenzene were identified and quantified during Phase I and Phase II of the experiment.
These compounds were not listed as constituents of side A or side B reagents on the Safety Data
Sheets for the product. Results for 2-methyl-2-pentenal are not reported due to failure to meet the
DCC acceptance criteria for 16 of 26 analyses.
SPFI Methods Development
Page 101
-------
4.3.5.1 Phase I Air Concentrations of Selected VOCs Determined by TD-GC/MS
As seen in Figures 4-14 through 4-17, concentrations of the VOCs changed rapidly in the
chamber and exhaust duct during the period of application. The concentration changes are
consistent with the timing of the spray events. Spraying of Frames 1 and 2 was completed at 0.15
h followed by OVS and isocyanate sampler sample collection, then Frames 3 and 4 were sprayed
between 0.26 and 0.37 h. Peak concentrations in the chamber and exhaust duct were observed
shortly after the completion of each spraying event.
There are some differences between reported chamber concentrations during the spray event that
may be due to the sampling strategy. The gas sampling syringe that was employed to collect 0.1
L samples at 0.04, 0.25 and 0.42 h (port CS-11-P3) pulled the air through the multi-bed sorbent
tube in about 10 seconds whereas the 0.5 L samples with sampling time midpoints at close to the
same times pulled from a port located within 10 cm (CS-11-P4) over a 5-minute period by using
vacuum pump and mass flow controller to maintain sampling flow rate. The concentrations of
1,2-dichloropropane, 1,4-dioxane, and chlorobenzene determined for the samples collected with
the gas sampling syringe are higher than the concentrations determined for the samples collected
over a 5-minute period using vacuum pump and mass flow controller. This pattern was not
observed for 3-chloropropene. These differences may be indicative of the variability of
emissions during application. These observations suggest that for purpose of emissions
characterization, sample collection periods during periods of rapidly changing concentrations
need to be the same to avoid bias introduced by differences in concentration spikes that might
occur during application. Concentrations of the VOCs dropped rapidly following cessation of
spraying and decreased slowly within an hour of completion of spraying.
DO
3
c
O
O Chamber A Exhaust Duct OChamber (100 mL grab samples)
400
350
300
250
200
150
100
<1>
u
c
O
u 50
0
*
§
ti t)
w A
jp.
A
©
%
£
A
• A <¦*
1 © *
i
©
A
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
Elapsed Time (hours)
Figure 4-14. Phase I concentrations of 1,2-dichloropropane
SPFI Methods Development
Page 102
-------
O Chamber A, Exhaust Duct • Chamber (100 mL grab samples)
800
700
600
•= 500
c
£ 400
TO
£ 300
0)
£ 200
° 100
£
DO
I
I ti
r T
» %
1L
* a *
© 4
© *
0.0 0.5 1.0 1.5 2.0 2.5
Elapsed Time (hours)
3.0
3.5
Figure 4-15. Phase I concentrations of 1, 4-dioxane
O Chamber A Exhaust Duct • Chamber (100 mL grab samples)
180
„ 160
m
£ 140
3: 120
.2 100
£ 80
S 60
§ 40
u 20
0
0.0 0.5 1.0 1.5 2.0 2.5 3.0
Elapsed Time (hurs)
Figure 4~ 16. Phase I concentrations of chlorobenzene
I *
as &
f
m
SPFi Methods Development Page 103
-------
4.3.5.2 Phase II Air Concentrations of Selected VOCs Determined by TD-GC/MS
Multi-bed sorbent traps were collected from the chamber side port and the exhaust duct sampling
port at 14 sampling events over the Phase II period. Concentrations of four of the five VOCs
determined by sampling from the side port of the chamber (CS-11) and from the exhaust duct
port (DE-1) are shown in Figures 4-18 through 4-21. Concentrations of 2-methyl-2-pentenal,
which are not plotted, increased from 5 to 10 |ig m"3 initially, then decreased to <1 |ig m"3 by 168
h. For all compounds, concentrations determined in the chamber compared well with
concentrations determined in the exhaust duct.
RPD between duplicate samples for nine samples collected from the chamber side port averaged
1.3% except for one suspect concentration determination for 3-chloropropene, an apparent
outlier that can be seen in Figure 4-19. In general, the VOCs followed a similar pattern, in that
concentrations determined at 10.6 h were all higher than those determined at 5 h following the
decrease in chamber air change rate from 4 to 0.4 h"1. The initial increase in concentration was
followed by a period of decreasing concentration through approximately 75 to 100 h.
Concentrations then decreased slowly over the remainder of Phase II except when the air
temperature of the chamber increased due to failure of the air handling system for the building
housing the chamber. The most apparent increase in concentration with increased temperature
was observed for 3-chloropropene.
SPFI Methods Development
Page 104
-------
O Chamber a Exhaust Duct O Suspected outlier - chamber
()
I
I
I ,
o
f 1
f
i
i
i
* §
0 100 200 300 400 500 600 700
Elapsed Time (hours)
Figure 4-17 Phase II concentrations of 3-chloropropene
01,2-DCP Chamber A 1,2-DCP Exhaust Duct
£
)
©
a
&
&
® $ a
©
m A
0 100 200 300 400 500 600 700
Elapsed Time (hours)
Figure 4-18 Phase II concentrations of 1,2-dichloropropane
SPFI Methods Development
Page 105
-------
O Chamber A Exhaust Duct
T
K
i;
o
£
ft
ft
ft
ft ft ft
ft
ft
ft ft
0 100 200 300 400 500 600 700
Elapsed Time (hours)
Figure 4-19. Phase II 1,4-dioxane concentrations
OChamber a Exhaust Duct
|
I I
T
T X L
I
1
9 9 f
I
|
X
0 100 200 300 400 500 600 700
Elapsed Time (hours)
Figure 4- 20. Phase II concentrations of chlorobenzene
SPFI Methods Development
Page 106
-------
4.3.6 Phase I and II Concentrations of Aldehydes and Ketones Determined by HPLC-
DA
Plywood and wood are known sources of aldehyde and ketone emissions [33], To minimize the
impact of emissions from the plywood substrate and wood framing materials on chamber air
concentrations, the wood frame edges of the substrate frames and the gaps between the edges of
the frames and the walls of the chamber were sealed with aluminized tape. Prior to the start of
the test, background samples were collected at the low and high air change rates planned for the
test with the substrates in place.
Concentrations of aldehydes determined for the background samples at high and low air change
rates were presented in Table 3-26. Formaldehyde, acetaldehyde, acetone, propionaldehyde,
valeraldehyde, and hexaldehyde were observed at quantifiable concentrations in the chamber at
the low air change rate and hexaldehyde was quantified at the higher air change rate. The factor
of ten increase in hexaldehyde concentration at the low air change rate was inversely
proportional to the ten-fold decrease in air change rate.
4.3.6.1 Phase I Air Concentrations of Aldehydes and Ketones
The Phase I concentrations of the aldehydes shown in Figures 4-22. Concentrations were highest
during the period of application and decreased rapidly. Concentrations observed throughout
Phase I were well above the background concentrations determined the air change rate of 4 h"1.
Acetaldehyde was not observed after 0.5 h until the air change rate was decreased to 0.4 h"1 in
Phase II.
—O--Formaldehyde o Acetaldehyde —A—Propionaldehyde —Acetone
300
crT"
E 250
txo
3.
o 200
'+-»
(O
i—
-t—1
S 150
c
o
u
100
50
0
0 0.5 1 1.5 2 2.5 3 3.5
Elapsed Time (hours)
Figure 4- 21. Phase I aldehyde concentrations.
iA
1
v—
\
\
\
0
\
>»
/~
t
t
/
Vv
\,
't>-
SPFI Methods Development
Page 107
-------
4.3.6.2 Phase II Air Concentrations of Aldehydes and Ketones
Acetaldehyde and isovaleraldehyde Phase II concentrations are shown in Figure 4-23.
Acetaldehyde concentration reached a maximum of 27 |ig m"3 at 10 h, then decreased to the low-
flow background level by 200 h. The concentration of isovaleraldehyde, which was not observed
in any of the background or Phase I samples, averaged 11.1±1.5 |ig m"3 until rising to 18.3 |ig m~3
when the chamber temperature peaked at 480 h. The dashed line shows the acetaldehyde low-
flow rate background concentration.
O Isovaleraldehyde o Acetaldehyde
f
L
T
*
$
* $ a
$>
$
$
$
i
TT T T
¦ *
T
f
T
T "IT
z
H-i—5
—f—
4™
"l—k-
—f
0 50 100 150 200 250 300 350 400 450 500
Elapsed Time (hours)
Figure 4- 22. Phase II aldehyde concentrations.
Formaldehyde and hexaldehyde, shown in Figure 4-24, were quantified during Phase II, however
concentrations were below low-flow background levels of 11±0.01 and 55±4 |ig m"3,
respectively. It is not possible to determine from the data if the concentrations in the air of the
chamber were due to emissions from the substrate materials or the SPF. The Phase I
concentration data indicate brief emissions of formaldehyde, acetaldehyde, propionaldehyde, and
acetone. The Phase II data indicate emissions of isovaleraldehyde. Additional information is
needed to interpret the Phase II emissions of formaldehyde and hexaldehyde.
SPFI Methods Development
Page 108
-------
A Formaldehyde O Hexaldehyde
———• Formadehyde BKG — • - Hexaldehyde BKG
A 4
o
A
2
ft
* &
O
A
£
0 50 100 150 200 250 300 350 400 450 500
Elapsed Time (hours)
Figure 4- 23. Phase II formaldehyde and hexaldehyde concentrations
4.3.7 Phase I Particle Size and Number Characterization
Particle emissions size and number were characterized by pulling air from the chamber through a
1 m long stainless-steel sampling probe that extended into the chamber through a sampling port
in the ceiling of the chamber. The sample stream was divided between the ELPI and the time-of
flight APS particle counting instruments. The ELPI quantified particle number counts in 12
stages ranging from 0.039 |im to 8.220 |im whereas the APS quantified particles in 52 bins over
the range <0.53 to 19.8 |im. The sum of particle counts determined by the ELPI peaked at
121,000 and 91,000 per cubic centimeter at 0.12 and 0.38 h, consistent with the timing of the two
spray events.
Summarized results, in log format, from the ELPI, are shown in Figure 4-25 where particle
number concentrations have been summed for bins <0.5 |im, <3 |im, and across all size bins. As
can be seen in the figure, the differences between the sum of the <3 |im particle number bins and
sum of all size bins cannot be distinguished in the graph. For the 1077 observations between
initiation of spraying and 1.5 h, the sum of particle numbers <3 |im aerodynamic diameter
averaged 99.8±0.06% of the total particle counts whereas the sum of particle numbers <0.5 |im
SPFI Methods Development
Page 109
-------
averaged 81.4±4.3% of the total particle counts. Similar distributions were observed with the
APS with average sum of particles <3 |im aerodynamic diameter averaging 99.8% of the sum of
particles counted in all particle size bins. For the APS, the average percent of particles in bins
>3<10 |im diameter was 0.2% and the average percentage of particles >10<20 |im was <0.001%.
It is possible that larger particles were generated but are not entrained in the 15 L min"1 flow of
the vertical particle sampling line. However, this seems unlikely given that the settling velocity
of particles with aerodynamic diameter of 60 |im is about 10 cm sec"1 in still air and the face
velocity at the sampling probe was about 350 cm sec"1.
- Sum Particles <0.5 |im + Sum Particles <3 |im O Sum Particles 0.039 to 8.22 [xm
12
6
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6
Elapsed Time (hours)
Figure 4- 24. Particle concentrations during the first 1.5 h of Phase I
Inspection of Figure 4-28 suggests a log4inear decay rate of concentration following the
completion of the spray event. The slope of the log of the particle concentration - time plot
between the time periods of 0.5 and 1.0 h is -4.7 with r2 of 0.9996 whereas the air change rate
determined by tracer-decay was 4.1 h"1. The 0.6 h"1 difference between slopes of the tracer decay
gas concentration and the particle decay rate indicates that the overall particle deposition rate
was equivalent to about 0.6 ACH of additional ventilation air. Because the particle number
counts were dominated by fine particles, this analysis obscures the faster decay patterns for
larger particles. For example, the slope of the log concentration particle count versus time plot
for 8.22 |im diameter particle bin was -5.9 with r2 of 0.9195. The plot also shows that
SPFI Methods Development
Page 110
-------
concentrations decreased to about ambient particle number concentrations of 1000 - 2000
particles per cm3 by 1.5 h elapsed time.
4.4 Discussion of Results
The following sections discuss the results for each class of compound and identify what worked
well and what could be improved, from the standpoint of quantifying emissions during and
following application of SPF insulation to substrates in an environmental chamber. Discussion of
air sampling methods, including calculation of mass emitted during each phase, is presented first,
followed by discussion of the results of the surface and material sampling and analyses.
A key step in constructing a source emissions model from chamber concentration data is to
calculate the mass of the chemical emitted from the source material. The mass balance approach
described by Equation 3 is employed to determine mass emitted. The concentrations of the
chemical emissions in the air of the chamber deposited on surfaces are utilized to calculate the
mass of each chemical emitted by the source. The following equations are used to estimate the
mass emitted.
The mass emitted is described by Equation 3,
Where:
We= the mass emitted (|ig)
Wy = the mass leaving the chamber in the ventilation air determined by Equation 3, (|ig)
Wm = the mass remaining in the chamber air after t hours, (|ig)
Wo = the mass of the emission in the air at the start of a period (|ig)
Ws= the mass of the emission in the sink, (Equation 2)
For the application phase of the experiment:
Ws = WWaiis + Wts +Wef Equation (4)
Where:
Wwaiis = mass deposited on the chamber wall surfaces, (|ig)
Wts = the mass deposited on the Tyvek® suits, (|ig)
Wef= the mass deposited on the exhaust duct filter, (|ig)
For the calculation of mass leaving the chamber in the air:
We= Wv+ Wtn-Wo+Ws
Equation (3)
Q yn Q+ Q+i
Equation (5)
SPFI Methods Development
Page 111
-------
Where:
O = the ventilation flow rate, (m3 h"1)
Ci = the mass concentration at time ti, (|ig m"3)
Ct+i = the mass concentration at time ti+i, (|ig m"3), and
n+1 = the number of data points.
And the mass in the chamber air at the beginning (Ww) and end of a period (IV,,?) is:
Wen = VCn Equation (6)
Wto = VCo Equation (7)
Where:
Fis the chamber volume (m3)
Cn is the concentration of the last data point (|ig m"3)
Co is the concentration at time = 0 (|ig m"3)
4.4.1 Isocyanate Sampling and Analysis
It has been suggested by Streicher [34] that impinger/filter samplers might collect higher
concentrations than the denuder/filter samplers during the application of SPF insulation due to
the more efficient derivatization of particle phase isocyanates in the impinger solution. Puscasu
[35] reported that denuder/filter samplers treated with DBA underestimated application phase
MDI emissions compared to impingers. We implemented two sampling approaches utilizing
impingers and denuder-filter samplers for collection and derivatization of isocyanates for
subsequent identification and quantification by LC-MS/MS, as described in ISO 17734. The
objectives were to quantify the emissions during application, investigate potential concentration
gradients in the chamber, investigate potential low bias of the denuder-filter sampler for
quantifying application phase emissions, and investigate how long isocyanate emissions persist
in the chamber.
4.4.1.1 Isocyanate Sampling Strategy
To facilitate collection of side-by-side impinger and denuder-filter samplers, we placed samplers
on sampling support stands in the chamber so that the personnel in the chamber could rapidly
collect samples. This strategy was employed due to lack of sampling ports that would permit
movement of impingers into and out of the chamber by personnel outside of the chamber.
Although this strategy facilitated rapid collection of samples during the application, the
variability of the results prevents us from confidently addressing some of the questions that we
wanted to answer.
SPFI Methods Development
Page 112
-------
4.4.1.2 Quality Issues Associated with Isocyanate Sampling and Analysis
The quality issues are seen in (1) apparent contamination of the denuder-filter field blank, (2)
failure to meet targets for RPD between duplicate samples, and (3) failure to meet quality goals
as demonstrated by results for the blind challenge samples where four of nine impinger spikes
and two of nine denuder-filter spikes failed to meet the 100±25% acceptance criteria. See Tables
3-7and 3-8 for comparisons of duplicate samples and Table 3-9 for recovery of blind spikes for
denuder-filter samplers and impingers.
4.4.1.2.1 RECOVERY OF BLIND SPIKES
The results of the blind spikes were very inconsistent. The low-level spikes were above the MQL
determined from the lowest calibration standard and above the IQL that had been determined by
the standard deviation of seven low level standards. For the denuder-filter sampler, at the low
spike levels of 5, 3, and 1 ng for MDI, p3-MDI, and p4-MDI, recoveries failed for MDI and p3-
MDI and passed for p4-MDI. The mid and high-level spikes passed for the analysis of the
denuder-filter spiked samples. For the impinger spikes, p4-MDI failed at all three levels and
MDI failed at the mid spike level.
4.4.1.2.2 COMPARABILITY OF IMPINGER AND DENUDER-FILTER SAMPLES
Four samples were collected with each media at two time periods during the application process.
As seen in Figures 4-2 and 4-3, where all four data points for each sampler are averaged at 0.07
and 0.31 hours, the mean concentrations determined with each sampler type overlap within the
error bars. The overall ratio of the averages of sum of MDI, p3-MDI, and p4-MDI denuder-filter
to impinger was 1.1 and 1.3 for the four samples collected on each media at two time periods that
encompass the application (see Table 4-1). There are too few data and the variability are too high
to draw conclusions regarding concentration differences due to sampling location or sampling
media. The results do suggest that during the period of application, sampling at multiple
locations may be advisable until reliable data are available to demonstrate that a single sampling
location is adequate to represent isocyanate concentrations during application.
4.4.1.3 Implications for Sampling Isocyanates During Spray Application
The results indicate a need to modify sampling procedures for isocyanates so that the
application/sampling team is not handling the samplers from inside the chamber. This change is
straightforward for denuder-filter samplers because they can be easily placed into the chamber
through sampling ports located on the walls or ceiling of the chamber. Impingers present
challenges to this approach due to the dimensions of an impinger apparatus and the need to
maintain the impinger in an upright position to prevent spillage of solvent.
SPFI Methods Development
Page 113
-------
4.4.1.4 Calculation of Isocyanate Mass Emitted
The isocyanate concentrations in the air of the chamber during and following the application
were due to the gas and aerosol emissions from the stream of material that was sprayed from the
gun as it traveled to the substrate and emissions from the SPF insulation that was deposited on
the substrate. The mass of emissions determined from the Phase I chamber air concentration data
were assumed to be the emissions attributable to the application process. In the absence of sink
terms to account for particle deposition, losses to chamber walls and chemical reactions, the
emissions mass determined by this process underestimates mass emitted. Mass emitted
determined from the Phase I chamber concentrations using Equation 3 was determined to be
2.7E+04 |ig from the denuder-filter samplers and 2.4E+04 |ig from the impinger samplers.
4.4.2 Blowing Agent Sampling and Analysis
The use of the PAS to quantify the application phase emissions demonstrated strengths and
weaknesses to this approach. The strengths were evident in the ability to track rapidly changing
concentrations over a concentration range of several orders. Weaknesses include potential
interferences and lack of sufficient dynamic range to capture the peak concentrations. We
demonstrated that the collection of whole air samples for subsequent dilution can address the
dynamic range issues. A dynamic dilution system could also be used to ensure that the
concentrations remain within range of an instrument. Also, there is a need to identify appropriate
sampling and analysis approaches for the new class of oxygenated blowing agents.
Equation 3 was utilized to calculate the mass of blowing agent emitted during Phase I. and was
estimated to be 1.17 kg, or 1.17E+06 mg.
4.4.3 TCPP and PMDETA Sampling and Analysis
TCPP sampling and analysis strategies were implemented to characterize the emissions during
and following application by measuring concentrations in the air and on multiple surfaces.
Multiple approaches were implemented for air and surface measurements. Each method provided
information that helps paint the picture of the flame retardant emissions. Some of the methods
were clearly works-in-progress that need further development.
4.4.3.1 TCPP and PMDETA Air Sampling and Analysis Approaches
Two approaches were implemented for quantification of TCPP in the air of the chamber,
sampling with multi-bed sorbent traps with TD-GC/MS analysis and sampling with modified
OVS with subsequent extraction and analysis by GC/MS. The objectives for use of the modified
OVS were (1) to evaluate the use of the modified OVS to quantify TCPP and PMDETA
emissions during application where concentrations are very high and multi-bed sorbent traps may
be easily overloaded and (2) provide insight into the gas and aerosol concentrations during
application since the multi-sorbent bed sampler does not separate gas and aerosol phases unless,
for example, a filter is placed in front of the sampler.
SPFI Methods Development
Page 114
-------
4.4.3.1.1 SAMPLING WITH MULTI-BED SORBENT TRAPS
Sampling with multi-bed sorbent traps to quantify application phase emissions was successful
with the exception that the three 0.1 L grab samples collected during the application were not
particularly useful due to uncertainty about the actual midpoint of sampling. The grab samples
allowed quantification of very high concentrations of TCPP and PMDETA during the application
where overloading of multi-bed sorbent traps was observed on samples collected at a sampling
flow rate of 0.1 L min"1 over a 5-minute period. However, there is some uncertainty regarding
the actual sampling time and rate using the syringe pump. Variable volume samples collected at
lower rates over the same time might have resolved this issue. However, the larger issue is that
the application takes place very quickly and concentrations rise from background to very high
concentrations in a very short time. Many samples must be collected very quickly to plot the rise
in concentration, or alternatively, few samples may be collected over the period of the
application to determine an average concentration during application. This strategy is less
resource intense but does not provide information regarding peak concentrations.
The mass of TCPP emitted during Phase I was determined from the mass leaving the chamber in
the exhaust duct air, mass collected on the exhaust duct surfaces, and mass deposited on surfaces
in the chamber. The mass of TCPP leaving the system in the exhaust duct was determined by
applying Equation 3 to the concentrations determined in the exhaust duct. TCPP mass
determined on chamber surfaces and the exhaust duct filter are presented in Table 4-6.
4.4.3.1.2 SAMPLING WITH MODIFIED OVS
Our results indicate that the modified OVS approach as implemented in this test was not a
mature procedure and needs further development. That is, the OVS results during the application
phase exhibited variability in the amount of TCPP mass collected on the filters and some
samples exhibited breakthrough from the front XAD-2 resin bed to the backup bed. There are
several possible reasons for the TCPP variability seen with the modified OVS, including
oversampling during the application, failure of the Teflon™ ring holding the Teflon™ filter in
place in front of the XAD resin bed, TCPP migration from the Teflon® filter to the XAD-2 resin
beds and migration from the front bed to the backup resin bed.
The breakthrough from front XAD-2 to backup XAD-2 resin beds observed for sampling high
concentrations of TCPP was not observed for PMDETA. PMDETA was not detected on the
Teflon filters placed ahead of the glass fiber filters and breakthrough was not observed from the
front XAD-2 resin bed to the backup bed. Since maximum concentrations during application
were several times higher for PMDETA than TCPP during the application, it appears that the
compounds interact differently with the modified OVS sampling media. The lack of PMDETA
on the Teflon filter is surprising since PMDETA was recovered from the Phase I exhaust duct
HVAC filter and from the PPE. Given the potential for very high concentrations of flame
retardant and amine catalyst during application, it is important that reliable sampling and analysis
methods be developed and demonstrated.
SPFI Methods Development
Page 115
-------
4.4.4 VOC Air Sampling and Analysis
The air samples collected on multi-bed sorbent traps during each phase of the test with analyses
by TD-GC/MS provided the concentration - time data needed for determination of mass emitted
and subsequent generation of source emissions models for the target VOCs, PMDETA, and
TCPP.
The mass emitted during Phase I determined using Equation 3 expressed as mass emitted, mass
per unit of surface area, per unit of mass of SPF insulation, and per unit of volume of spray foam
is presented in Tables 4-7 and 4-8 for mass determined from concentrations measured at the
chamber sampling port and exhaust duct port, respectively.
Table 4-7 Phase I mass emitted determined from sampling directly from the chamber sampling port.
Phase 1
3-Chloro-
propene
1,2-Dichloro-
propane
1,4-Dioxane
Chlorobenzene
PMDETA
Emitted (mg)
38.3
19.3
142
6.1
774
Emitted (mg rrr2)
5.4
2.7
20.1
0.9
110
Emitted (mg kg"1)
3.6
1.8
13.3
0.6
73
Emitted (mg rrr3)
126
64
487
20
2560
Table 4- 8 Phase I mass emitted determined by sampling at the exhaust duct sampling port
Phase I
3-Chloro-
propene
1,2-Dichloro-
propane
1,4-Dioxane
Chlorobenzene
PMDETA
Emitted (mg)
42.5
20.6
148
4.9
799
Emitted (mg rrr2)
6.0
2.9
21.1
0.7
114
Emitted (mg kg"1)
4.0
1.9
13.9
0.5
75
Emitted (mg rrr3)
141
68.2
492
16.2
2645
The average ratio of mass determined at the exhaust duct compared to mass determined by
sampling from the chamber side port for the five compounds was 1.05±0.15 (average ± standard
deviation). The differences in estimated mass emitted may be due to integration of
concentrations at two locations with small differences in sampling times over the period of
maximum concentration.
Mass emitted during Phase II, determined with Equation 3 is presented in Tables 4-9 and 4-10.
As with the Phase I comparisons, differences between mass emitted determined from the exhaust
duct compared to mass emitted determined by sampling directly from the chamber are small. For
the five compounds, the ratio of mass emitted determined from sampling at the exhaust duct
compared to sampling from the chamber averaged 0.98±0.04.
SPFI Methods Development
Page 116
-------
Table 4- 9 Phase II emissions determined from the chamber concentrations
Phase II (3.3 - 669
hours)
3-Chloro-
propene
1,2-Dichloro-
propane
1,4-Dioxane
Chloro-
benzene
PMDETA
Emitted (mg)
197
20.6
51.6
11.7
621
Emitted (mg rrr2)
27.9
2.9
7.3
1.7
88
Emitted (mg kg"1)
18.4
1.9
4.8
1.1
58
Emitted (mg rrr3)
651
68.1
171
38.6
2060
Table 4-10 Phase II emissions determined from exhaust duct concentrations
Phase II (3.3 - 669
hours)
3-Chloro-
propene
1,2-Dichloro-
propane
1,4-Dioxane
Chloro-
benzene
PMDETA
Mass leaving (mg)
196
20.6
51.7
10.7
600
Emitted (mg rrr2)
27.8
2.9
7.4
1.5
85
Emitted (mg kg"1)
18.4
1.9
4.9
1.0
56
Emitted (mg rrr3)
648
68.1
171
35.3
1985
The apparent differences between estimates of Phase II mass emitted determined at the two
sampling locations are very small and within the experimental error of the measurements.
4.4.5 Aldehyde Air Sampling and Analysis
The sampling and analysis approach for determination of aldehyde concentrations during
application was demonstrated to be adequate, however the Phase II results are confounded by the
aldehyde emissions from the plywood substrates. Plywood substrates were selected because they
are the material to which SPF insulation is usually applied and use of plywood supports
determination of adhesion and cohesion testing to demonstrate that the SPF insulation meets
product specifications. To minimize the potential impact of aldehyde emissions from the
plywood and wood substrates, the wood frames and support legs were covered with aluminized
tape and the perimeter of the frames were sealed to the walls of the chamber to inhibit migration
of emissions from the backside of the plywood into the chamber.
The strategy was at least partially effective. During the application, concentrations of
formaldehyde, acetone, acetaldehyde, and propionaldehyde were well above the chamber
background concentrations determined at the high air change rate. Mass emitted during Phase I
determined using Equation 3 was 8.1E+03, 3.1E+04, and 1.4E+04 |ig for formaldehyde, acetone,
and propionaldehyde, respectively. However, in Phase II, formaldehyde and hexaldehyde
concentrations were below the low flow background concentrations, acetone and
propionaldehyde concentrations were below background concentrations after 24 h, and
acetaldehyde was above the low flow background for 168 h. Isovaleraldehyde, quantified
throughout Phase II, was not observed in the background samples or the Phase I emissions. The
SPFI Methods Development
Page 117
-------
isovaleraldehyde mass emitted during 480 h of Phase II was determined to be 7.0E+04 |ig.
Although the plywood substrates were useful in terms of facilitating evaluation of foam
properties such as adhesion and cohesion, the use of plywood substrate appears to have created
several problems in terms of quantifying aldehyde emissions from the SPF insulation.
4.4.6 Particle Size and Number Sampling and Analysis
The number of particles emitted per size bin during Phase I was calculated by converting particle
#/cc to #/m3 and using Equation 3, substituting #/m3 for |ig m"3 to calculate the number of
particles per size bin that would leave the chamber in the exhaust air, assuming no filtration or
deposition, based upon the ELPI particle count data. This analysis does not include particles
beyond the 8.22 |im aerodynamic diameter size bin. Although the APS counted particles as large
as 16 |im, the cumulative percentage of particles leaving the system approached 100% near the
2.5 |im size bin with both instruments. The ELPI data shown in Figure 4-26 indicate that about
67% of the particles emitted had an aerodynamic diameter of 0.32 |im or less. Thus, it is not
surprising that the MERV 13 HVAC filter was only about 50% efficient in removing the TCPP
from the air exhausted from the chamber.
The ELPI particle count data indicate that large numbers of fine particles are emitted during the
spray event. The increase in particle numbers as particle size decreases below 0.122 |im suggests
full characterization of particle emissions might include use of ultrafine particle counting
systems. Also, it would be useful to output particle mass from the instruments which can be done
if density and shape factors are known. Gravimetric techniques could also be employed to
characterize the particle mass emissions.
SPFI Methods Development
Page 118
-------
¦o
OJ
t
re
Cl
#
E
i/i
^ 1
'¦5
(.6
^ 4*S
8.9°/
!'9.93
9.96
16 9
9.99
161(
o.oc
l.E+14
l.E+13
l.E+12
l.E+11
l.E+10
l.E+09
l.E+08
l.E+07
l.E+06
l.E+05
l.E+04
l.E+03
l.E+02
l.E+01
1.E+00
0.039 0.072 0.122 0.204 0.320 0.490 0.771 1.242 1.970 3.117 5.216 8.220
Particle Size Bin (|im)
Figure 4- 25. Phase I cumulative particle emissions
%
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
4.4.7 Surface and Material Sampling
The surface and material sampling produced information regarding the impact of spray
application of foam insulation on concentrations of flame retardant on surfaces. The data does
not address whether the TCPP determined from deposition samplers, wipes, and extractions from
materials was deposited as vapor or associated with aerosol or polymer.
4.4.7.1 Deposition Sampling
4.4.7.1.1 ESTIMATION OF MASS OF TCPP ON WALLS, FLOOR, AND CEILING OF THE CHAMBER
Summary results for the floor, ceiling, walls, and area behind the frames are presented in Table
4-11. The mass of TCPP on each surface of the chamber was estimated as the product of the
exposed surface area (m2) and average TCPP concentration determined from the deposition
samplers associated with that surface. The deposition samplers were attached to foil sheets that
were taped to specific locations in the chamber near center points of areas of the chamber not
covered by the frames. The samplers were not randomly deployed on walls, ceiling, and floor
areas, and the results may not be representative of the average concentrations of deposition of
SPFI Methods Development
Page 119
-------
TCPP on each surface. The differences between concentrations observed on the floor and ceiling
at sites near and away from the area of application demonstrate that the amount of deposition is
related to distance from the source.
As discussed in Section 4. 3.3.2.1, the concentration differences between the sampling locations
near the spray event on the floor were 2.5 times the concentration at the location away from the
area of application. The Phase I concentrations for the wall samplers shown in Figure 4-27 show
that the lowest wall concentrations were observed at the door wall, the furthest wall from the
area of application. The uncertainties between duplicate deposition samplers (average RPD 24%,
for 26 pairs) were much smaller than the differences between sampling locations, as shown in
Table 4-11 and Figure 4-27.
Overall, the summary results indicate a downward trend of TCPP concentrations on the surfaces
with time. For the walls and ceiling, the largest percentage decrease in concentration occurred
after removal of the frames from the chamber. Following removal of the frames, deposition
samplers were placed on the unexposed areas of the walls that were previously behind the
frames. Concentrations of TCPP increased at those sites during Phase III whereas concentrations
at the other locations decreased.
Table 4-11 Concentrations and estimated TCPP mass on surfaces of the chamber
Location
End of
Phase
Exposed
Surface
Area (m2)
TCPP Loading
(mg rrr2)
Estimated Mass
(mg)
Average Floor*
Phase I
10.1
11.5±4.9
116±49
Phase II
10.1
8.8±3.4
89±34
Phase III
10.1
8.3±2.8
84±28
Average Ceiling *
Phase I
10.1
0.52±0.18
5.2±1.8
Phase II
10.1
0.67±0.23
6.8±2.3
Phase III
10.1
0.33±0.11
3.3±01.1
Average Walls**
Phase I
22.7
0.88±0.37
20.0±8.4
Phase II
22.7
0.80±0.33
18.2±7.5
Phase III
22.7
0.43±0.18
9.8±4.1
Behind Frames 2,3,4**
Phase III
6.6
0.20±0.02
1.3±0.13
Behind Frame 1
Phase III
2.2
0.86
1.9
* Average of two
** Average of the
sampling locations for the chamber floor and ceiling ± 0.5 times the range,
deposition samplers at three wall locations ± standard deviation.
4.4.7.1.2 SURFACE DEPOSITION CONCENTRATIONS AND AIR VELOCITY
The relationship between the natural log of the TCPP concentration on the deposition samplers
and air speed determined at the high air change rate at the wall locations is shown in Figure 4-27.
The plot suggests a linear relationship between the log of concentration determined for the
deposition samplers and air speed near the surface at the location of the deposition samplers. It
should be kept in mind that the concentration of TCPP on the deposition samplers at the end of
Phase I is the net result of multiple processes that have different time scales, including; aerosol
deposition during application and gas phase sorption and desorption from the surface of the
samplers. As previously discussed, the distance from the spray event also impacts deposition. An
improved experimental design would randomize the placement of samplers on each chamber
SPFI Methods Development
Page 120
-------
surface to provide a more accurate picture of the deposition and the relationships between
distance from the area of application and near surface air speeds.
Right Wall
I 771 > 1
y = 1.6077x + 5.7094
R2 = 0.9907
0J
re 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
_Q
m Air Speed (m s-1)
Figure 4- 26. Correlation between air speed 1 cm above the surface of the deposition samplers and log
TCPP CONCENTRATION DETERMINED AT THE END OF PHASE I.
4.4.7.2 Wipe Sampling
Wipe sampling was conducted to determine TCPP concentrations on surfaces of the sampling
support trees, fans, and areas near the deposition samplers to evaluate comparability.
4.4.7.2.1 TCPP RECOVERED FROM FAN SURFACES
Results for wipe samples of fans presented in Table 4-5 indicate that more than 60% of the mass
was recovered from the fan blades, 32% from the front vanes or grill, 6% from the back grill and
less than 1% from the fan base and support arms. The highest concentrations were observed on
the front vanes however the results are uncertain due to low recovery of the RCS at the 25 fold
dilution level.
4.4.7.2.2 TCPP RECOVERED FROM EXHAUST DUCT SURFACES
The TCPP wipe results for the exhaust duct, areas behind the frames, and areas adjacent to the
deposition samplers are presented in Table 4-6. Each surface was wiped three times. The re-wipe
percent of total indicates that the initial wipe recovered about half of the mass recovered with
three wipes at the sampling location.
C
O
c
<3J
u
c
o
u
a3
7.5
Cl
^ E
£ so
.2 =L
o
Cl
0J
O
6.5
Left Wall
Door Wall
c. c,
SPFI Methods Development
Page 121
-------
4.4.7.2.3 TCPP RECOVERED FROM AREAS BEHIND THE FRAMES
Comparisons between wipe and Phase III deposition samplers demonstrate that wipe sampling
results tend to produce lower estimates of surface concentrations of TCPP and the results vary
considerably from surface to surface. The ratio of concentrations of wipes collected adjacent to
deposition samplers ranged from 0.22 to 0.84. For wipe samples taken near the ceiling deposition
samplers the ratio was 0.25 (85±11: 330±86 |ig m"2), the ratio was 0.22 for wipes collected from
the wall area behind the frames after their removal for Phase III (64±32 :290 ±248 |ig m"2) and
0.84 for the average of left, right and back wall areas not covered by sprayed frames
(364±262:434±156 |ig m"2). Comparison between deposition and wipe samples should be seen in
light of the wipe - re-wipe data in Table 4-6. The percent of the total mass recovered from each
wipe location with the second and third wipes ranged from 23 to 67%. These results demonstrate
the qualitative nature of wipe sampling.
4.4.7.2.4 TCPP RECOVERED FROM SAMPLING SUPPORT FRAMES
Concentrations determined by wiping the sampling support system were 8.9 and 4.0 |ig m"2 for
Sample Tree 1 and Sample Tree 2, respectively. These results are consistent with the
observations of higher concentrations for the deposition samplers placed on the base supports for
the sampling Tree 1 which was located nearest the area of application.
4.4.7.3 Material Sampling
TCPP and PMDETA concentrations were determined for the HVAC filters placed over the
opening to the chamber air exhaust duct and for the PPE worn by the applicator and helper. The
averages of concentrations for each material were used to estimate the mass of TCPP and
PMDETA collected on the material.
4.4.7.3.1 TCPP AND PMDETA DEPOSITED ON HVAC FILTER MEDIA
Spray foam applicators following industry guidance place filters over the inlet or outlet of the
exhaust fans that move air from the area of application to the outdoors. Homeowners and
building managers use HVAC filters to remove particles from indoor air to protect equipment
and reduce exposure to dust and allergens. We placed a MERV 13 filter over the exhaust duct
opening to investigate the efficiency of the filter material to remove SPF insulation emissions
from the air exhausted from the chamber. MERV 13 filters are rated to be 50% efficient in
capture of particles with diameters between 0.3 and 1 |im. The inefficiency of the MERV 13
filter to capture TCPP during the application phase may be due to several factors: (1) MERV 13
filters are rated to be 50% efficient in capture of particles between 0.3 and 1 |im in diameter, (2)
at elevated temperatures, SVOCs may behave as VOCs and although air temperatures did not
exceed 32 °C, the mid-depth foam reached temperatures of 100 °C. Therefore, preventing the
exhaust of TCPP to the environment during application and curing may require use of carbon
filters and filters that capture particles <1 |im in diameter. Regarding the efficacy of removing
gas-phase TCPP during Phases II and III, the results demonstrated that the MERV 13 filters used
SPFI Methods Development
Page 122
-------
in the experiment have little to no impact on airborne concentrations of TCPP after
approximately 4 days of use.
Over the course of Phase II, the mass of TCPP recovered from the exhaust duct filter amounted
to about 1% of the mass leaving the chamber system in the air. Following removal of the frames
from the chamber, the amount captured by the exhaust duct filter in Phase III was about 4% of
the mass leaving the chamber via the exhaust duct.
Table 4-12 TCPP collected on HVAC exhaust duct filters
Filter
ID
Average TCPP
Concentration
(M9 m-2)
Standard
Deviation
(n=5)
RSD
(%)
Mass per
filter (mg)
Notes
1
ND*
Blank Filter (blank)
2
2.04E+05
2.02E+04
9.9
66±6.6
Phase I
(0 - 3.3 hours)
3
2.3E+04
2.3E+03
9.8
7.6±0.7
Phase II
(3.3 - 670 hours)
4
6.2E+03
0.8E+03
12.7
2.0±0.3
Phase III
(670 - 960 hours)
*Non detect
TCPP Limit of quantitation = 55 |ig m"2
4.4.7.2 TCPP and PMDETA Deposited on PPE
Surface area of the Tyvek® suits was estimated to be 2.9 and 2.5 m2 for the suits worn by the
applicator and helper, respectively. Sampling locations included two locations on the torso, one
on each leg and one on the back of each suit. TCPP and PMDETA mass per suit for the sprayer
and helper are shown in Table 4-13.
Table 4-13 Estimate of TCPP and PMDETA deposited on Tyvek® suits
PPE Tyvek® Suit
Sampling location
Sprayer
Helper
TCPP*
(mg/suit)
PMDETA
(mg/suit)
TCPP
(mg/suit)
PMDETA
(mg/suit)
Front (n=4)
30.3±10.5
17.7±11.3
3.4±1.2
2.2±1.1
Back (n=1)
11.7
31.7
2.1
<1.4 (0.2)**
Total
42
49.4
5.4
2.3
*TCPP Limit of Quantitation = 2.7 ng cm"2 or 0.08 mg per suit.
**PMDETA Limit of Quantitation =1.4 mg per suit.
The estimates of mass on each suit underestimate mass leaving the chamber on PPE of the
sprayer and helper because we did not sample booties, head covers, and gloves. A minimum of
47±15 mg TCPP and 52±12 mg PMDETA was transported out of the chamber environment on
PPE when the sampling team exited the chamber.
SPFI Methods Development
Page 123
-------
4.4.7.3.3 TCPP RECOVERED FROM SPF INSULATION MATERIAL COLLECTED FROM THE FLOOR
If the concentration of TCPP in the foam recovered from the floor is the same as was determined
for the foam applied to the substrates, (12.7±1.8%), then the nodules would contain 3.95 grams,
or 3,950 mg which is about 40 times the 100 mg of TCPP estimated as the sum of TCPP on the
walls, ceiling, and floor from the deposition samplers at the end of the test. Although we did not
investigate the emissions from the nodules separately from the chamber surfaces, it is safe to
assume that they had some impact on the Phase III TCPP emissions. The projected planar surface
area of the nodules is roughly 0.001 times the surface area of the chamber so the actual impact
TCPP emissions from the overshoot nodules on overall TCPP emissions from the chamber may
be small.
4.4.8 Summary of Mass Determined for TCPP in Various Compartments
The mass determined for TCPP in various compartments for each Phase is shown in Table 4-14.
The data are useful in understanding what happens to TCPP in the chamber system during and
following application of SPF insulation. The estimated mass amounts in the table do not include
the mass estimated to be in the applied foam or overshoot (3,950 mg), or that determined by
wipe sampling the fans (2.7 mg), sampling trees, and exhaust duct (2.4 mg).
For Phase I, roughly 50% of the mass emitted due to spray application remained in the chamber
on surfaces and about 50% left the chamber via the exhaust duct or was captured on the exhaust
duct filter. Specifically, 34 % of the 386 mg of the TCPP accounted for in the air and on surfaces
left the chamber system via the exhaust duct, 17% was captured on the exhaust duct filter, 37%
was deposited on chamber surfaces, and 12% was deposited on PPE of the applicator and helper.
Table 4- 14 Mass of TCPP determined in various compartments
Phase 1
Phase II
Phase III
Sum
Mass Leaving in Air of the Exhaust Duct (mg)
131
606
55
792
Mass Collected on Exhaust Duct Filter (mg)
66
8
2
76
Mass Determined on PPE (mg)
47
47
Mass Estimated on Chamber Surfaces (mg)
142
114
99
Total Estimated TCPP (mg)
386
732
156
% Mass leaving in Air Collected on HVAC Filter
33%
1%
4%
The impact of the TCPP emissions from the chamber walls and overshoot can be estimated by
comparing the average emission rate during Phase II with the average during Phase III. In each
case, the average system emission rate is determined as the sum of the mass leaving the chamber
in the air plus the mass collected on the HVAC filter. The Phase III TCPP chamber system
emission rate average of 197 |ig h"1 is 21% of the Phase II average emission rate of 922 |ig h"1.
The implications of this observation are (1) the test method needs to account for emissions of
flame retardant from chamber surfaces and overshoot, (2) building simulation models and
exposure models need to consider the impact of secondary emissions, and (3) strategies designed
to reduce long-term exposure to flame retardants through, for example, encapsulation of the SPF
SPFI Methods Development
Page 124
-------
insulation, need to consider control of emissions due to deposition on other surfaces during
application.
The methods development research reported here was conducted with a low pressure two
component kit where the operator sprayed each frame in a continuous manner. High pressure
applications reportedly generate more aerosol due to the high-pressure spray process and
generate more waste in the form of globules of SPF insulation because the spray gun is
repeatedly triggered on and off by the operator to obtain the desired application. Globules form
at the end barrel of the spray gun and fall to the floor. Cardboard was placed over the floor of the
chamber to prevent contamination. In terms of developing a test method that characterizes TCPP
emissions from SPF insulation, these observations raise questions about how to structure a test
protocol so that the results are relevant to ways that the product is used.
If the overall goal is to generate an emissions term due to the spray application process, then it
may be sufficient to assign all of the emissions determined by measuring air concentrations to
the application to the target surfaces, recognizing that the concentrations in the air result from
emissions from the spray process and from emissions from multiple surfaces. However, for the
purposes of exposure modeling and developing source management strategies, it makes sense to
have knowledge of the impact of the spray application on concentrations on surfaces that may
impact long-term emissions. Deposition samplers placed on the ceiling and floor could be
employed to estimate the amount of flame retardant, or other semivolatile compounds on on-
target surfaces.
4.5 Summary of the Results
Spray application of SPF insulation to plywood substrates in an environmental chamber supplied
with conditioned air resulted in rapid increases in concentrations of isocyanates, blowing agent,
flame retardant, amine catalyst, VOCs, aldehydes, and aerosols.
• Air temperatures in the chamber increased to 30 °C during application and decreased to 26
°C within an hour of completion of the spray events whereas foam temperatures reached 110
°C and decreased to 40 °C within an hour.
• Concentrations of emissions increased rapidly by several orders during application and
decreased rapidly for most compounds following cessation of spraying.
• Following cessation of spraying, isocyanate, flame retardant, and particle concentrations
decreased more rapidly than the air change rate whereas blowing agent and VOC
concentrations decreased at rates slower than the air change rate.
• Concentrations of aerosols with aerodynamic diameters of 0.039 to 8.22 |im rose rapidly with
each spray event to a maximum of 1.2E+06 particles per cm"3.
o Particle counts were dominated by particles of <3 |im aerodynamic diameter with on
average, >80% of particles with aerodynamic diameter <0.5 |im.
• Flame retardant was deposited on all surfaces in the chamber during the application.
SPFI Methods Development
Page 125
-------
o Concentrations of TCPP measured near the floor were about ten times higher than
observed on samplers placed on the walls and ceiling.
o Approximately 50% of the flame retardant emitted during application deposited on
chamber surfaces and PPE of the sprayer and helper.
o Of the 50% of the TCPP leaving the chamber in the air 33% was captured by the
HVAC filter over the exhaust duct opening.
• The wood-framed plywood substrates emitted aldehydes into the chamber air.
o Formaldehyde, acetaldehyde, acetone, and propionaldehyde emissions were above
chamber/plywood substrate background concentrations during the application and
decreased rapidly at the cessation of the spray event.
• Air concentrations of emissions increased following the decrease in chamber air change rate
from 4.1 to 0.4 h"1.
• Concentrations of HFC-134a increased from 8.1E+04 to 1.4E+05±1.2E+03 |ig m"3 within 4 h
of the decrease in air change rate and were relatively constant for 2 h.
• Concentrations of HFC-134a increased 45% during the period where chamber air
temperature increased by 3 °C.
• Concentrations of TCPP in the air of chamber and exhaust duct increased for a period 10 to
24 h then decreased slowly until the unplanned rise in chamber air temperature.
• TCPP concentrations in the air of the chamber and exhaust duct increased by 42% during the
period of temperature rise of 3 °C.
• The HVAC filters were <5% effective in removing the TCPP emitted from the SPF
insulation and from chamber surfaces during Phase II and Phase III.
• The TCPP emissions from the chamber walls, floor, and ceiling may have accounted for
more than 20% of the emissions from the chamber system with sprayed frames.
• Isovaleraldehyde was not observed in background samples or Phase I but was quantified
during Phase II.
4.6 Lessons Learned
Over the course of this research, the following lessons were learned:
• Additional work is needed to evaluate the comparability of denuder-filter samplers and
impingers for quantitation of application phase isocyanates.
• Variability was greater for denuder filter samples than for impingers, however the
contamination of the denuder-filter trip blank confounds interpretation of comparisons
between the two sampling approaches.
SPFI Methods Development
Page 126
-------
• Foreknowledge of expected concentrations is needed to scale sampling rates and volumes
and to prevent overloading of sampling media during application events.
• Dilutions of the impinger samples collected during the application events were needed to
bring analyte concentrations into the calibration range of the LC-MS/MS.
• HFC-134a concentrations exceeded the calibration range during the application- a dilution
system or collection of more frequent bag samples during application were needed to
quantify peak concentrations.
• Variable volume samples would be needed during application to quantify the concentrations
of the amine catalyst by thermal desorption of multi-bed sampling media during the
application events.
• Deposition samplers were essential for quantifying TCPP emissions.
• The apparent relationship between deposition of TCPP on walls and air speed suggests that
this variable should be considered in SPF insulation test protocols.
• The MERV 13 HVAC filter reduced the TCPP leaving the chamber system during the
application by 33%.
• The HVAC was <5% effective in removing gas phase TCPP from leaving the chamber
system in the air leaving the chamber in the exhaust duct.
• Aldehyde emissions from the wood-framed plywood substrates complicated interpretation of
the aldehyde emissions from SPF insulation.
• The modified OVS sampler needs more work before consideration for use in separation of
gas and aerosol phase TCPP and quantification of PMDETA emissions during application.
5.0 Emissions Rates and Emission Factors
5.1 Calculation of Emission Rates and Emission Factors
Emission rates and emission factors describe the rate at which chemicals move from a source to
the environment and are central components of indoor air quality models that predict
concentrations of emissions in buildings, a key component of exposure and risk modeling.
Source emissions models fall broadly into categories of empirical models where model
parameters are derived from the conditions of the test, and mass transfer models where key
parameters are derived from the properties of the chemicals that make up the product, the
physical characteristics of the product and of the environment.
Mass transfer-based source emissions models are potentially more useful than empirical models
because the validity of the model extends beyond those of the test environment employed to
generate model parameters. However, obtaining values for key parameters for mass transfer
models may be challenging or unavailable for specific chemicals in a specific product. In that
regard, experimental data generated in a controlled environment provides information that can be
SPFI Methods Development
Page 127
-------
used to generate parameters for empirical models and generate data that can be used to evaluate
the ability of empirical and mass transfer models to predict emissions from a specific source.
Measurement and modeling of emissions that include spray application of reactive chemicals to
substrates introduces complexities beyond characterization of emissions from a material or
product that is produced at an off-site manufacturing facility and transported to the site as a
completed product. For SPF insulation, concentrations of emissions in the chamber during
application may be due to emissions from the stream of reagents ejected from the nozzle of the
gun, emissions from the SPF insulation applied to the substrate, and emissions from overshoot
that deposits on surfaces in the chamber. No attempt is made to apply mass transfer models to the
application process or to separate the emissions due to the application process including those
due to evaporation from aerosol generated by spray application. In that regard, during the
application, the emissions due to the process of making the product are attributed to the product.
Source emissions rate terms for use in simulation systems such as IECCU can be generated from
the chamber concentration data using several techniques. Emission rate terms or emissions
factors can be calculated directly from chamber concentration data in an electronic spreadsheet
with a numeric solution to Equations 9 orlO, rearrangements of Equation 8, the mass balance
equation for the chamber system, assuming that the concentrations of the emissions are uniform
in the chamber.
where,
V = chamber volume (m3),
= incremental change in concentration over the period dt (pig m~3),
R(t) = emission rate at time t, (/ugh'1),
O = supply airflow rate (nf h'1),
C(t) = concentration of emission at elapsed time t,
S(t) = loss rate of the emission to chamber surfaces through deposition, chemical reaction, or
other mechanism (/ugh1)
Assuming that S(t) is negligible, the emission rate by rearrangement of Equation 8 is:
Equation (8)
Equation (9)
The emission rate can be divided by area of the source to obtain the emission factor.
_ (f+wc))
L
Equation (10)
Where:
SPFI Methods Development
Page 128
-------
E(t) = emission factor at time /, (|ig m"2 h"1)
AC
— = change in concentration over the time interval between samples
N= air change rate (If1)
C = chamber concentration at time (t), and
L = chamber loading factor (source area, m2/chamber volume, m3)
Equation 7 can be solved numerically from the time -concentration data in an electronic
spreadsheet.
We demonstrate several approaches to generation of source emission rate terms for the Phase I
data, including the method of direct calculation of emission rates from chamber concentrations,
dual first order decay model, and models that increment mass emitted based upon increments of
spray time or area. For the Phase II data, we employed the direct calculation method to generate
emission factors at each sampling event, then fit a power-law function to the emission rates over
the time of the test to generate an empirical source emissions model that describes the changes of
emission rate with time for selected compounds. We also demonstrated the use of a mass
transfer-based model for the Phase IITCPP emissions data that does not rely upon chamber
concentration data.
5.1.1 Phase I Calculation of Isocyanate Emission Factors
The isocyanate concentrations in the air of the chamber during and following the application
were due to the gas and aerosol emissions from the stream of material that was sprayed from the
gun and emissions from the SPF insulation that formed on the substrate. The mass of emissions
determined from the Phase I chamber air concentration data was assumed to be the emissions
attributable to the application process. In the absence of sink terms (Equation 2) to account for
particle deposition, losses to chamber walls and chemical reactions, the emissions mass
determined by this process underestimates mass emitted. It also has the potential to overestimate
the mass attributable to the application process because the mass determined by integration of the
Phase I time-concentration data is attributed to the very short period of the spray events.
5.1.1.1 Estimation of Mass Emitted
Equation 3 was used to calculate mass emitted from the sum MDI time-concentration data. Mass
emitted was determined separately from concentrations determined by sampling with denuder-
filter samplers and with impingers. The difference between mass emitted by the two sampling
approaches, expressed as the relative percent difference, was 12% (2.5±0.3 |ig).
5.1.1.2 Emission Rate Model
The simulation software IECCU (Version 1.1) contains four application-phase source models.
Three of the models increment mass emissions on the basis of increments of area sprayed and
SPFI Methods Development
Page 129
-------
one, Model 21, increments mass emitted on the basis of time. Model 21, "rapid evaporation"
(Equation 11) requires no additional information outside of the application start and stop times
and determination of the mass of each chemical emitted (We) from the time-concentration data
generated in the chamber emissions test. The model is an empirical approach to generating an
emission rate from the chamber air isocyanate concentration data. The simplified application
phase model is:
R = w (for to
-------
Inspection of Figure 5-1 indicates that concentrations of isocyanates did not reach the observed
peak concentrations and decreased much more rapidly than would be predicted from air change
rate alone. Therefore, integration of the time-concentration profile underestimates the mass
emitted which is the basis for the calculation of emission rate and this may be one reason that the
model predictions appear to underestimate the peak chamber concentrations. A second reason is
that the simulation apportions the mass over the period of application and concentration decay
due to ventilation. Additional information is needed to improve the fit between the data and the
simulation, including loss rate terms of isocyanates due to chemical reactions and deposition to
surfaces.
700
600
500
CuO
c
a>
u
c
o
u
400
300
200
100
- Denuder/filter data
i
j
simulation
n\
7 \
X\\
Impinger data
simulation
* M V
% ii\
'/\\ A1 \\
¦
Denuder filter data
'/ \V/ v
'/ V
if
Impinger data
If
i
t
1
i
1 V
-
0.0
0.5 1.0
Elapsed Time (hours)
1.5
2.0
Figure 5-1. IECCU simulations of sum MDI Phase I concentrations
5.1.1.4 Direct Calculation of Emission Rates from Chamber Concentrations
The numeric solution to Equation 7 was used to calculate emission rates directly from the
chamber concentration data. The calculated emission rates were entered into IECCU as a text file
(.txt) along with chamber volume and air change rate. IECCU calculated predicted chamber air
concentrations based upon the data in the emission rate table. The Figure 5-2 shows simulated
sum MDI concentrations based upon the emission rates calculated from the denuder-filter
samplers and from the impinger samplers. The difference between the simulations shown in
Figures 5-1 and 5-2 is that for the simulation shown in Figure 5-1, that mass emitted based upon
integration of the time concentration data was apportioned to the length of each spray event.
There is insufficient data to generate emission rate tables for separate spray events. Also note
that differentiation to determine emission rates may result in a negative emission rate where
concentrations decrease very rapidly. Therefore, although the direct calculation method appears
to provide a better fit between the data and the IECCU predictions, the direct calculation method
does not incorporate area sprayed and rate of application.
SPFI Methods Development
Page 131
-------
OD
=L
700
600
500
~ Impinger — — Impinger Simulation
A Dry Sampler - — — -Dry Sampler Simulation
O
1
3
to
*
[ /
I
\\
/
/
V
v
w
I
J
A
B-
= 400
o
2 300
¦*->
C
-------
Ri(t) = Ai [M1k1e~klt + M2k2e~k2t]
Equation (12)
where:
Ri(t) = emission rate for an incremental area at time t (ugh1)
At = area of the incremental source (m2)
Mi = amount of chemical for rapid emission (jugm~2)
Mi = amount of chemical for slow emission (jugm~2)
k\ = first-order decay constant for rapid emission (If')
ki = first-order decay constant for slow emission (If')
t = elapsed time (/?)
to = time when the incremental area is applied (/?)
M0 = Mi + M2 and is the total emittable amount of the chemical available for emission during
Phase I. M0 is determined by integration of the Phase I time-concentration data using Equation 3.
The user employs a least-squares fitting routine to estimate values of the first order decay
coefficients, ki and k2. Due to the break in application between the second and third frames, the
application process was modeled as two separate applications and the predicted time-
concentration values were added together. Note that for HFC-134a the units of mass and
concentration in Table 5-2 and Figure 5-3 have been converted to mg and mg m"3.
The parameters and details of the application have been implemented within the simulation
program IECCU to predict the time-course of air concentrations in the full-scale chamber.
Varying the estimated proportion of mass available for "rapid" emissions impacts the estimates
of ki and £2 and impacts the fit between observed and model predictions. Steps for constructing
the source emissions model for the application and initial curing phase include:
• Determine the input parameters for fitting Equation 10 to the time-concentration data.
• At = area of the source (m2)
• Mo= total emittable amount of chemical available for emission during the application phase
determined as mass per unit projected surface area (|ig m"2)
• t = elapsed time (h)
• to = time when the application to the incremental area starts
• Utilize a software least squares fitting program to obtain estimates of ki and £2 through non-
linear least square fit of the equation to the time-concentration data.
• Determine the best estimate of % mass (Mo) available for "rapid" emissions.
• Mi = amount of chemical available for rapid emission (|ig m"2)
• M2 = amount of chemical available for slow emission (|ig m"2)
SPFI Methods Development
Page 133
-------
Model parameters ki and k} determined by least squares fitting of the experimental data using
Scientist® are provided in Table 5-2. The assumption that 95% of the mass of HFC-134a
emitted during Phase I was available for rapid emissions {Mi is 95% of Mo) provided the best fit
to the data. Assumptions of 90 and 99% available for rapid emission yielded estimates of ki or k -2
with standard deviations greater than the parameter estimates.
Table 5- 2 Model Input Parameters for HFC-134a Phase I Emissions
Mo
(mg rrr2)
ki±standard
deviation
k2±standard
deviation
Correlation
Percent of Mass
"Fast" Emissions
1.66E+05
75.6±6.6
0.63±0.27
0.995
95
The assumption that Mi comprises 95% of the application/curing phase emissions provides the
best fit between the predicted and observed chamber air concentrations for HFC-134a after 2
hours. Although the data demonstrate that with careful selection of parameters, the simulation
predictions closely match the observed chamber concentrations, these data do not address how
well the parameters scale between the test chamber environment and other environments.
A IECCU Prediction • Experimental Data
1.0E+05
1.0E+04
, 1.0E+03
1.0E+02
O
<_>
1.0E+01
1.0E+00
£
A
A
•
A
•
•
•
•
0.0
0.5
1.0 1.5 2.0
Elapsed Time (hours)
2.5
3.0
Figure
5-3. Observed and predicted Phase I concentrations for HFC-134a assuming Mi = 95% of M0
SPFI Methods Development Page 134
-------
5.1.2.2 Phase II Emissions Modeling for HFC-134a
The first order decay modeling approach demonstrated for the Phase I data can be used to
generate Phase II emission factors. However, as described earlier, the direct calculation approach
does not require estimation of model parameters and can be performed using an electronic
spreadsheet. The Phase II HFC-134a emission factors determined with the direct calculation
approach are plotted in Figure 5-4. The Y axis is presented in log scale to show the changes in
emission factor when the chamber air temperature increased to 26.9 °C and then returned to 23.9
°C between 435 and 535 h. The maximum emission factor of 325 mg m"2 h"1 was determined at
4.7 h, or about an hour after the chamber air flow was set to 0.4 h"1.
5.1.2.2.1 POWER FIT OF EMISSION FACTORS
A power equation was fit to the emission factor-time data in an Excel® spreadsheet. The form of
the equation is:
y = ci*x~b Equation (13)
where:
y = EF (mg m"2 h"1)
a = constant
x = elapsed time, t, (h)
b = exponent
The power law fit of emission fact versus time data, excluding the period of temperature
excursion, is shown on the graph. The plot demonstrates the decreasing emission factor with
SPFI Methods Development
Page 135
-------
time. The initial half-life (period-of-time in hours for the emission factor to decrease by 50%)
increases initially from 7 h to >350 h by the end of Phase II at 670 h.
GO
£
u
03
LL.
c
o
• EF calculated from chamber air concentrations
a EF at 26.9 C
Power (EF calculated from chamber air concentrations)
1000
100
00 . _
¦2 10
t
y = 1030.4X0-769
\
R2 =
0.9992
A*
*
0 100 200 300 400 500 600 700
Elapsed Time (hours)
Figure 5-4. Phase II emission factors showing impact of time and of the chamber air temperature
EXCURSION ON HFC-34A EMISSION FACTOR.
5.1.2.2.2 PREDICTION OF EMISSION FACTORS FROM SHORT-TERM DATA
The period-of-time for which power law fit is valid is unknown, however the experimental data
can be used to test the relationship over the period of the experiment. The equation for the power
fit of the emission factor versus time for the period time from 4.7 to 72 h is: EF = 1112.6x"0 797
with R2 of 0.9987. This equation predicts an emission factor of 6.2 mg m"2 h"1 after 669 h and
that is 3% lower than the observed emission factor of 6.5 mg m"2 h"1. This bias is well within the
uncertainty range of the measurement data and supports a hypothesis that short-term emissions
SPFI Methods Development
Page 136
-------
test data may predict longer term emissions factors for blowing agents for closed cell foam. The
ratio of predicted to observed emission factors at 164 and 348 h was 0.91 and 0.86, respectively.
5.1.2.3. Impact of Chamber Air Temperature on HFC-134a Emission Factor
At 442 hours after the start of the test, the HVAC system for the building housing the chamber
malfunctioned, and chamber air temperature rose from 23.9 to 26.9 °C in a 35-hour period. The
rise in temperature was accompanied by an increase in HFC-134a concentration. The emission
factors determined by the direct calculation from chamber concentration data are plotted against
chamber air temperature in Figure 5-5. Each data point represents a 5-hour average of
temperature or emission factor. The emission factors have not been adjusted for the decrease of
emission factor with time which would be about 6% over the period of temperature rise. The
emission factor increased by 41% as the chamber air temperature increased from 23.9 to 26.9 C.
A least-squares fit of the emission factor with temperature (r2 = 0.96) indicates that the HFC-
134a emission factors increased by 1.3 mg m"2 h"1 for each 1 °C increase in air temperature over
the range of the experimental data. The apparent linear fit may be a product of the limited
temperature range of the experimental data and the fact that the linear model does not include the
decrease of emission factor with time as shown in Figure 5-4. The increase of an HFC-134a EF
over the predicted EF at time of peak temperature (476 h) is closer to 48%. The relationship
between temperature and diffusion is well-studied in polymers and the relationship between
temperature and diffusion in a material is expected to follow an Arrhenius relationship, as
discussed by Zulu et.al. [29], IECCU Version 1.1 has built-in tools for estimating partition and
diffusion coefficients as a function of temperature. This data set may be useful for evaluating the
predictive temperature dependent functions built into IECCU.
SPFI Methods Development
Page 137
-------
• HFC-134a Linear (HFC-134a)
CO
CO
15
14
13
12
11
10
o 9
DO
E
u
a:
t i
*•
• * * *
•
* *
f
V
= 1.3425X-22.31
R = I
1.9601
8
& 7
6
23.5 24.0 24.5 25.0 25.5 26.0 26.5 27.0 27.5
Chamber Air Temperature (°C)
Figure 5-5 Impact of chamber air temperature on HFC-134a emission factor.
5.1.3 TCPP Emissions Modeling
During the spray application, TCPP was emitted to the air in gas and aerosol phases. As shown
earlier, TCPP deposited on all surfaces in the chamber, the exhaust duct filter, and surfaces of the
exhaust duct. The data in Table 5-3 indicate that about 50% of the Phase I TCPP was accounted
for in the sum of the TCPP leaving the chamber system in the exhaust duct air and collected on
the exhaust duct filter. Therefore, estimating emission factors from the chamber or exhaust duct
air concentrations will significantly underestimate the TCPP emission rates during Phase I.
Given that the largest percentage of the deposited TCPP was estimated from the deposition
coupons near the floor, it is likely that much of the deposition was associated with particle
deposition that occurred during and immediately after the spray event.
An empirical approach was utilized to model the Phase I emissions and empirical and mass
transfer approaches were employed to model the Phase II emissions.
5.1.3.1 Phase I TCPP Emissions Modeling
The mass of TCPP emitted during Phase I is summarized in Table 4-14. The data indicate that
roughly 50% of the mass of TCPP emitted during Phase I deposited on chamber surfaces and
PPE of the sampling team and 50% was exhausted from the chamber in the air of the exhaust
duct or was captured on the surfaces of the filter covering the exhaust duct opening. The Phase I
total mass emissions of TCPP determined by measuring concentrations on surfaces and in the air
of the chamber exhaust duct were utilized to estimate a constant emission rate term shown in
Table 5-3 by dividing the sum of mass determined in various compartments (|ig) with the total
SPFI Methods Development
Page 138
-------
time of application (h). Simulations were run with no sink terms which overestimate the air
concentrations.
Table 5- 3IECCU Phase ITCPP simulation inputs.
IECCU Source Type
21 (Instant Evaporation)
Chemical
Sum TCPP
Zone ID
1 (30 m3 chamber)
Constant Emission Rate (|jg lr1)
1.66E06
Start time (h)
0.021
0.256
End time (h)
0.146
0.366
Application
Frames 1 and 2
Frames 3 and 4
As can be seen in Figure 5-6, the simulation over-predicts the air concentrations during the
application due to lack of sink terms for gas and aerosol deposition. The detailed TCPP
deposition data may be useful for estimating particle deposition to chamber surfaces. However,
the TCPP deposition data does not differentiate between gas-phase and aerosol deposition so
additional information is needed to fully utilize this approach. What is clear is that surface
deposition has a significant impact on chamber air concentrations of TCPP during application.
O Chamber Concentration A Exhaust Duct Concentration -Simulation
7000
6000
5000
W)
3
c
o
4000
c
QJ
U
c
o
(J
3000
2000
1000
i
it
1\
i i
i i
i t
* : \
t\: \
11 • *
' \: 1
i
i V
i
' X
%
\
\
\
T
! $
\
*
\
V
£>
i
i
«
\
\
\
I 4 i
1 A £
k—
4 ® A
—ft—*
-A-—A
0.0
0.5
1.0
2.5
3.0
3.5
1.5 2.0
Elapsed Time (hours)
Figure 5-6 Phase I IECCU simulated TCPP chamber concentrations shown with chamber and exhaust duct
AIR CONCENTRATIONS.
5.1.3.2 Phase II TCPP Emission Rate Modeling
Two approaches for modeling Phase II TCPP emissions are presented. Emission rates and
emission factors were calculated directly from the chamber air concentrations using the numeric
SPFI Methods Development
Page 139
-------
solution to Equation 7. Also, use of the modified space-state diffusion model, Equations 12 and
13, to predict chamber concentrations with the simulation system IECCU is presented and
discussed. As with the Phase I modeling, there is no attempt to account for emissions from
surfaces in the chamber so assigning the emissions to the source overstates the emission factors
for the SPF insulation.
5.1.3.2.1 PHASE II TCPP EMISSION FACTORS DETERMINED BY DIRECT CALCULATION FROM CHAMBER
CONCENTRATIONS
Area specific emission factors were calculated from TCPP Phase II chamber concentrations
using the direct calculation method. The emission factors (|ig m"2 h"1) are plotted versus elapsed
time in Figure 5-7. The peak Phase II emission factor observed was 320 |ig m"2 h"1 at elapsed
time of 4.96 h, or about 1.5 h after the chamber air change rate was decreased from 4.1 to 0.4 h"1.
Except for the period of the temperature excursion, the emission rate decreased over the Phase II
period, although the rate of change was slow after 200 h.
5.1.3.2.2 IMPACT OF TEMPERATURE ON EMISSION RATE
A power equation was fit to the TCPP emission factors (Figure 5-7), exclusive of the data point
at the elevated temperature. The emission factor of 173 |ig m"2 h"1 determined from the
concentration (EF = C*Q/A) determined at the elevated chamber temperature of 26.9 °C is a
54% increase over the emission factor of 112 |ig m"2 h"1 predicted at 480 h by the power fit
equation. Thus, a three degree increase in chamber air temperature appears to have resulted in a
>50% increase in the TCPP emission rate.
It is important to note that the increased chamber air concentration of TCPP and apparent
increase in emission rate are likely due to increased emissions of TCPP from the walls, ceiling,
and floor of the chamber as well as emissions from the sprayed foam. Therefore, additional
information is needed to evaluate the impact of the secondary emissions and the impact of the
temperature increase on TCPP emission rates from the foam.
The recent publication by Liang [36] investigated the relationship between emissions of TCPP
from polyisocyanate foam in microchambers. The study demonstrated a nonlinear impact of
temperature on the diffusion parameter (l)m) and partition coefficient (Kma). The apparent
increase in TCPP emission factor due to the unplanned rise in chamber temperature in this
experiment may be useful in the evaluation of the temperature correction for Dm andiC„M inputs
to the diffusion model in IECCU.
SPFI Methods Development
Page 140
-------
O SumTCPP ~ TCPP Emission Factor (26.9 C) Power (Sum TCPP)
350
300
E
j£ 200
L_
o
tS 150
ro
U—
o 100
LO
U)
£ 50
0
0 100 200 300 400 500 600 700
Elapsed Time (hours)
o
y = 410.51)
(-0.21
1
Ft2 = 0.9707
c
o
° -o-]
A
o
o
G
O
O
O
O
Figure 5-7 Power law Fit of Phase II TCPP emission factors determined with the direct calculation
METHOD.
5.1.3.2.3 PHASE II TCPP EMISSIONS MODELING USING THE MODIFIED SPACE-STATE DIFFUSION MODEL IN
IECCU (EQUATIONS 10 AND 11)
Bevington applied the modified state-space method for prediction of emissions of an amine
catalyst and the flame retardant TCPP from SPF insulation [25], An objective of this experiment
was to provide a data set for calibration of the modified space-state diffusion-based source model
built into IECCU. Three parameters for the modified space-state model include initial
concentration of the chemical in the material (Co), the diffusion coefficient of the chemical in the
material (Dm), and the material-air partition coefficient (Kma). The general form of the source
model is:
E = Ha ~ c) Equation (14)
Ha = f(h, Kma, Dm) Equation (15)
Where:
Remission factor (|ig m"2 h"1)
Ha = overall gas-phase mass transfer coefficient (m h"1)
Cspf= Concentration in the surface layer of SPF insulation (|ig m"3)
C = concentration of the chemical in the air (|ig m"3)
Kma = material air partition coefficient (dimensionless)
SPFI Methods Development
Page 141
-------
h = gas-phase mass transfer coefficient, (m h"1), and
Dm = solid-phase diffusion coefficient, (m2 h"1).
The variables in Equation 15 are a function of the specific material and vary with temperature.
Bevington et., al. provides guidance on techniques for parameter estimation [26], The
relationship between temperature and the diffusion coefficient is also discussed by Bevington
and the algorithm has been built into Version 1.1 of IECCU. Inputs to the diffusion source model
include the volume and ventilation for the modeled space, area of the source, initial
concentration of the chemical in the material (Co), the gas-phase mass transfer coefficient (/?), the
diffusion rate of the chemical in the material (Dm), and the material-air partition coefficient
(Kmc), both of which are temperature dependent.
Figure 5-7 shows IECCU predictions of Phase II TCPP concentrations for the 30 m3 emissions
chamber with ventilation rate 0.4 h"1. Co was determined to be 4.5 E+09 |ig m"3 by extraction of
TCPP from cores taken from the foam and h was estimated at 1.785 m h"1 using the method of
Sparks (Params, Version 1.1) from measurements of air speed near the surface of the foam at the
end of the test. The initial TCPP concentration in the chamber air at 3.8 hours was estimated to
be 89 |ig m"3 based upon IECCU predictions using the application phase dual first order decay
model with supply air flow between 3.3 and 3.7 hours computed from air change rates estimated
from the tracer decay concentration. The simulated concentrations based upon the 2017 estimates
of Bevington et. al., and those estimated for this data set are shown in Figure 5-7 along with the
TCPP air concentrations determined by sampling from the chamber side port. Sorption
parameters were not included in the model so the differences between model predictions and
observed concentrations may be misleading.
The simulation using Bevington et.al. parameters predict that the low-flow post application
Phase I concentrations peak 12.4 hours after application which is about 2 h after the time
determined from the chamber concentration data. The partition coefficient, Km estimated by
Bevington et al., appears to overestimate the peak post-application concentrations and
underestimate the concentrations after about 200 hours. However, the impacts of deposition and
sorption on peak concentrations and impacts of emissions from deposition on chamber surfaces
and from overshoot are not accounted for in Figure 5-8. As discussed in Section 4.4.8, the
apparent TCPP emission rate from the chamber system during Phase III implies that 21% of the
emissions may be due to emissions from chamber surfaces. Thus, the long-term concentrations
based upon on the parameters estimated by Bevington et.al. may be much closer to the long-term
data than they appear in Figure 5-8.
SPFI Methods Development
Page 142
-------
W)
1
c
c
c
c.
A Chamber Concentration Data
350
300
250
200
150
100
50
0
Bevington (2017)
¦D & K to fit data
*
**
: V
Bevington et. al (2017
Kma = 1.23E06
i •.
iv
t
;
A A A
A
Kma = 1.5E07
100
200 300 400 500
Elapsed Time (hours)
600
700
Figure 5-8IECCU model output overlaid on chamber concentration data. Diffusion and partition
COEFFICIENTS ESTIMATED BY BEVINGTON (2017) AND ESTIMATED TO FIT THIS DATA SET
Other challenges to use of the modified space-state approach include obtaining estimates of Dm
and Kma. IECCU Version 1.1 provides five methods (Huang et al., 2017, Baner et al. 1994,
Begley et al., 2005, Deng et al., 2009, and Millington et al., 1961) for estimation of the diffusion
coefficient An. Each method requires experimentally determined coefficients for diffusion of a
specific chemical in a specific material. Lookup tables are provided for Dm where data is
available from the literature. Coefficients for diffusion of TCPP in low density PUF insulation
and for diffusion of TCPP in polyisocyanurate foam are provided however, Version 1.1 of
IECCU does not provide estimates of diffusion coefficients for TCPP in medium density SPF
insulation.
Two methods, Guo (2000) and Zhang (2006), are provided in IECCU for the estimation of the
material-air partition coefficient, Kma. The method or Guo requires input of the vapor pressure at
the relevant temperature. Reported vapor pressure values of TCPP vary over orders of
magnitude. The method of Zhang computes K,„a for a chemical-material combination utilizing
two experimentally derived material-specific constants and the temperature (K), however there
are currently no constants provided for TCPP and medium density SPF insulation. For the EU
reported vapor pressure of TCPP, (1.05E-05 mm Hg) the method of Guo estimates Kma to be
5.16E+07. Simulations employing this estimate (not shown) of K,„a did not track the time-
concentration profile of the data set.
In summary, the diffusion model tracks the pattern of the long-term concentrations. Additional
source terms are needed to address impacts of TCPP emissions from particle deposition,
SPFI Methods Development
Page 143
-------
sorption, and overshoot on long-term emissions as well as material specific estimates of Dm and
Kma-
5.1.4 PMDETA Emissions Modeling
The PMDETA emission rates for Phase I and Phase II were determined from the chamber air
concentration data. In terms of determining mass emitted, it is not clear what impact deposition
may play during Phase I. PMDETA was quantified in extracts of the Phase I exhaust duct filter
and extracts of the PPE worn by the spraying/sampling personnel. However, if PMDETA was
present in extracts of the Teflon filters in the modified OVS or in extracts from the deposition
and wipe samplers, it was below levels of detection.
5.1.4.1 Phase I PMDETA Emissions Modeling
Two approaches are demonstrated for generating Phase I emission rates for PMDETA from
chamber concentrations, the dual first order decay model, and the method of direct calculation of
emission rates from chamber concentration data.
5.1.4.1.1 PHASE I PMDETA EMISSIONS MODELING WITH DUAL FIRST ORDER DECAY MODEL
During the application, PMDETA concentrations in the air of the chamber were observed in the
range of 11,000 to 13,000 |ig m"3. Concentrations in the air determined by sampling from the
exhaust duct during the spray events were not used for determining model parameters for
Equation 10 because mass amounts on some of the multi-bed sorbent traps were above the
calibration range of the TD-GC/MS. The estimate of mass emitted per unit area sprayed during
Phase I, determined using the chamber air concentrations and Equation 3 is shown in Table 5-4.
Total mass emitted per area sprayed (Mo) during Phase I was estimated to be 1.34E+05 |ig m"2by
Equation 3. The first-order decay constants for the source emissions model were determined by
non-linear least squares fit assuming 10 and 15% PMDETA mass available for "slow" emissions.
As seen in Figure 5-9, the dual first order model provides a reasonable fit to the data, however
neither fit captures the concentrations determined at 1.1 and 2.8 h. The concentrations are shown
in log format so that the impact of assumptions regarding mass available for "slow" emissions
are evident. The model fits the data well except for the chamber concentrations determined at 1.1
h and 2.8 h where the model predictions are 8 and 52% higher and 37 and 51% lower than
observed, respectively. The method for estimating the first order decay constants is based upon
the assumption that the mass emitted during Phase I can be evenly apportioned to the two spray
events that were separated in time by sample collection in the chamber. Since the application
went off-ratio towards the end of the second application event, the assumption of constant
emission rate during Phase I may not be valid for PMDETA.
SPFI Methods Development
Page 144
-------
Table 5- 4 Model parameters for fitting the dual first order equation to the application phase emissions
Compound
Mo
(M9 m-2)
/oistandard
deviation*
foistandard
deviation*
Concentration-
time
Correlation
Percent of
Mass "Rapid"
Emissions
PMDETA
1.34E+05
4.44E+01±9.5
0.16±0.21
0.996
90
4.77E+01±8.7
1.0±0.42
0.998
85
* Scientist® 3.0.
A PMDETA Experimental Data
— • -Model 10%Slow Emissions
— • Model 15% Slow Emissions
100000
ST 10000
E
ClO
3
c 1000
o
ro
i—
-I-'
S 100
u
c
o
u
10
1
0 0.5 1 1.5 2 2.5 3 3.5
Elapsed Time (hours)
/ S
• %
s.
1 >
• —s.
Figure 5-9 Phase I predicted and observed chamber concentrations for PMDETA assuming 10% or 15% of
APPLICATION PHASE EMISSIONS (M0) ARE DUE TO SLOW DECAY (K2)
5.1.4.2 Phase II PMDETA Emissions Modeling
The Phase II PMDETA emission factors were calculated by Equation 9 in an electronic
spreadsheet. The calculated emission factors are plotted for the Phase II in Figure 5-10. A power
fit of the data, excluding the data point where chamber temperature was elevated is plotted with
the calculated emission factors.
5.1.4.2.1 IMPACT OF TEMPERATURE ON PMDETA EMISSION
The equation for the power fit was solved for emission factor at 479 hours when the chamber
temperature reached 26.9 °C to obtain an estimate of predicted emission factor at 23.9 °C. The
emission factor determined by direct calculation at 479 h (107 |ig m"2 h"1) from the chamber
concentration is 45% greater than the emission factor predicted by the power fit equation (74 |ig
SPFI Methods Development
Page 145
-------
m~2 h"1). The difference between observed and predicted emission factors is not apparent because
of the large range of the Y axis.
PMDETA A PMFETA (27 C)
1400
*—i
-«= 1200
fN
I 1000
r 8oo
o
£ 600
U—
| 400
CO
¦| 200
LU
0
Power (PMDETA)
L
»
y = 3088.5x 0-604
ii
= 0.9763
T
•—
A
••
0 100 200 300 400
Elapsed Time (hours)
500
600
700
Figure 5-10 Power fit of PMDETA emission factors determined by direct calculation from chamber
CONCENTRATION DATA
Emission factors at 168, 383, and 669 h were estimated by power fit of the first 5 data points
representing 75 h test duration. The equation for power fit of the truncated data predict emission
factors at 168, 383, and 669 h of 189, 127, and 97 |ig m"2 h"1, respectively. These estimates are 41,
48, and 37% above the emission factors calculated directly from the chamber concentration data.
5.1.5 Emissions Modeling of 3-Chloropropene, 1,2-Dichloropropane, 1,4-Dioxane,
and Chlorobenzene
Phase I and Phase II emission rates were calculated from the emissions test chamber
concentration-time data using the concentrations determined by TD-GC/MS of multi-bed sorbent
samples as presented earlier in Figures 4-14 through 4-21.
5.1.5.1 Phase I Emissions Modeling of 1,2-Dichloropropane, 1,4-Dioxane,
and Chlorobenzene
The dual first order decay model was fit to the Phase I chamber concentration data for each
compound with a least squares fitting program (Scientist®) to estimate model parameters ki and
k2 of Equation 7. To extend the time period for integration of mass emitted (Mo) using Equation 3
to the time at which the spraying/sampling personnel exited the chamber at the end of Phase I,
SPFI Methods Development
Page 146
-------
the chamber concentration at 3.3 h was estimated using a power law fit of the time -
concentration data for each compound (see Appendix E for examples).
The Phase I mass emitted (Mo) determined for each compound by applying Equation 3 to the
chamber concentrations determined from the samples collected through the chamber port and
normalized by area of application (7.04 m2) are presented in Table 5-5. Also presented in Table
5-5 are the first order decay constants ki and k.2 estimated by fitting Equation 7 with a least
squares fitting program (Scientist®) to the Phase I concentration - time data for each compound.
Estimated masses emitted in Phase I ranged from 0.87E+03 |ig m"2 for chlorobenzene to
20.1E+03 |ig m"2 for 1,4-dioxane.
As is evident from the estimates of standard deviations output by the statistical report in the
Scientist® software system, the uncertainties in the decay constants are in some instances, larger
than the estimate of the parameter. Several attempts were usually needed to find the percent of
Mo attributable to ki for "rapid emissions" so that the predictions fit the lower concentrations for
the time after one hour. This was determined by judgement. The relatively high correlations
between the experimental data and fitted model concentrations, except for chlorobenzene,
demonstrate that the dual first order decay model has the potential to provide reasonable fits to
the data during the period of application.
The standard deviation for the ki estimate for chlorobenzene indicates a serious problem with
fitting the model to the data. Iterations were run with percent rapid emissions estimates for Mo
ranging from 25 to 95%. Although the software returned estimates for ki and k'2 for each
iteration, the program was unable to successfully complete the statistical analysis to output
standard deviations for the parameters except for the 50% iteration.
To demonstrate the use of the source model (Equation 7) to predict concentrations in an
environment, the estimates of Mo and the decay constants determined by fitting the data to
Equation 7 have been employed in the simulation system IECCU, Model 23, to predict
concentrations in the emissions test chamber. Plots of Phase I chamber concentrations plotted
along with the concentrations predicted by IECCU for each VOC are shown in Figures 5-10
through 5-13. The Phase I plots are not smooth due to the six-minute pause in application to
harvest the first set of impingers, denuder/filter samplers, and OVS samplers between application
to the first two frames and the last two frames. This results in two discrete application periods
and two peaks in concentration.
Table 5- 5 Model parameters for fitting the dual first order equation to the Phase I emissions
Compound
Mo
(M9 m"2)
/oistandard
deviation*
foistandard
deviation*
Concentration
-time
Correlation*
Mo%
Rapid
Emissions
3-Chloropropene
5.4E+03
511±1367
1.05±0.40
0.982
70
1,2-Dichloropropane
2.7E+03
304±281
0.573±0.350
0.995
80
1,4-Dioxane
20.1 E+03
88.2±29.7
0.26±0.1.1
0.996
95
Chlorobenzene
0.87E+03
5.0E03±9.6E05
0.491 ±0.395
0.950
50
*Output by Scientist® regression and statistics report
SPFI Methods Development
Page 147
-------
1 Mo determined by Equation 3 from chamber concentrations determined by sampling at chamber
side port.
Except for chlorobenzene, the IECCU simulations appear to provide a reasonable representation
of the concentrations determined by sampling from the chamber side port during Phase I. The
simulations also provide insight into potential peak concentrations during the application.
SPFI Methods Development
Page 148
-------
O Experimental Data IECCU Simulation
o
500
450
400
3 50
300
250
200
150
100
50
0
4
''
L' >
%
%
iV
%
\
\
\
X
X
—
0.0 0.5 1.0 1.5 2.0
Elapsed Time (hours)
2.5
3.0
3.5
Figure 5-11 Phase I IECCU simulation of 3-chloropropene concentrations assuming 70% of the emissions
WERE AVAILABLE FOR "RAPID" EMISSION (Mi = 70% OF M0)
¦ IECCU Simulation
O Experimental Data
BOO
250
co
E
M 200
I 150
c
QJ
u
c
o
(J
100
50
A
II
1 1
, /}
*\ * I1
;\ i *
V
i
*
\
i X
1
*
t
i
\
\
\
\
\
\
f
N
N
0.0 0.5 1.0 1.5 2.0
Elapsed Time (hours)
2.5
3.0
3.5
Figure 5-12. Phase I IECCU simulation of 1,2-dichloropropane concentrations assuming 80% of the
EMISSIONS WERE AVAILABLE FOR "RAPID" EMISSION (Mi = 80% OF M0)
SPFI Methods Development
Page 149
-------
A Experimental Data IECCU Simulation
2500
2000
W)
3 1500
£=
O
£ 1000
-------
60
ro
50
E
ClO
3
40
£=
o
30
no
i_
c
20
aj
u
c
o
10
(J
0
O Experimental Data (Port CS-11P4)
• — IECCU Simulation
A Scientist calculated concentrations
4
T °
L *
9 *
A
/ ~
~ —
11—
~
-------
Figure 5-15 illustrates the use of IECCU to simulate concentrations in a modeled environment
from emission rates determined from direct calculation of chamber concentration data. Note that
there is one less emission rate calculation than the number of samples. Also, the emission rates
during the application do not relate the increase in emission rate to the increase in sprayed area
as the application proceeds.
5.1.5.2 Phase II Emission Factors and Emissions Modeling of 3-Chloropropene, 1,2-
Dichloropropane, 1,4-Dioxane, and Chlorobenzene
The Phase II chamber concentrations starting with the samples collected from the chamber side
port at 5 h were utilized in Equation 9 to calculate emission factors at each Phase II sampling
period. A Power fit of the emission factors versus time was generated in an electronic
spreadsheet to create models of emission factors as a function of time. Emission factors and
power fit of the emission factors calculated from 10 h until the sprayed frames were removed
from the chamber are shown in Figures 5-16 through 5-19.
O 3-Chloropropene A 3-Chloropropene (26.9 C)
Power (3-Chloropropene)
100
90
i 80
rsi
E 70
CuO
3 60
% 50
tc
c 40
o
30
.52
w 20
10
0
o
1 •
o
y = 129.66x"0,217
R2 = 0.908
o
o
o
o
A
oo o
100 200 300 400 500 600 700
Elapsed Time (hours)
Figure 5-16 Phase II emission factors calculated for 3-chloropropene
SPFI Methods Development
Page 152
-------
O 1,2-dichloropropane A 1,2-dichloropropane (26.9 C)
Power (1,2-dichloropropane)
25
20
GO
3 15
o
u 10
TO
LL_
o 5
1/5
to
E 0
o
v = 33.:
3 3 6x 0-378
Q
°>
M
II
D.S945
O""
-O-Q--0.
Q
p
0 100 200 300 400 500
Elapsed Time (hours)
600 700
Figure 5-17 Phase II emission factors calculated for 1,2-dichloropropane
O 1,4-Dioxane ~ 1,4-Dioxane (26.9 C) Power (1,4-Dioxane)
140
_120
T—I
Jc
V 100
60
u
tc
80
60
40
20
0
o
o
y = 6 74x 0,826
R2 = 0.99
o
o
~
o
100
200 300 400
Elapsed Time (hours)
500
600
700
Figure 5-18 Phase II emission factors calculated for 1,4-dioxane
SPFI Methods Development
Page 153
-------
O Chlorobenzene Power (Chlorobenzene)
_ 5-°
^ 4.0
r 3.0
o
¦M
£ 2.0
c
o
£ 1.0
£
LU
0.0
0 100 200 300 400 500 600
Elapsed Time (hours)
Figure 5-19 Phase II emission factors calculated for chlorobenzene
The equations for the power fit of emission factors for each compound are displayed in the
graphs. The concentrations determined during the period of the temperature excursion where
chamber air temperature reached 27 °C were not utilized. Correlations between data and fitted
model range from 0.99 to 0.57 are >0.9 for all compounds except chlorobenzene where Phase II
chamber concentrations ranged from a high of 2.7 |ig m"3 at 10 h to 1.5 |ig m"3 at 669 h.
Concentrations of all of the compounds dropped below LOD for each compound except TCPP
following removal of the sprayed frames from the chamber. Within the context of the air change
rate, loading factor and length of the experiment, the power fit equations may be useful in
calculating time needed for an emission factor to drop below a specific rate.
Table 5-6 compares Phase II average emission factors calculated by dividing mass emitted per
m2 in Tables 4-9 and 4-10 by the total hours of Phase II expressed in units of |ig m"2 h"1.
Table 5- 6. Comparison of average Phase II emission factors determined from chamber and exhaust duct
PORT SAMPLING LOCATIONS
Phase II
Average EF
(|jg m 2 h1)
3-Chloro-
propene
1,2-Dichloro-
propane
1,4-Dioxane
Chloro-
benzene
PMDETA
Chamber
41.9
4.4
11.0
2.5
133
Exhaust Duct
39.5
4.2
10.8
2.3
128
Average
40.7
4.3
10.9
2.4
130.5
RSD (%)
6%
5%
2%
8%
4%
O
y = 4.9416x°-
r2 = n
129
¦.
¦ O
O
SPFI Methods Development
Page 154
-------
As seen in Figures 4-18 through 4-21, there are no apparent differences between Phase II
concentrations of 1,2-dichloropropane and 1,4-dioxane determined from chamber and exhaust
duct samples. Concentrations of PMDETA appear higher in the chamber relative to the exhaust
duct for samples collected in the first 24 hours of the post-application phase. These differences
may be due to initial uptake of PMDETA by the exhaust duct filter. Differences in chamber and
exhaust duct concentrations of TCPP appear to persist for 167 hours or more, indicating TCPP
uptake by the exhaust duct filter and sorption by chamber and exhaust duct surfaces.
Other empirical approaches such as the first order decay model or mass transfer models could
also be employed to create source models for the VOCs emitted during Phase II. Use of the mass
transfer models require determination of emittable mass in the source at the start of the Phase II,
and partition and diffusion coefficients for each compound in the specific material. See for
example, Zhao, Little, and Cox, "Characterizing Polyurethane Foam as a Sink for or Source of
Volatile Organic Compounds in Indoor Air" [37], Given the expense of full-scale chamber
emissions tests, use of the diffusion model approach for predicting long-term emissions for
exposure evaluations of product emissions makes financial sense. Use of emissions factors
generated in chamber emissions tests can provide emissions data for evaluation of the efficacy of
model parameters and assumptions.
5.1.6 Emissions Modeling of Aldehydes and Ketones
Emission rates were calculated from the Phase I and Phase II chamber concentrations presented
in Figures 4-22, 4-23, and 4-24.
5.1.6.1 Phase I Modeling of Aldehydes and Ketones
The concentrations of acetone, formaldehyde, acetaldehyde, and propionaldehyde were
quantified in the Phase I emissions. The Phase I mass emitted calculated with Equation 3,
emission rate, expressed as |ig h"1 and |ig m"2, are presented in Table 5-7. The emission rates
expressed in terms of the time of application (0.235 h) could be used as inputs to IECCU using
the instant evaporation Model 21 (Equations 6). Alternatively, the emission rate per unit area
could be utilized as Mo in Equation 7 to generate first order decay terms using a least square
fitting program.
Table 5- 7 Phase I Emission rates for selected aldehydes and ketones
Compound
Mass Emitted
Emission Rate
M9
M9 h"1
M9 m"2
Formaldehyde
8.1E+03
3.5E+04
1.2E+03
Acetaldehyde
4.2E+03
1.7E+04
5.8E+02
Acetone
3.1E+04
1.3E+05
4.4E+03
Propionaldehyde
1.4E+04
5.8E+04
1.9E+03
SPFI Methods Development
Page 155
-------
5.1.6.2 Phase II Emissions Modeling of Aldehydes and Ketones
Isovaleraldehyde (3-methylbutanal), acetaldehyde, formaldehyde, and hexaldehyde were
quantified during Phase II. Formaldehyde and hexaldehyde concentrations were below the low-
flow background concentrations at every sampling period and acetaldehyde concentrations were
below the low-flow chamber background concentrations within 200 h. Emission factors were not
calculated from the Phase II time-concentration data for those compounds.
Isovaleraldehyde was not identified in chamber background samples or in the Phase I emissions.
Concentrations during Phase II averaged 11.8±0.8 |ig m"3 from 48 h through 380 h prior to the
rise in chamber temperature. An emission factor of 20.0±1.4 |ig m"2 h"1 is determined by the
method of direct calculation from the average chamber concentration. The apparent
isovaleraldehyde emission factor determined from the chamber concentration determined at the
peak of the unplanned chamber temperature rise is 31 |ig m"2 h"1 which implies that the emission
factor increased of 55% due to the 3 °C rise in chamber air temperature.
5.2 Impact of Chamber Air Temperature Emission Factors
The power-fit equations shown in Figures 5-3, 5-6, 5-9, and 5-14, generated by fitting the Phase
II emission factors, exclusive of the emission factor for the 480-h sampling time where chamber
temperature increased to 26.9 °C, were employed to predict the emission factor at 480 h with no
temperature rise for compounds listed in Table 5-8. The emission factors at 26.9 °C were
determined from the chamber concentrations using the direct calculation method, Equation 6.
The table presents the estimated percent increase in emission factor due to the temperature
increase from 23.8 to 26.9 °C for the selected compounds. Estimates of impact of temperature on
emission factor are not provided for 1,2-dichloropropane, 1,4-dioxane, and chlorobenzene as the
apparent concentration changes (see Figures 4-19, 4-20, and 4-21) were in the range of the
uncertainty of the concentration measurements.
Table 5- 8 Estimated impact of 3°C temperature rise on emission factors for selected compounds
Compound
*EF Predicted
at 480 Hours
(|jg m 2 h1)
**EF Calculated
with Equation 6
(|jg m 2 h1)
EF Increase
at 26.9 °C
(%)
Vapor Pressure
(mm Hg)
At 25 °C
HFC-134a
9.0E+03
13.1 E+03
46
5.0E+03
3-Chloropropene
34
47
38
368
PMDETA
71
105
46
0.3 (21 °C)
***TCpp
112
173
54
9.23 E-03
*EF predicted at 480
l assuming no temperature change with the power fit of emission factors
through 670 hours.
**EF at 480 hours at chamber temperature of 26.9 °C determined from concentration using the
direct chamber calculation method.
***TCPP emission factor includes emissions from chamber surfaces.
5.3 Emission Factor Predictions from 72-hour Test Data
A power law model was fit to the Phase II emission factors calculated through 72 h for the
compounds listed in the table below. Emission factors predicted by the 72-hour power law
SPFI Methods Development
Page 156
-------
equation at 168, 334, and 669 h for selected compounds were compared to the emission factors
determined from the chamber data at those selected times. The ratio of predicted to observed
emission factor and parameters for the power law fit are presented in Table 5-9.
Table 5- 9 Comparison of predicted emission factors calculated from power fit of Phase II emission
FACTORS THROUGH 72 H TO EMISSION FACTORS DETERMINED FROM THE EXPERIMENTAL DATA AT THOSE TIME PERIODS.
Hours
Power law parameters 72-h fit
168
334
669
Constant
Exponent
Correlation
Compound
Ratio Predicted to Observed
a
b
R2
HFC134a
0.91
0.86
0.97
1113
-0.797
0.9987
TCPP
0.92
0.91
0.98
441
-0.237
0.9552
PMDETA
1.41
1.48
1.37
2272
-0.485
0.9311
3-Chloropropene
0.76
0.62
0.55
230
-0.395
0.9814
1,2-
0.77
0.53
0.44
87
-0.662
0.9825
Dichloropropane
1,4-Dioxane
0.79
0.62
0.51
1256
-1.014
0.9944
Chlorobenzene*
0.84
0.62
0.59*
8.7
-0.301
0.8355
Average
0.91
0.81
0.77
Standard Deviation
0.23
0.33
0.34
*Last emission factor at 550 hours
For the VOCs 3-chloropropene, 1,2-dichloropropane, 1,4-dioxane, and chlorobenzene, the bias
between predicted and observed increases with time beyond the 72-h data set used to create the
emission factor model.
The bias for HFC-134a, TCPP, and PMDETA are fairly constant and do not appear to follow the
pattern of the VOCs.
The data suggests that emission factors determined from a 72-h test could be used to make
reasonable emission factor predictions for a seven-day period following application.
This experiment amounts to a single data point and additional experimental data would be
required to evaluate this observation.
6.0 Summary and Conclusions
6.1 Summary of the Experimental Process
A proof-of-concept experiment was conducted to investigate the feasibility of characterizing
application and post-application emissions from SPF insulation in a full-scale test chamber. The
experiment quantified emissions of the isocyanates, MDI, p3-MDI, and p4-MDI, the blowing
agent HFC-134a, various VOC emissions including several chlorinated volatile compounds, 3-
chloro-l-propene, 1,2-dichloropropane, 1,4-dioxane, chlorobenzene, the amine catalyst,
pentamethyl-diethylenetriamine, the flame retardant tris (l-chlor-2-propyl) phosphate, and
SPFI Methods Development
Page 157
-------
several aldehydes. Particle emissions were monitored during the application of the spray foam to
plywood substrates that were placed against walls of the chamber. The spray application was
completed in two spray events, each lasting less than eight minutes with a 6-minute period
between spray events for sample collection. Air and surface samples were collected to quantify
concentrations of emissions in three phases; Phase I, during and for three hours following the
application, Phase II, for four weeks following Phase I, and Phase III for 12 days following
removal of the sprayed frames from the chamber. The Phase I chamber air change rate of 4.1 h"1
was reduced to 0.4 h"1 at the start of Phase II and remained at that rate until the end of the test.
Concentrations of emissions in the air of the chamber and the chamber exhaust duct were
determined by sampling with various instruments and sampling media. Concentrations of the
flame retardant on chamber surfaces were determined at the end of each phase by extraction of
TCPP from deposition samplers. TCPP mass collected on the MERV 13 HVAC filters covering
the opening to the exhaust duct was determined by extraction from sections cut from the filters.
Wipe samples were collected at the end of the test to estimate flame retardant concentrations on
surfaces of sampling equipment and fans in the chamber and on inner surfaces of the exhaust
duct. Samples were collected from the PPE of the spraying/sampling personnel at the end of
Phase I to estimate the amount of TCPP deposited on PPE.
6.1.1 Summary of the Chamber Environmental Conditions
Air change rate averaged 4.1±0.06 h"1 during Phase I, and 0.4±0.006 h"1 during Phases II and III.
Air temperatures in the chamber rose from 23.7±0.3 prior to the spray event to nearly 30 °C
during the application process and dropped to <26 °C within three quarters of an hour of
completion of the spray events. The chamber air temperature dropped by 0.5 °C when the
personnel exited the chamber at the end of Phase I.
The chamber air temperature averaged 25.4 ±0.5, 24.2±0.6, and 23.7±0.1 °C (average ± standard
deviation) during Phases I, II, and III, respectively. Relative humidity averaged 40.6±1.6,
39.0±2.0, and 39.7±6.7% during Phases I, II, and III, respectively. At 440 h the temperature in
the chamber began to increase from 23.8 to 26.9 °C during a 35-hour period due to malfunction
of the air conditioning system of the building that houses the chamber. The experiment was
extended to allow the chamber air temperature to stabilize at target temperatures once the
building air conditioner was repaired.
6.1.2 Summary of the Emissions Source
Average maximum foam temperatures measured with thermocouples placed at the mid depth of
the foam varied between 95 and 112 °C for three of the four substrates. The spray kit
malfunctioned during application to the fourth substrate frame and application was halted prior
to spraying the fourth quadrant of that frame. The average maximum temperature of the foam
applied to the fourth frame did not reach 75 °C and the physical characteristics of the foam in the
second and third quadrants of the fourth frame were characteristic of off-ratio foam.
SPFI Methods Development
Page 158
-------
6.1.3 Summary of Phase I Air Sampling to Characterize Emissions
The application to the four substrate frames occurred as two events where two frames were
sprayed, and samples were collected by the spray team prior to spraying the second two frames.
During the break in spraying the sampling team time collected isocyanate and modified OVS
samplers and prepared the samplers for collection of emissions during the second spray event.
Other sampling support team members positioned outside of the chamber collected air samples
from the chamber and from the chamber exhaust duct by periodically placing multi-bed sorbent
traps, and DNPH samplers into the chamber through ports in the side wall and ceiling, and from
a port in the exhaust duct. The support team also operated gas and particle analyzers positioned
outside of the chamber. Three grab samples of 100 mL volume were collected with a 100 mL gas
sampling syringe during the spray events due to concern for overloading the multi-bed thermal
desorption traps.
6.1.4 Summary of Phase II Air Sampling to Characterize Emissions
Once the sampling team exited the chamber, the flow control system was re-configured to
control flow at 0.4 ACH. Periodic sampling from the chamber and exhaust duct with multi-bed
sorbent traps, DNPH samplers, and modified OVS sampling media continued until 670 h, at
which time personnel entered the chamber, harvested deposition samplers, collected surface wipe
samples, replaced the HVAC filter, and removed the sprayed frames from the chamber.
6.1.5 Summary of Phase III Air Sampling to Characterize Emissions
Sampling from the chamber and exhaust duct resumed after removal of the frames from the
chamber and was continued for 290 h to monitor concentrations of the flame retardant in the air
of the chamber and the exhaust duct.
6.2 Summary of the Results
Concentrations of gases and particles in the air increased rapidly and peaked shortly after each
application event then decreased rapidly. Concentrations of gas phase emissions increased again
following the factor of ten reduction of chamber air flow at the end of Phase I and for most
emissions, concentrations increased again when chamber air temperature increased by 3 °C due
to malfunction of the building HVAC system. Concentrations of all emissions except the flame
retardant TCPP rapidly fell below detectable levels with removal of the sprayed frames from the
chamber at the end of Phase II. The flame retardant TCPP was deposited on surfaces of the
chamber, sampling equipment in the chamber, mixing fans, PPE of the personnel in the chamber
during Phase I, on the HVAC filters covering the exhaust duct and the inner surfaces of the
exhaust duct. Concentrations were highest on floor of the chamber and higher on the walls than
the ceiling.
6.2.1 Summary of Phase I Air Concentrations
Phase I air concentrations are summarized for the isocyanates, blowing agent, flame retardant,
amine catalyst, and four VOCs in the Table 6-1 as average concentrations during the spray
events and concentration of the last sample collected during Phase I. The concentration data may
be useful for scaling test parameters and estimating sampling volumes for specific analytes in
SPFI Methods Development
Page 159
-------
tests conducted with similar area specific loading ratios. Table 6-1 Phase I air concentration
summary
Phase 1 Air Concentrations Summary (|jg m~3)
Time Period
Average During Application
End of Phase I
Sampling location
Chamber
Exhaust Duct
Chamber
Exhaust Duct
Sum MDI
542±169 (ASSET)
486±133 (Impinger)
<5
HFC-134a
1.8E+07
6.0E+04
TCPP
2.3E+03±0.5E+03
5.5E+02±2.7E+02
97
142
PMDETA
9.0E+03±5.2E+03
NA*
170
270
3-Chloropropene
2.8E+02±1.0+02
3.1 E+02±1.1E+02
17
39
1,2-Dichloropropane
1.5E+02±5.9E+01
1.7E+02±5.6E+01
5.6
17
14-Dioxane
1.3E+03±5.8E+02
1.5E+03±6.7E+02
15
25
Chlorobenzene
37±6
32±14
3.6
3.9
*The loading on the 0.5 L sample volumes col
ected from the exhaust duct during the application
were above the calibration range of the TD-GC/MS.
The results do not support the hypothesis that denuder/filter samplers underestimate MDI
concentrations during spray events compared to impinger samplers nor do the aggregate results
(impingers and denuder-filter samplers) indicate higher isocyanate air concentrations in the
proximity of the application. However, the variability of the data, rapid mixing in the chamber,
and the relatively proximity of the sampling systems in the chamber may preclude use of the data
to evaluate isocyanate concentrations as a function of distance from the area of application.
6.2.1.1 Summary of Phase I Aldehyde Concentrations
Phase I concentrations of formaldehyde, acetaldehyde, acetone, and propionaldehyde were
quantified well above the Phase I chamber air flow background concentrations. The highest
concentrations were observed for acetone at 276 |ig m"3 whereas maximum observed
concentrations for the other three compounds were in the range of 50 to 75 |ig m"3.
Concentrations ranged from 2.6 (formaldehyde) to 17 |ig m"3 (acetone) at the end of Phase I
except for acetaldehyde which was not quantified after 0.5 h.
6.2.1.2. Summary of Phase I Particle Concentrations
Particle concentrations determined with the ELPI summed over the bin size range of 0.039 to
8.22 |im peaked at 1.21E+05 particles per cm3 during the application and decayed to 1.10E+03
particles per cm3 by 1.5 h. Particles with aerodynamic diameter <0.5 and <3 |im accounted for
81 ±4% and 99.9±0.1% respectively of the particle emissions determined as number per cm3.
6.2.2 Summary of Phase II Concentrations
Phase II air concentrations of HFC-134a, TCPP, PMDETA, and the VOCs are summarized in the
following sections. Concentrations of particles and isocyanates were not measured in Phase II.
SPFI Methods Development
Page 160
-------
6.2.2.1 Summary of Phase II Air Concentrations of the Blowing Agent HFC-134a
• Concentrations increased to 1.4E+05 |ig m"3 at 7.44 h, or about 4 hours after the chamber
air flow rate was reduced by a factor of ten.
• Concentrations peaked at 7.44 h, however, between 6.34 and 8.4 h the concentrations
were nearly constant and averaged 1.38E+05 |ig m"3 with a relative standard deviation of
0.85%.
• Concentrations decreased by a factor of 10 within 144 h.
• Concentrations increased from 5.5E+03 to 8.1E+03 |ig m"3 during the time where
chamber air temperature increased by 3 °C.
• Concentrations decreased to 3.75E+03 |ig m"3 by the end of Phase II.
6.2.2.2 Summary of Phase II Air Concentrations of TCPP
• Concentrations increased to 133 |ig m"3 measured in the chamber sampling port at 10.4 h
or about 7 h after the chamber air flow rate was decreased from 123 to 12 m3 h"1
• Concentrations increased to 103 |ig m"3 measured at the exhaust duct sampling port at
23.9 h.
• Chamber air concentrations remained above the exhaust duct concentrations over the
period of Phase II, though after 100 h the differences were <5 |ig m"3.
• Concentrations increased from 70 to 100 |ig m"3 when the chamber air temperature
increased from 23.9 to 26.9 °C due to malfunction of the air conditioner for the building
housing the chamber.
• Concentrations were 60 |ig m"3 measured at the chamber and exhaust duct sampling ports
at the end of Phase II.
6.2.2.3 Summary of Phase II Air Concentrations of PMDETA.
• Concentrations appeared to peak at 5 h in the chamber whereas the highest concentrations
determined in the exhaust duct were observed at 24 h.
• Concentrations ranged from 323 |ig m"3 after the decrease in chamber air flow rate to 37
|ig m"3 at the end of Phase II determined by sampling from the chamber sampling port.
• Concentrations ranged from 298 at 24 h to 35 |ig m"3 at the end of Phase II determined by
sampling from the exhaust duct sampling port.
• The ratio of concentrations determined in the exhaust duct to concentrations determined
in the chamber averaged 0.96±0.04 for 11 of 12 samples collected between 24 and 669 h.
6.2.2.4 Summary of Phase II Air Concentrations of the Target VOCs
SPFI Methods Development
Page 161
-------
• The Phase II concentrations of the target VOCs 3-chloropropene, 1,2-dichloropropane,
1,4-dioxane, and chlorobenzene followed very similar patterns.
• Maximum concentrations were observed at 10.5 hours for each compound and ranged
from 66 to 2.7 |ig m"3.
• Concentrations at the end of Phase II ranged from 19 |ig m"3 for 3-chloropropene to
-------
Phase I estimate for TCPP also includes the mass deposited on surfaces in the chamber,
including walls, floor, and ceiling.
Table 6- 2 Summary of Phase I mass emitted for target compounds
Compound
(pg)
(M9 m-2)
Sum MDI (Impinger)
2.4 E+04 *
3.4E+03
(Denuder-Filter)
2.7E+04**
3.8E+03
HFC-134a
1.2 E+09
1.7E+08
3-Chloropropene
4.2E+04
6.0E+03
1,2-Dichloropropane
2.1 E+04
2.9E+03
1,4-Dioxane
1.5E+05
2.1 E+04
Chlorobenzene
5.3E+03
0.8 E+03
Formaldehyde
8.0E+03
1.1E+03
Acetaldehyde
4.0E+03
5.7E+02
Acetone
2.8E+04
4.0E+03
Propionaldehyde
1.2E+04
1.8E+03
*Impinger samplers
** Denuder-filter samplers
Table 6- 3 Summary of mass emitted during phase I for TCPP and PMdeta
Compound
Phase I Mass Emitted Summary (pg)
Emitted per
unit area
sprayed
Air
Exhaust
Filter
PPE
Chamber
Surfaces
Total
(|jg m-2)
TCPP
1.3E+05
6.6E+04
4.7E+04
1.4E+05
3.9E+05
5.5E+04
PMDETA
8.0E+05
5.1 E+03
5.2E+04
NA
8.6E+05
1.2E+05
6.2.6 Summary of Phase II Mass Emitted
The chamber concentration time data for the blowing agent, VOC emissions and amine catalyst,
were used to estimate mass emitted by the trapezoid method. For TCPP, the Phase II mass
emitted was determined as the sum of the mass leaving in the exhaust duct and the mass
collected on the exhaust duct filter.
Table 6- 4 Summary of mass emitted during Phase II
Air (pg)
Exhaust
Filter (pg)
Total (pg)
Emitted per unit area
sprayed (pg rrr2)
HFC-134a
1.0 E+08
1.0 E+08
1.4E+07
TCPP
6.1E+05
7.6+03
6.2+05
8.8E+04*
PMDETA
5.9E+05
5.9E+05
1.2E+05
3-Chloropropene
1.96E+05
1.96E+05
2.8E+04
1,2-Dichloropropane
2.1 E+04
2.1 E+04
7.3E+03
1,4-Dioxane
5.2E+04
5.2E+04
2.1 E+04
Chlorobenzene
1.1 E+04
1.1+04
1.5 E+03
Acetaldehyde
3.7E+04
3.7E+04
5.3E+02
Isovaleraldehyde
1.9E+03
1.9E+03
2.7E+02
*A11 TCPP leaving the system (exhaust duct air + mass on outlet duct filter) attributed to sprayed frames.
SPFI Methods Development
Page 163
-------
6.2.7 Generation of Emission Rates from Chamber Concentration Data
6.2.7.1 Phase I MDI emissions modeling
The mass emitted determined from the chamber concentration data were utilized to generate
constant and instant emission factors which were utilized in IECCU to generate predicted time-
concentration plots for isocyanate emissions in the chamber system. The IECCU concentration-
time predictions under predicted the peak concentrations during the first spray event and
overpredicted the concentrations following the completion of the second spray event. In the
absence of sink terms for the isocyanates, the software calculated the post-application
concentration based upon the air change rate. The underprediction of the concentration peak
during the first spray event may be due to how the software balances mass emitted with
concentration - time predictions. Particle deposition rates and chemical reaction rate terms
would improve the fit between the measured concentrations and the simulations.
The direct calculation of emission rates from chamber concentration data provided a better fit
between the simulation and the measured concentrations. However, the simulation did not
capture the apparent decrease in concentrations during the period between the two spray events.
This is due to the fact that the simulation which is based upon direct calculation of emission rate
from the chamber concentration data, is not coupled to the timing of the spray events. As
observed with the use of the instant and constant emission rate approach, loss rate terms are
needed to account for mass emitted by the spraying events.
6.2.7.2 Phase I Emissions Modeling for Blowing Agent, the VOCs, and Amine Catalyst
The chamber concentration time data for the blowing agent, VOC emissions and amine catalyst,
during the application phase were used to estimate mass emitted by the trapezoid method. The
calculation of mass of HFC-134a emitted during the application phase was consistent with 10%
weight percent of the blowing agent emitted during the application. Mass emitted was
apportioned to area and time of application. This information was used to fit a dual first order
decay model to the time-concentration data. A least squares curve fitting routine was employed
to estimate the decay constants, ki and fo. Reasonable fits between the data and model
predictions demonstrated that this approach may be useful for generating application phase
emission models from the test protocol.
Application phase emission factors were estimated for several VOCs, and the amine catalyst
PMEDTA, using the approach where mass emitted during application phase was used to
determine emittable mass, Mo, which was apportioned to mass available for rapid emissions (Mi)
and mass available for slower emissions (Mi) based upon area sprayed in each spray event. The
chamber concentration time data were then used to generate dual first order decay constants, ki
and k2 for the application phase emissions model using a non-linear least-squares fitting routine.
For post-application phase emissions, the direct calculation method was used to generate
emission factors over the 665-hour post application period. A power fit was found to represent
the decrease in emission factors with time, suggesting that the post-application phase emissions
are controlled by material-air diffusion processes. Analysis of the emission factors determined
SPFI Methods Development
Page 164
-------
with the power fit suggest that emission factors determined in a 72-h test may be useful in
predicting emission factors beyond the period of the test.
6.2.7.3 Phase II Emissions Modeling of VOCs, Amine Catalysts
Concentrations of HFC-134a were within the instrument calibration range during Phase II and it
was estimated that 104 g of HFC-134a were emitted over the 665 h period. Emission factors as a
function of time were determined using the direct calculation from chamber concentration
method. The relationship between temperature and HFC-134a emission rate was investigated by
plotting emission factor versus chamber air temperature during the period where the air handler
for the building housing the chamber malfunctioned. A linear relationship between emission
factor and temperature was observed over the 2.9 °C temperature rise.
Phase II chamber concentrations of emissions increased for a period of approximately four to ten
hours following the decrease in chamber air change rate from 4 to 0.4 h"1, then began to decline.
Emission factors were generated from the chamber concentration-time data using the method of
direct calculation from chamber concentrations. It was found that a power function fit the decay
of the emission factors with time, suggesting diffusion-controlled emissions processes. The
power fit of the emission factors through 72 h was used to predict emission factors at 168, 338,
and 669 h. The comparison between predicted emission factors and those calculated at each time
from the experimental data demonstrates the potential for predicting emission factors beyond the
time period of a 72-hour test.
6.2.7.4 Phase II Modeling of TCPP
The modified space-state diffusion model in IECCU was run with diffusion and partition
coefficients (Dm and Kma) input parameters for the flame retardant TCPP suggested by Bevington
and with parameters selected based upon the data. In each case, Co, the concentration of TCPP in
the SPF was that determined by extraction from the foam that was applied to the substrates. This
exercise demonstrated the potential of a diffusion-based model to predict concentrations in a
controlled environment. The Phase III TCPP emissions data indicated that about 20% of the
emissions during Phase II may be due to emissions of TCPP from chamber walls and overshoot
nodules remaining in the chamber after removal of the sprayed frames. This indicates the need
for improved understanding of the impact of emissions from secondary sources and an approach
to modeling the secondary emissions from contaminated chamber surfaces.
6.2.7.5 Summary of Calculation of Emission Rates from Chamber Concentrations
The power fit of emission factors determined at 23.8 °C was used to predict the emission factors
of several compounds at the time that the chamber temperature rose to 26.9 °C. The predicted
emission factor at 477 h, the time of maximum chamber air temperature, was compared to the
one determined by direct calculation of emission factor from the concentration data point
corresponding to peak temperature to estimate % increase in emission factor with increased
temperature. The apparent increase in emission factor with the 3 °C temperature rise varied from
38 to 54% for selected compounds with vapor pressures ranging from 9E-3 to 5E+3 mm Hg.
These apparent increases in emission factor with temperature are indications of the temperature
dependence of emission factors for emissions of VOCs and SVOCs from SPF insulation. This
SPFI Methods Development
Page 165
-------
implies a need for data that characterizes temperatures in uncontrolled spaces of buildings and
development of temperature dependent An andiC„M parameters for SPF insultation materials.
6.3 Major Findings
The mass of TCPP in several environmental compartments was determined periodically from air
concentrations, deposition samplers, and extractions from PPE and filters placed over the
chamber exhaust duct. The results inform understanding of the processes that impact transport of
flame retardant. During the 3.3-hour application phase, 122 mg TCPP left the chamber system
via the exhaust duct, 66 mg was collected on the exhaust duct filter, 34 mg left the chamber
system on the PPE of the spraying/sampling personnel, and 141 mg was deposited on chamber
surfaces. During the 665-hour post-application phase 600 mg TCPP left the chamber system via
the exhaust duct, 7.6 mg was collected on the filter covering the exhaust duct and an estimated
114 mg remained on chamber surfaces. During the 290-hour decay phase, 55 mg left the
chamber system in the air of the exhaust duct, 2 mg were collected on the exhaust duct filter and
approximately 100 mg remained on chamber surfaces. An additional estimated 3,950 mg of
TCPP was collected from overshoot recovered from the floor of the chamber at the end of the
experiment.
Concentrations of several aldehydes were quantified as emissions. However, the use of plywood
substrate complicated interpretation of the data because several of the aldehydes were present in
background samples collected prior to the spray event. Also, since only 15 of 16 quadrants were
sprayed, a portion of the plywood substrate remained exposed to the chamber air. The spray
event produced short-term aldehyde emissions above the background levels. Isovaleraldehyde
was the only aldehyde that remained above background concentrations during the post-
application low flow period. No attempt was made to generate emission factors from the
chamber concentrations due to the uncertainty regarding exposed plywood. The data indicate a
need to understand diffusion of aldehydes through SPF insulation to inform exposure modeling
to emissions from plywood substrates.
Particle number and size were monitored during the application phase of the experiment. The
results indicate that the greatest particle number is in the size range <3 |im. As of the time of this
reporting, emission factors have not been generated from the particle data.
6.4 Conclusions
The following conclusions are drawn from the test results:
• The test results are suitable for informing development of a consensus full-scale integrated
test method or guidance document within ASTM or other standard setting body.
o Approaches for generation of source emission rates and emission factors for a wide
range of emissions during and following application of SPF insulation have been
demonstrated.
• The test results inform characterization of potential SPF insulation emissions to indoor and
ambient air and may be useful for evaluation of emissions control strategies.
SPFI Methods Development
Page 166
-------
o Results demonstrated that emissions may be orders of magnitude higher during the
application than during the post-application/curing phase.
o Results demonstrated the potential for blowing agent and flame retardant emissions
from closed cell SPF insulation to persist long after application.
o Results indicated that MDI and p-MDI emissions decreased rapidly following
cessation of the spray events.
¦ Results did not indicate that denuder/filter samplers underestimated MDI and
p-MDI concentrations during application compared to impinger samplers for
the low-pressure kit.
o Results demonstrated that a common HVAC filter captured about one-third of the
TCPP emissions leaving the chamber in the exhaust duct air flow during Phase I.
o Results demonstrated that the HVAC filters had little impact on transport of gas phase
TCPP from the chamber to the exhaust duct following the application and initial
curing.
o Results demonstrated that a significant portion of the TCPP emitted during
application was deposited on surfaces in the chamber, including the PPE of the
sprayer and helper.
o Results demonstrated that TCPP emissions from material deposited on chamber
surfaces persisted following removal of the sprayed frames from the chamber.
o Results indicated that acetone may not be a reliable solvent for collecting TCPP from
chamber surfaces with wipe samplers.
• Results demonstrated that additional work may be needed to develop and demonstrate the use
of the modified OVS sampler for quantification of particle and gas-phase TCPP emissions
during application.
• Results demonstrated the importance and potential challenges of post-test cleanup of surfaces
of the chamber.
7.0 Recommendations
• Continue to work with stakeholders in ASTM D22.05 to develop the integrated full-scale
protocol guidance document that includes characterization of the application phase
emissions.
o Encourage stakeholders to demonstrate the integrated testing approach with high
pressure SPF insulation products.
¦ Evaluate the impact of air change rate on emission factors during the
application to determine if emission rates increase with air change rate, an
indication of gas-phase limited mass transfer.
SPFI Methods Development
Page 167
-------
• Develop and demonstrate methods to determine total emittable mass (Co), D, and K (and
variance with temperature) for semivolatile emissions from SPF insulation products as this
could potentially greatly reduce the need for expensive and time-consuming emissions tests
and improve long-term predictions of emissions.
• Encourage research to support use of IECCU by stakeholders, including:
a. Characterization of temperatures in unconditioned spaces of buildings where SPF
insulation is applied.
b. Characterization of air flows and movement of SPF insulation emissions from
unoccupied to occupied spaces.
• For the consensus test protocol, develop a chamber design that allows collection of all types
of samples (including impingers) without the need for personnel to enter the chamber.
• Avoid using wood substrates that emit aldehydes unless the experimental goals are to
investigate interactions between substrate and aldehyde emissions.
• Employ the data set generated in this test to evaluate/calibrate IECCU for additional
chemicals, products, and installation conditions.
• Encourage stakeholders to develop flame retardants that are chemically bound to the
material.
• Encourage stakeholders to investigate methods to remove gas-phase flame retardant from
indoor air.
8.0 References
[1] Tichenor BA, Sparks LA, White JB, Jackson MD. Evaluating sources of indoor air pollution.
J Air Waste Manage Assoc. 1990 Vol. 40(4):487-92. doi: 10.1080/10473289.1990.10466703.
PMID: 2340148.
[2] Tichenor, Bruce A., Guo, Zhishi, Dunn, James E., Sparks, Leslie E., Mason, Mark A. The
Interaction of Vapour Phase Organic Compounds with Indoor Sinks, Indoor Air, Vol 1, (1),
0905-6947. https://doi.Org/10.llll/i.1600-0668.1991.03-ll.x
[3] Tichenor, Bruce A., Guo, Zhishi, Sparks, Leslie E. Fundamental Mass Transfer Model for
Indoor Air Emissions fi'om Surface Coatings, Indoor Air, Vol. 3, (4), 0905-6947,
https://doi.Org/10.llll/i.1600-0668.1993.00007.x
[4] Brown, S. K., Sim, M. R., Abramson, M. J., Gray, C. N., Concentrations of Volatile Organic
Compounds in Indoor Air - A Review Indoor Air, Vol 4, (2) 0905-6947,
https://doi.Org/10.llll/j.1600-0668.1994.t01-2-00007.x
[5] Center for the Polyurethanes Industry, 2012 End-Use Market Survey on the Polyurethanes
Industry in the United States, Canada, and Mexico, American Chemistry Council, Washington,
DC, 2013.
SPFI Methods Development
Page 168
-------
[6] Huang, Y. and Tsuang, W., Health Effects Associated with Faulty Application of Spray
Polyurethane Foam Insulation in Residential Homes, Environ Res., Vol. 134, 2014, pp. 295-300.
[7] Tsuang, W., Asthma Induced by Exposure to Spray Polyarethane Foam Insulation in a
Residential Home, J. Occup.Environ. Med., Vol. 54, No. 3, 2012, pp.272-273.
[8] Krone, C. A., and Klinger, T.D., Isocyanates, polyarethane and childhood asthma, Pediatric
Allergy and Immunology, Volume 16, Issue 5, August 2005, pp.368-379.
[9] Redlich, C.A., Karol, M.H., 2002. Diisocyanate asthma: clinical aspects and immuno-
pathogenesis. Int. Immunopharmacol. 2, 213-224
[10] Bello, D., Herrick, C. A., Smith, T.J., Woskie, S.R., Streicher, R.P., Cullen, M.R., Liu, Y.,
and Redllich, C.A. 2007. Environ Health Perspectives. 115, (3), 328-335.
[11] K. Sleasman, Carol Hetfield, and Melanie Biggs, "Investigating Sampling and Analytical
Techniques to Understand Emission Characteristics from Spray Polyurethane Foam Insulation
and Data Needs." In STP1589-EB Developing Consensus Standards for Measuring Chemical
Emissions from Spray Polyarethane Foam (SPF) Insulation, edited by Sebroski, J. and Mason,
M., (pp. 277-2017). West Conshohocken, PA: ASTM International, 10.1520/STP158920150037.
doi:https://doi.org/978-0-8031-7624-9.
[12] Van der Veen, I., de Boer, J., 2012. Phosphorous flame retardants: Properties, production,
environmental occurrence, toxicity and analysis. Chemosphere. 88, 1119-1153.
[13] Marklund, A., Andersson, B., and Haglund, P., "Organophosphorous Flame Retardants and
Plasticizers in Air from Various Indoor Environments", J. Environ. Monitor., Vol. 7, No. 8,
2005, p. 814-819.
[14] EINECS No: 237-158-7, Summary of Risk Assessment Report, May 2008, last accessed
8/14/2019 at: TCPP Risk Assessment Report).
[15] www.epa.gov/snap
[16] Tesser et.al., Modeling of Polyarethane Foam Formation, Journal of Applied Polymer
Science, Vol 92, 1875-1886 (2004)
[17] Developing Consensus Standards for Measuring Chemical Emissions from Spray
Polyurethane Foam (SPF) Insulation. STP1589. Editors: John Sebroski and Mark Mason. ASTM
International, West Conshohocken, PA (2017).
[18] Woods, Estimating Re-entry Times for Trade Workers Following the Application of Three
Generic Spray Polyurethane Foam Formulations. In Developing Consensus Standards for
Measuring Chemical Emissions from Spray Polyurethane Foam (SPF) Insulation, Editors; John
Sebroski and Mark Mason. ASTM STP 1589
[19] IECCU. a Simulation Prosram for Estimating Chemical Emissions from Sources and
Related Changes to Indoor Environmental Concentrations in Buildings with Conditioned and
SPFI Methods Development
Page 169
-------
Unconditioned Zones, https://www.epa.gov/tsca-screening-tools/users-guide-and-download-
ieccu. Accessed on 9/20/2017.
[20]_ASTM D6670 Standard Practice for Full-Scale Chamber Determination of Volatile
Organic Emissions from Indoor Materials Products. ASTM International, West Conshohocken,
PA (2013). www.astm.org
[21] Wenhao Chen, Andrew K. Persily, Alfred T. Hodgson, Francis J. Offermann, Dustin
Poppendieck, Kazukiyo Kumagai, Area-specific airflow rates for evaluating the impacts of VOC
emissions in U.S. single-family homes, Building and Environment, Volume 71, 2014, Pages 204-
211, ISSN 0360-1323, https://doi.Org/10.1016/i.buildenv.2013.09.020.
[22] T. G. Matthews , D. L. Wilson , A. J. Thompson , M. A. Mason , S. N. Bailey & L. H.
Nelms (1987) Interlaboratory Comparison of Formaldehyde Emissions from Particleboard
Underlayment in Small-Scale Environmental Chambers, JAPCA, 37:11, 1320-1326, DOI:
10.1080/08940630.1987.10466328
[23] ASTM E741-11 (2017) Standard Test Method for Determining Air Change in a Single Zone
by Means of a Tracer Gas Dilution. ASTM International, West Conshohocken, PA (2017).
www.astm.org
[24] Guo, Z., "On Validation of Source and Sink Models: Problems and Possible Solutions,"
Modeling of Indoor Air Quality and Exposure ^ASTM STP 1205. NirenNagda, Ed. American
Society for Testing and Materials, Philadelphia, 1993, pp. 131-144.
[25] Bevington, C., Guo, Z., Hong, Tao., Hubbard, H., Wong, E., Sleasman, K., and Hetfield, C.
"A Modeling Approach for Quantifying Exposures fi'om Emissions of Spray Polyurethane Foam
Insulation in Indoor Environments'", in Developing Consensus Standards for Measuring
Chemical Emissions from Spray Polyurethane Foam (SPF) Insulation. STP1589. Editors: John
Sebroski and Mark Mason. ASTM International, West Conshohocken, PA (2017).
[26] ASTM D8142 Standard Test Method for Measuring Chemical Emissions from Spray
Polyurethane Foam Insulation using Micro-Scale Environmental Test Chambers. ASTM
International, West Conshohocken, PA (2017).
[27] Howard, E. M., Mason, M. A., Zhang, J., and Brown, S., "A comparison of Design
Specifications for Three Large Environmental Chambers'", in Engineering Solutions to Indoor
Air Quality Problems. VIP 51, Air and Waste Management Association, One Gateway Center,
Third Floor, Pittsburgh, PA 15222.
[28] ISO 17734-1 Detenrrination of organonitrogen compounds in air using liquid chromatography and mass
spectrometry - Part 1: Isocyanates using dibutylamine derivatives.
https://www.iso. org/standard/58006. html
[29] NIOSH 5600 Organophosphorous Pesticides https://www.cdc.gov/niosh/docs/2003-
154/pdfsZ5600.pdf
SPFI Methods Development
Page 170
-------
[30] Compendium Method TO-17, Determination of Volatile Organic Compounds in Ambient
Air Using Active Sampling onto Sorbent Tabes. EPA/625/R-96/010b
[31] Compendium Method TO-11 A, Determination of Formaldehyde in Ambient Air Using
Adsorbent Cartridge Followed by High Performance Liquid Chromatography (HPLC) (Active
Sampling Methodology) EPA/625/R-96/010b
[32] National Standard of Canada CAN/ULC-S705.2-05 (ICS 91.120.10, 03.120)
Standard for Thermal Insulation - Spray Applied Rigid Polyitrethane Foam Medium Density -
Application. Underwriters Laboratories of Canada - approved by Standards Council of Canada
Second Edition - February 2005
[33] Hodgson, A., Beal, D., Mcllvaine, J. (2002) Sources of formaldehyde, other aldehydes and
terpenes in a new manufactured house. Indoor Air 12(4): 235-242.
[34] Streicher, R. P., E. R. Kennedy and C. D. Lorberau (1994). "Strategies for the simultaneous
collection of vapours and aerosols with emphasis on isocyanate sampling." Analyst 119(1): 89-
97.
[35] Puscasu, S. Aubin, S., Cloutier, Y., Sarazin, P., Van Tra H., Gagne S. Comparison between
the denuder-filter EZ4 NCO and Impinger Sampling Devices for Aerosol Sampling of 4,4'-
Mthylene DiphenylDiisocyanate in Spray Foam Application, (2015). The Annals of
Occupational Hygiene, 59(7), 872-881. doi: 10.1093/annhyg/mev025
[36] Liang, Y., Liu, X., Allen, M.R. The Influence of Temperature on the Emissions of
Organophosphate Ester Flame Retardants from Polyisocyanurate Foam: Measurement and
Modeling. Chemosphere 233 (2019) 347-354.
[37] Zhao, Little, and Cox, "Characterizing Polyurethane Foam as a Sink for or Source of
Volatile Organic Compounds in Indoor Air". Journal of Environmental Engineering (2004)
Vol.130, Issue 9.
Acknowledgements
Mr. Ken Krebs was responsible for the isocyanate sample processing and analysis by LC-
MS/MS, Mr. Dale Greenwell operated the Innova multi-gas monitor, the particle characterization
instruments, and the GC-ECD and collected samples from within the chamber during
application. Mr. Mark Barnes operated the full-scale emissions test chamber, sprayed the foam,
collected samples from within the chamber following application and conducted the physical
properties measurements. Mr. Gary Folk (Jacobs Technologies) performed the duties of safety
observer, collected samples and operated the DI-GC/MS system and was responsible for
compilation of the sampling events. Mr. Folk was assisted by Mr. John Ulrich (formerly of
Jacobs Technology) who assisted with sample collection and analysis by TD-GC/MS. Mr. Calvin
Whitfield, Jacobs Technology, assisted with operation of the supplied air systems and provided
additional communications support during the application phase of the experiments.
SPFI Methods Development
Page 171
-------
We also wish to acknowledge Ms. Carol Hetfield, OPPT, (retired) for her leadership of the
intergovernmental polyurethanes work group, and efforts to support development of reliable
consensus standards for characterization of emissions from SPFI. We also acknowledge the
support of Charles Bevington, formerly OPPT, currently CPSC, for leading development of the
simulation tool IECCU (Dr. Zhisihi Guo, ICS) and for his continued support of the emissions
characterization research. We acknowledge Katherine Sleasman (OPPT) for leadership on MDI
and support to the SNAP program.
We thank Cheryl Estill and Robert Streicher, NIOSH for discussion of sampling and analysis
methods for flame retardants and isocyanates. We also thank Drs. Gunnar and Marianne
Skarping and Daniel Karlsson for discussions regarding ISO 17734 for measurement of
isocyanates.
We also thank the EPA and NIST, NIOSH, and CPSC peer reviewers and the Air and Energy
Management Division QA staff for review of the report generated during this project.
SPFI Methods Development
Page 172
-------
Appendices
Appendix A Emissions Test Chamber Facility
A-l Emissions Test Chamber
EPA's room-sized emissions test chamber referred to as the Full-Scale Emissions Test Chamber
(FSETC), served as an SPF insulation emissions characterization method development test
facility [19], The FSETC (Figure A-l) is constructed of Type 304 stainless steel (SS) with all
interior seams welded and all exposed surfaces polished to a No.4 finish. The approximate
interior dimensions of the chamber are: 2.75 m (H) x 3.80 m (L) x 2.85 m (W) having an interior
volume of approximately 29.8 m3 (1053 ft3). The exterior is made of 20-gauge enameled sheet
metal with pipe lock seams. Space between the interior and exterior walls is insulated to aid in
temperature control of the chamber environment. The clean air generation and conditioning
system can supply the chamber with clean and conditioned air at up to 10 h"1 or 300 m3 h"1.
The chamber is configured to accommodate supply and return process airflow, instrumentation,
power and pneumatic needs and air sampling. Two windows are provided for observation and a
door for entry. The door is sealed airtight with the aid of pneumatic piston actuators. Supply air
enters the chamber through four process air supply penetrations that are positioned on the lower
centerline of each wall. Air is exhausted from the chamber through a centrally located opening in
the ceiling of the chamber. Two fans mounted on steel plates are attached to the ceiling to mix
the air. The fans are directed downward at a 10° angle from vertical and expel air in the direction
of the front (door) wall. Fan speed and direction were manually adjusted to provide desired air
speed of 0 to 0.25 m sec"1 0.01 m from the surface of the substrates.
SPFI Methods Development
Page 173
-------
LEGEND
@ COMPUTER GC/HjGEN
PRE-CONDITIONING BOX
DEHUMIDIFIER
FAN 1. SUPPLY
CARBON ADSORBER
MAIN CONDITIONING BOX
LARGE CHAMBER
FILTER TEST MODULE
FAN 2, REC1RC
SINK
CHILLER
CEM PANEL
HUMIDIFIER
FAN 3, EXHAUST
CONTROL PANEL
- MEZZANINE LIMITS
MESWINE UMITS
1
2
3
4
s
8
T
a
9
10
11
12
13
14
Figure A-1 Schematic of test chamber and clean air conditioning and delivery system
For the SPF insulation methods development tests, clean, conditioned air was supplied to the
chamber by Fan 1 shown in the diagram above and was exhausted from the chamber by Fan 3.
Pressure sensors on either side of calibrated orifice plates are continuously monitored by the
control system which adjusts the fan speeds to maintain set point flows and pressures. Chamber
flow is determined at the exhaust duct flow measurement station. As a result of this control
strategy, air released into the chamber from supplied air respirators worn by personnel in the
chamber results in a corresponding decrease in supply air flow from the clean air system with no
net change in exhaust air flow rate. The chamber flow control system allows for zero, partial, or
complete recirculation the of air exiting the chamber. For the experiments reported here, the
chamber was operated in single pass mode with no recirculation. Air withdrawn from the
chamber during sampling was returned to the exhaust duct upstream of the flow control station.
Figure A-2 shows the process flow diagram for the chamber system.
SPI I Methods Development
Page 174
-------
Figure A- 2 Large chamber control computer screen shot of air flow control
A-2 Modifications to the Chamber for Characterization of SPFI Emissions.
The test chamber was modified by addition of an air lock vestibule, aluminum sheeting wall
covering, air lines to supply breathing air to personnel working in the chamber, and fans to
provide additional mixing of chamber air and provide a means to control air velocity near the
surfaces of the SPF insulation sprayed onto the substrates.
SPFI Methods Development
Page 175
-------
A-2,1 Air Lock Vestibule
The test chamber was modified by addition of an air lock vestibule (Figure A-3) to prevent
Figure A- 3 Air lock vestibule
source emissions from contaminating surrounding laboratory space. The vestibule consists of
polyethylene sheets stretched over a metal frame that has the same width and height as the
FSETC. The polyethylene sheeting overlaps the outer wall of the chamber and is taped to the
outer walls of the chamber, to the metal frame, and to the floor. Zip-doors (ZipWall, 4 ml plastic
sheet, (lame retardant), was installed in openings on either side of the vestibule for entry and
egress from the laboratory to the vestibule. With the zip doors closed, air enters the vestibule
through a 20" X 20" XI" filter mounted in an opening next to one of the doors and is withdrawn
from the vestibule through a 4" outlet duct that penetrates into the vestibule. The outlet duct is
connected via flex duct to the inlet of the large blower and filtration unit. A ball valve installed
in the outlet duct can be closed to isolate the vestibule from the exhaust system. This
arrangement (1) allows personnel to enter and exit the chamber from the vestibule while
minimizing exchange of air between the chamber and the vestibule, (2) permits rapid ventilation
of the vestibule once the personnel have closed the door to the chamber and (3) ensures
maintenance of negative pressure. The air exchange rate of the vestibule, determined using a
tracer decay method [20] was 0.07 h"' with the vestibule isolated from the vestibule exhaust
system and 20 h"1 with the ball valve partially open. A portable high efficiency particulate air
SPI I Methods Development Page 176
-------
(HEP A) filtration unit was placed in the vestibule to reduce the particulate concentrations of the
vestibule space following a door opening event.
A-2.2 Chamber Wall Covering
To minimize potential waste issues associated with solvent cleaning to remove spray foam
insulation and flame retardant from the stainless steel walls of the chamber, the chamber walls
were covered with aluminum film (Grainger part #4UGG8, soft temper foil thickness of 0.127
millimeter, 30.5 m length, 1.2 m width).
A-2.3 Air Supply for Supplied Air Respirators
The chamber and vestibule were each equipped with air lines that provided breathing air to the
supplied air respirators worn by personnel working in the chamber and vestibule. This allowed
personnel to enter the chamber using air supply lines terminating in the vestibule, switch to the
lines that penetrate the chamber through a port in the wall and close the chamber door after
passing the vestibule air supply lines out of the chamber to personnel equipped with supplied air
respirators stationed in the vestibule. Upon egress from the chamber, personnel switched from
chamber to vestibule respirator air supply lines, closed the chamber door and ventilated the
vestibule. This protocol minimized potential inadvertent exposure to emissions. Coiled air supply
lines inside of the chamber are shown in Figure A-4.
SPFI Methods Development
Page 177
-------
Figure A- 4 Air supply lines for supplied air respirators
A-2.4 Air Mixing Fans
To ensure rapid chamber air mixing and create the desired velocity range near the surfaces of the
substrates, two fans (Soleus Air Model FT-25-A) were installed on a bar attached to the ceiling
near the door end of the chamber. The fans were oriented towards the front wall at a downwards
angle of 10° so that air deflected off the wall towards the frames at the opposite side of the
chamber. Fan speeds were set to position 1, the lowest of three set points.
An omnidirectional hot-wire air velocity sensor with temperature sensor for temperature
compensation (Clinomaster Model 6501 Series Multifunction Anemometer, Kanomax, Andover,
NJ) was used to measure air velocity across the surfaces of the substrate panels. The
measurement range is 0.01 to 5.00 m s"1, resolution is 0.01 m s"1, and accuracy is +/- 0.02 m s"1
for velocities from 0.01 to 0.99 m s"1. Air velocity measurements were conducted prior to the test
to set the speed and direction of the fans and following completion of the test to document air
velocities across the uneven surfaces of the SPF insulation.
SPI I Methods Development
Page 178
-------
A-2.4.1 Mixing Fan Placement and Verification of Air Velocity Near the Surface of the Substrates
Two mixing fans were mounted on a bar fixed near the ceiling of the chamber at the end of the
chamber away from the frames. The speed and direction of the fans were adjusted to obtain desired air
velocity and turbulence at the predicted surface of the sprayed foam. Air speed was determined prior to
the experiment using empty frames and following the experiment with foam in the frames to provide
insight into the reliability of the pretest measurement procedure because the surface of the empty frames
is smooth whereas the surface of the sprayed foam is very uneven. The general setup of the fans is shown
in Figure A-5. Pretest velocity measurements were conducted at the low-flow conditions with a
polyethylene sheet fastened to the outer surfaces of the frames.
Panel 2
Panel 3
Panel 1
Panel 4
\V
A\
*
*
Door
Figure A- 5 Mixing fan placement and directions
SPFI Methods Development
Page 179
-------
A-2.4.2 Measurement of Air Speed at the Surface of the Frames
The locations for collecting air speed measurements on each of the four frames from quadrants 2
and 3 are shown in Figure A-6. At each location a series of twenty measurements were collected.
Air speed measurements were also conducted following the completion of testing at the locations
where deposition samplers were placed on the walls, floor, and ceiling of the chamber shown in
Figure 2-3. The measurements were repeated after the end of the test with foam in place in the
frames to determine if the pretest measurements provided a reliable depiction of air speed near
the irregular surface of the foam.
Position 1
X
Position 2
X
Figure A- 6 Location of air speed measurements on each frame
A-2.5 System for Measurement of Chamber Air Change Rate
Chamber air change rate was determined by the tracer decay method as described in ASTM E-
741. An automated system consisting of a programmable event timer, electrically operated
solenoid valves, and mass flow controllers periodically injected the tracer gas (SF6) into the
supply duct from a gass=
. The concentration of the tracer gas was monitored at three-minute intervals with a gas
chromatograph (Agilent 6850) equipped with an electron capture detector. Air was pulled from
the sampling port located in the chamber exhaust duct through a four-port gas sampling valve
and injected from the 1 mL sampling loop onto the column of the GC shown in Figure A-7. Air
change rate (h"1) was determined as the negative slope of the log of the tracer gas concentration
SPFI Methods Development
Page 180
-------
with time. The average air change rate determined by the tracer decay method during each phase
of the experiment was used in calculations of emission rates.
Figure A- 7 GC ECD with gas sampling valve
Appendix B Sampling Systems
Sampling systems are described for collection of emissions from the air, deposition of emissions
on surfaces of the chamber and equipment, and on materials.
B-l Air Sampling Systems
Air sampling systems consist of vacuum pumps, flow controllers, sampling media, and sampling
lines that connect the sampling media to the vacuum pumps. Air sampling systems are described
for collection of isocyanates, the blowing agent, VOCs, flame retardant and amine catalyst,
aldehydes and ketones, and particles.
B-l.l Isocyanate and Gas-Particle Phase TCPP Sampling System
The spraying/sampling personnel collected isocyanate and OVS samplers from inside the
chamber using a custom-made sampling system that supported collection of up to 16 samples
SPFI Methods Development
Page 181
-------
simultaneously at flow rates of 0.2 or 1.0 liter per minute. The vacuum pump shown in Figure B-
1, located outside of the chamber, was connected via PFFTE line to a shutoff valve and vacuum
gage. A stainless steel sampling line penetrated the chamber wall through a bored-through
stainless steel fitting and connected to the manifolds. Each sampling line was equipped with a
shutoff valve and critical orifice as shown in Figure B-2. Color-coded PTFE sampling lines
connected each sampling line to a sampling device, shown in Figure B-3. Sampling devices were
transported in the air-tight container shown in Figure B-4
Figure B-1 Vacuum pump, gage,
SHUTOFF VALVE, OUTSIDE OF CHAMBER
WITH LINE INTO CHAMBER.
Figure B- 2 View of 4-port manifolds and sampling lines with
CRITICAL A CRITICAL ORIFICE IN EACH LINE
SPFI Methods Development
Page 182
-------
Modified OVS samplers were utilized to separate aerosol and gas-phase TCPP emissions during
the application. The OVS samplers were modified due to concerns that the glass fiber would
collect vapor phase flame retardant. The OVS samplers were modified by addition of a PTFE
filter ahead of the glass fiber on the inlet side of the sampling device. Previous unpublished pilot
experiments demonstrated uptake of gas-phase TCPP by the glass fiber filter whereas the TCPP
was demonstrated to pass through the Teflon filter. The modified OVS sampler is shown in
Figure B-5 and B-6. The pilot experiments were carried out at low concentrations with the gas-
phase TCPP generated by diffusion vial placed inside of a 53 L small chamber.
Figure B- 5 Modified OVS with PTFE filter placed
AHEAD OF THE OVS GLASS FIBER FILTER AND XAD RESIN
BEDS.
Transport container for isocyanate samplers.
SPI I Methods Development
Page 183
-------
B-1.2 Air Sampling System for High Concentrations of Blowing Agent
A whole-air sampling system was constructed in-house to collect air samples from the chamber exhaust
duct during periods where the concentrations of the blowing agent were expected to be beyond the
calibration range of the PAS. The system consisted of Tedlar bags fitted with a quick-connect that
attached to the sample inlet line (Figure B-7), an outlet line from the air-tight box to the mass flow
controller and vacuum pump (Figure B-8). To collect samples of known volume, air was pulled through
the sample line from the exhaust duct port into the Tedlar bag by pulling air from the box. The flow rate
of air from the box was controlled by a mass flow controller. Air was later withdrawn from the Tedlar
bags with a gas tight syringe and diluted by injection into clean Tedlar bags containing known volumes of
clean air.
Figure B- 6 OVS with end caps and PTFE filter
MATERIAL
SPFI Methods Development
Page 184
-------
Figure B- 7 Air-tight box for collection of whole air samples in tedlar bags
SPI I Methods Development
Page 185
-------
B-2 Deposition Samplers
Deposition samplers were prepared inhouse using a cutting jig shown in Figure B-9 to cut 64 mm
diameter samplers from foil used to line the chamber. The samplers were attached to larger pieces of
foil in arrays of six or seven samples. The arrays were placed at the locations indicated in Figure 2-3 of
Figure B- 8 Jig for cutting deposition samplers from aluminum foil.
Figure B- 9 deposition sampler array at the base of a sampling tree
SPFI Methods Development
Page 186
-------
82.5 mm diameter samples were cut from the MERV 13 HVAC filters shown in Figure B-ll using a arch
style cutter shown in Figure B-12. Upon collection from the chamber outlet duct opening, the filter
material was wrapped in foil and transported to the laboratory. The pleated filter material was
expanded on an aluminum foil covered lab bench. 82.5 mm diameter samples were cut from the
expanded filter and extracted as described in Section 3.
Figure B-11 MERV 13 HVAC filter section prior to mounting over
CHAMBER OUTLET DUCT OPENING.
¦¦ m
Figure B-10 Cutout jig to collect samples from the HVAC filter.
SPFI Methods Development
Page 187
-------
Appendix C Analytical Systems
Operating parameters for the TD-GCMS, HP GCMS with liquid autosampler, and the LC MSMS
are presented in Section C-l through C-3.
C-l. Thermal Desorption System for Analysis of VOCs Collected on Multi-
bed Sorbent Traps
Figure C-1 Markes tdIOO thermal desorption system and agilent Gcms
Table C-1 markes td- 100 thermal desorption sample introduction system
Method Name:
SPF-TD-IS DOUBLE SPLIT.mth
Operating Mode
Standard Two Stage
Flow Path Temp:
180
°C
Min Carrier Pressure:
8.0
psi
GC Cycle Time:
19.0
min
Standby Mode
Split:
On
Split Flow:
20.0
mL/min
Pre-desorption
Purge Settings
Inject Std:
1.0
min
IS Flow:
40.0
mL/min
Loop Fill Time:
0.5
min
Prepurge Time:
0.1
min
Split On:
20.0
mL/min
Tube/Sample Desorption
SPI I Methods Development
Page 188
-------
Method Name:
SPF-TD-IS DOUBLE SPLIT.mth
Desorption Time:
8.0
min
Desorption Temp:
300
°C
Trap In Line
Trap Flow:
40.0
mL/min
Split:
On
Split Flow:
20.0
mL/min
Inlet Split Ratio:
1.5:1
Trap Settings
Trap Desorption Settings
Pre Trap Fire Purge:
1.0
min
Trap Flow:
20.0
mL/min
Split Flow:
19.0
mL/min
Outlet Split Ratio:
20.0:1
Total Split Ratio:
30.0:1
Trap Low Temp:
25
°C
Heating Rate:
MAX
°C/min
Trap High Temp:
310
°C
Trap Hold Time:
3.0
min
Table C- 2 Operating parameters for the agilent gcms
Agilent 7890A/5975C inert MSD with Triple-Axis Detector
Method Name:
SPFI TD1 MARKES
Method Run Time:
21.0 minutes
Transfer Line:
Restek Base Deactivated Guard Column (Cat. # 10000)
Transfer Line Dimensions:
1.0 mx0.25 mm
Column Model:
Restek Rtx®-5 Amine (Cat. # 12338)
Column Dimensions:
30 m x 0.25 mm ID 0.5 jjm df
Carrier Gas:
Helium
Inlet Type:
Split/Splitless (bypassed)
Flow Rate:
1.2 mL/min
Flow Mode:
Constant Flow
Inlet Pressure:
13.31 psi at 40°C oven temperature
Purge Time:
999.99
min
Purge Flow:
20.00
mL/min
Total Flow:
21.45
mL/min
Septum Purge:
0.1
mL/min
Septum Purge Mode:
Standard
Gas Saver:
Off
Oven Temp. Gradient:
40 °C hold for 2.0 min; 20°C/min to 300°C for 6.0 min
Injector Type:
Direct connection with transfer line from Markes TD100
MSD Interface:
280
°C
Source Temperature:
250
°C
MS Quad Temperature:
200
°C
Drawout Plate:
6 mm
Agilent # G2589-20045
Tune File:
atune.u
Acquisition Mode:
SCAN
Solvent Delay:
1.0
min
SCAN Parameters:
35 - 450 amu
SPFI Methods Development
Page 189
-------
Table C-3 Sources of standards for calibration of the TD-gc/ms system
Analytes
CAS#
Company
Catalogue #
Purity (%)
N,N,N',N",N"-
Pentamethyl-
diethylenetriamine
3030-47-5
Tokyo Chemical Industry
P0881
98.0
Chlorobenzene
108-90-7
Sigma-Aldrich
08650-5ML-F
99.7
1,2-Dichloropropane
78-87-5
Sigma-Aldrich
D72182-100G
99.0
1,4-Dioxane
123-91-1
Sigma-Aldrich
296309-
100ML
99.8
3-Chloro-1-propene
107-05-1
Sigma-Aldrich
236306-5ML
99.0
2-Methyl-2-pentenal
623-36-9
Sigma-Aldrich
294667-25G
97.0
1,4-Dichlorobenzene-
d4
Sigma-Aldrich
329339-1G
98.0
Acenaphthene-dio
Sigma-Aldrich
451819-1G
99.0
tris(1 -Chlor-2-propyl)
phosphate
13674-84-5
Fluka
32952-100MG
67.2
TCPP Isomer 1
26.2
TCPP Isomer 2
4.1
C-2 GC MS System for Analysis of Liquid Extracts from Various Media
The analytical system used for analysis of PMDETA and TCPP extracted from various media
consisted of a Hewlett-Packard 6809 gas chromatograph equipped with a series 6890 liquid
autosampler. This system is shown in Figure C-2. Operating parameters are provided in Table C-
4 and sources of standards are provided in Table C-5.
SPFI Methods Development
Page 190
-------
Figure C- 2 GCMS SYSTEM FOR ANALYSIS OF EXTRACTS OF MEDIA
Table C- 4 Hewlett-Packard 6890/5973 GC/MS system with a 6890 Series injector
Injection of liquid standards and sample extracts
Method Name:
SPFI-SIM-FR.M
Method Run Time:
18.5 minutes
Column Model:
Restek Rtx-5 Amine
Column Dimensions:
30m x 0.25mm 0.50 jjm
Carrier Gas:
Helium
Flow Rate:
1.0 mL/min
Oven Temp. Gradient:
50 °C hold for 1.0 min; 20°C/min to 300°C for 5.0 min; Post Run 310 °C for
1 min
Injector Type:
Agilent Split/Splitless
Inlet Mode:
Pulsed Splitless
Injection Port
Temperature:
250 °C; Pulsed pressure 30 PSI for 2.10 min; Purge time 2.00 min; Purge
Flow 35 ml/min.
Injection Volume:
2.0 mL
MSD Interface:
280 °C
Source Temperature:
240 °C
MS Quad Temperature:
190 °C
Tune File:
atune.u
Acquisition Mode:
SIM
Solvent Delay:
4.5 mins
SIM Parameters:
Group 1 (4.50 min) m/z-152, 68, 70, 72, 103, 115, 136, 150, 152
Group 2 (8.00 min) m/z-164, 80, 94. 103, 131, 143, 151, 157, 160, 162,
164, 167, 188, 201, 205, 231, 249, 251, 277, 279
Group 3 (13.00 min) m/z-240, 120, 191, 200, 236, 240, 264, 265, 381, 383
SPI I Methods Development
Page 191
-------
C-3 Operating Parameters for the LC-MSMS
The target isocyanates were quantified by analysis of the DBA derivatives of the target analytes
using a Agilent 1100 HPLC coupled to an ABSciex 3200 Triple Quadrapole Mass Spectrometer.
C-3.1 LC-MS/MS Analysis for Quantification of MDI, P3-MDI, and p4-MDi
Samples collected in impingers or on dry samplers were worked up for analysis using the ISO
17734a common protocol. Deuterated analogues of each isocyanate analyte were added to the
Figure C- 3 LCMSMS used for identification and quantification of isocyanates
toluene extraction solution for the dry sampler or directly to the impinger solution as internal
standards. Following sonication, the extraction solution was taken to dryness under a nitrogen
purge and the sample was reconstituted with acetonitrile. Calibration standards which are the
deuterated analogues of isocyanates followed the same protocol. Analysis of the derivatized
isocyanates was accomplished by injection from an autosampler into an LC-MS/MS shown in
Figure C-3 The response for each DBA derivative for each isocyanate in the calibration was
determined relative to the response of the corresponding deuterated analogue. A multipoint
calibration curve was generated at the beginning and at the end of each analysis sequence
SPI I Methods Development
Page 192
-------
Quality control procedures required that slope of the pre and post calibration runs must agree
within 20% and the R2 for the fits of the relative response factors must be 0.95 or better. If these
criteria are met, the calibration data are combined into one calibration file for data reduction.
Other QC procedures included analysis of reagent blanks, field blanks, duplicate samples, and
analysis of denuder-filter samplers and impinger samplers spiked by an independent analyst.
1. Agilent 1100 HPLC
2. ABSciex 3200 Triple Quadrupole Mass Spectrometer
3. Ascentis Express C18 HPLC Column Supelco 53822-U
4. Ascentis Express C18 HPLC Guard Column Supelco 53501-U
5. Mobile Phase A: 95% Acetonitrile/5%Water/0.5%Formic Acid
6. Mobile Phase B: 95% Water/5% Acetonitrile /0.5%Formic Acid
Table C- 5 API 3200 LC/MS/MS System Operating Parameters for analysis of isocyanate-dibutylamine
DERIVATIVES
CAS
Name
Abbreviation
M.W.
(functional)
M.W.
(DBA-derivative)
101-68-8
4,4'-Methylenediphenyl diisocyanate
4,4-MDI
250.3
508.7
Table C-l Chromatographic conditions and parameters
Instrument
Agilent 1100 HPLC/ Applied Biosystem API 3200 MS
Column
Ascentis Express C18, 5 cm x 2.1 mm I.D., 2.7 um particle size
(53822-U)
Guard Column
Ascentis Express C18, 0.5 cm x 2.1 mm I.D., 2.7 um particle size
(53822-U)
Oven Temperature
35 °C
Injection Volume
20 |jL
Mobile Phase
A = 5:95 Acetonitrile:Water
with 0.05% Formic acid
B = 95:5 Acetonitrile:Water with 0.05%
Formic Acid
Gradient
Step
Time (min)
Flow Rate (jjL/min)
%A
%B
0
0
400
60
40
1
2.00
400
20
80
2
5.00
400
20
80
3
5.10
400
60
40
4
8.00
400
60
40
Table C- 63 Source/Gas Parameters
Ion Source
Turbo Spray
Curtain Gas (CUR)
10
Collision Gas (CAD)
7
lonSpray Voltage (IS)
4500
Temperature (TEM)
425
Ion Source Gas (GS1)
14
Ion Source Gas 2 (GS2)
4
Interface Heater (ihe)
ON
Scan type
MRM
SPFI Methods Development
Page 193
-------
Polarity
Positive
Resolution Q1
Unit
Resolution Q2
Unit
Table C- 7 Analyte specific parameters
Analyte
Ql Mass
Q3 Mass
Time
DP
EP
CE
CXP
(amu)
(amu)
(msec)
(volts)
(volts)
(volts)
(volts)
4,4'-Methylenediphenyl
diisocyanate
509.4
130.2
600
71
4.5
37
4
p3 MDI
769.5
130.2
600
86
4.5
63
6
p4 MDI
1029.8
130.2
600
86
4.5
89
6
4,4'-Methylenediphenyl
diisocyanate-d18
527.4
139.2
600
71
4.5
37
4
Appendix D Quality Assurance and Control Results
D-l Evaluation of Physical Properties of the SPFI
Physical properties are characteristics that determine the overall usefulness of the insulation
produced. They effect the ability of the foam to produce the air barrier, and barriers to heat
transfer. They also can help determine if the foam produced was mixed in the correct ratios. SPFI
kits are designed by the manufacturers to produce foam that meet these characteristics. Two
component low pressure kits are usually designed to be used in one application, or a few
applications over a short period of time. Some manufacturers add additional compounds such as
colorants, which aid the applicator in determining if the foam produced is of the correct ratio.
The ASTM D22.05 SPF emissions characterization method development committee selected
High Density Polyethylene (HDPE) sheets as the substrate material the foam is applied to for
production of samples to be tested using the micro-scale emissions test method [24], HDPE
allows for easy release once the foam has cured. However, this eliminates the ability of testing
for the adhesive and cohesive properties, which is the ability of the foam to grip to the surface it
is applied to, and the ability of the foam to grip to itself, or grip to additional layers of foam.
When the substrate is a material that the foam is normally applied to, for example plywood, or
thin metal sheeting, then testing for adhesion and cohesion can be done when the manufacturers
time for curing has been met or exceeded. This time is usually 24 hours after application.
Cellular structure covers several properties of the foam ranging from the foam being an open cell
or closed cell foam. This refers to the foam cells being designed to rupture during curing or
staying intact during curing. Foam that is open cell has larger cells, is not as effective in
reduction of heat transfer, but works better in reducing noise transfer. Closed cell foams will
have a larger number of smaller cells and will be more effective in the reduction of heat transfer.
So, by cutting a sample from the cured sample foam, and looking at the cells under magnification
you can tell if the cells have remained intact and determine the size ranges of the cells.
SPFI Methods Development
Page 194
-------
The surface of the foam, or the skin can also give clues to whether the foam was produced
correctly. The skin should be flexible, but not be like the foam found in a seat cushion. If the cell
walls are easily damaged, or fragile the foam is off ratio. Off ratio refers to the ratio of side A to
side B which in these kits are usually a 50/50 mixture. The foam color should be uniform, areas
that are dark yellow to light brown are often off ratio.
As discussed in Section 2, the application proceeded normally through application to the first
three substrate frames. During application to the third quadrant of Frame 4, the spray gun
operator noticed that the foam was off-color, and the application was stopped prior to spraying
the fourth quadrant. The physical properties examination revealed that the cells of the second
quadrant were small, so it is likely that the application started to go off-ratio after the application
to the first quadrant of Frame 4. The foam applied to all quadrants passed the adhesion and
cohesion tests. Some voids between substrate and foam were apparent from some areas of Frame
4 where samples were removed for cohesion tests. Density and depth were within the target
specifications for the first three substrate frames. Refer to Table D-l for summary of the physical
properties.
Table D-l: Physical Properties of Foam Testing Method and Sampling Information
Physical
Properties
Reference Method
QA/QC
Testable
Equipment
Other equipment
used during testing
Samples per
Plywood Panel
Depth
CAN/ULC-S705.2-05 [25]
Calipers
Letter Opener
48 per sheet - 12 per
Quadrant
Density
CAN/ULC-S705.2-05
Balance,
Graduated
Cylinder
Knife
4 per sheet - 1 per
Quadrant
Adhesion
CAN/ULC-S705.2-05
Balance
Hole-saw, plywood
disks, test frame
4 per sheet - 1 per
Quadrant
Cohesion
CAN/ULC-S705.2-05
lance
Hole-saw, plywood
disks, test frame
4 per sheet - 1 per
Quadrant
Cell
Structure
CAN/ULC-S705.2-05
None
3X Handheld
Magnifier
4 per sheet - 1 per
Quadrant
Voids
CAN/ULC-S705.2-05
None
3X Handheld
Magnifier
4 per sheet - 1 per
Quadrant
CAN/ULC - Canadian Method - Underwriters Laboratories of Canada
SPFI Methods Development
Page 195
-------
Appendix E Model Files
The simulation program IECCU has several routines built-in for simulation of concentrations
during the application of a product in an indoor environment. The dual first order decay model
(Equation 10 and Source Type 23 in IECCU) is useful for simulating application phase SPF
insulation emissions because the emissions are due to multiple processes. The initial rise in
concentration that occurs during the application process is due to emissions from the stream of
reagents emitted from the spray gun as well as the emissions from rapidly forming and curing
foam. Peak concentrations are observed at the end of the spray event and decrease rapidly
following cessation of spraying. The changes in concentration are due to the rapid removal of
emissions through ventilation and the emissions from the product. The rates of emissions from
the product are impacted by several factors including the temperature of the product and the air,
turbulence near the surface of the product, concentrations of emissions at the surface of the
product, and concentrations of emissions in the air of the chamber.
The dual first order decay equation (Equation 10) is a reasonable empirical approach to
simulating this complex process. However, the user is required to determine the Source Type 23
inputs for IECCU, including the decay constants ki and k.2 and estimates of the mass of emissions
attributable to each of the decay constants. Obtaining the decay constants requires use of least
squares fitting software to obtain a fit of Equation 10 to the time-concentration data generated in
the test. The steps were outlined in Section 5.1.2.1. Details for estimation of the decay constants
ki and k.2 using a commercially available software system are provided in the following section.
Examples are provided for HFC 134a, PMDETA, and 1,4-dioxane.
In the examples provided, the numeric form of the Equation 10 is used to generate the constants
ki and k2. Because the application was conducted as two spray events separated by the 5-minute
period of sample collection the process is modeled as two events (Source 1 and Source 2). For
each spray event, the area of the source is divided into segments (th) and time (h) for application
to each segment is (stop time - start time)///,. Mass emitted (Mo) is determined by Equation 3.
The mass emitted per unit area is determined as the Mot A. The mass attributed to each spray
event is determined by the area sprayed (m2). The area sprayed in event 2 was less than in event
1 because the application was stopped when it appeared that the application was off-ratio. The %
attributable to rapid emissions was determined by judgement.
E-l Model Files for HFC-134a
Fitting the dual first order decay equation to the data in Scientist is a multistep process. The time-
concentration data are saved in a .csv file for importation into Scientist®. For HFC-134a, ten-minute
averages of the time-concentration data were saved in a .csv file. The Model File copied below was
entered into the Model File page and successfully compiled. A Parameter Set file was generated for the
model file. Additional information about the .csv file is entered into the parameter set file so that the
program correctly interprets the data. The .csv file is imported into the parameter page and the program
is run. The program outputs several additional files, including a plot of the time-concentration prediction
of the model with time-concentration data from the .csv file, calculated parameters, and summary
statistics describing the goodness of fit between the model output and the input data.
SPFI Methods Development
Page 196
-------
E-l.l Model File for HFC-134a
// Micromath Scientist Model File
//HFC-134a application phase
// independent variable, dependent variables, and parameters to be estimated
IndVars: t
DepVars: CT
Params: kl, k2
//5/06/2020 Trial Version 4: M0=1.66e05 mg/m2,
//PCT 0.90, 0.95, 0.925 tested to determine best fit for time - concentration data
//Input time-concentration data set reduced from 195 time periods to 20 by averaging every 10 entries for the 3.3-hour period
//Therefore, the data is entered as 10 minute averages of the time-concentration data generated by the photoacoustic
//spectrophotometer.
// Chamber volume (m3) and ventilation rate (1/h)
V = 30
N =4.102
// Number of incremental segments for each source
ni = 10
// Source 1 - A10 = area of source l(m2); Ali = area of each increment; M10 =initial mass applied (mg/m2)
A10 = 3.755
Ali = A10 / ni
M10=166262
// percent of mass applied that is available for rapid emissions
//This parameter estimated by least squares or professional judgement
pet = 0.90
// initial mass for rapid emissions
Ml_l = M10 * pet
// initial mass for slow emissions
M1_2=M10-M1_1
//start time 0.02 hours after entry into chamber and spray into bucket to prime lines and gun
Sl_start = 0.02
Sl_end = 0.15
Sl_app = Sl_end - Sl_start
Sl_ti = Sl_app / ni
// Source 1 - start time for each increment
Sl_tl = Sl_start + Sl_ti / 2
Sl_t2 = Sl_tl + Sl_ti
Sl_t3 = Sl_t2 + Sl_ti
Sl_t4 = Sl_t3 + Sl_ti
Sl_t5 = Sl_t4 + Sl_ti
Sl_t6 = Sl_t5 + Sl_ti
Sl_t7 = Sl_t6 + Sl_ti
Sl_t8 = Sl_t7 + Sl_ti
Sl_t9 = Sl_t8 + Sl_ti
Sl_tl0= Sl_t9 + Sl_ti
// elapsed time for each increment of source 1
tll=t-Sl_tl
tl2=t-Sl_t2
tl3=t-Sl_t3
tl4=t-Sl_t4
tl5=t-Sl_t5
tl6=t-Sl_t6
tl7=t-Sl_t7
tl8=t-Sl_t8
tl9=t-Sl_t9
tll0=t-Sl_tl0
// Source 1 - PI, P2 are the first part of the exact solution for concentration for each increment
SPFI Methods Development
Page 197
-------
Pl=Ali*Ml_l*kl/V/(N-kl)
P2=Ali*Ml_2*k2/V/(N-k2)
// concentration contributed by each source increment of source 1
Cl_l=Pl*(exp(-kl*tll)-exp(-N*tll))+P2*(exp(-k2*tll)-exp(-N*tll))
Cl_2=Pl*(exp(-kl*tl2)-exp(-N*tl2))+P2*(exp(-k2*tl2)-exp(-N*tl2))
Cl_3=Pl*(exp(-kl*tl3)-exp(-N*tl3))+P2*(exp(-k2*tl3)-exp(-N*tl3))
Cl_4=Pl*(exp(-kl*tl4)-exp(-N*tl4))+P2*(exp(-k2*tl4)-exp(-N*tl4))
Cl_5=Pl*(exp(-kl*tl5)-exp(-N*tl5))+P2*(exp(-k2*tl5)-exp(-N*tl5))
Cl_6=Pl*(exp(-kl*tl6)-exp(-N*tl6))+P2*(exp(-k2*tl6)-exp(-N*tl6))
Cl_7=Pl*(exp(-kl*tl7)-exp(-N*tl7))+P2*(exp(-k2*tl7)-exp(-N*tl7))
Cl_8=Pl*(exp(-kl*tl8)-exp(-N*tl8))+P2*(exp(-k2*tl8)-exp(-N*tl8))
Cl_9=Pl*(exp(-kl*tl9)-exp(-N*tl9))+P2*(exp(-k2*tl9)-exp(-N*tl9))
Cl_10=Pl*(exp(-kl*tll0)-exp(-N*tll0))+P2*(exp(-k2*tll0)-exp(-N*tll0))
// Source 1 - The source is in effect only after the increment is applied to surface
C1_R1 = IFGEZERO (til, Cl_l, 0)
C1_R2 = IFGEZERO (tl2, Cl_2, 0)
C1_R3 = IFGEZERO (tl3, Cl_3, 0)
C1_R4 = IFGEZERO (tl4, Cl_4, 0)
C1_R5 = IFGEZERO (tl5, Cl_5, 0)
C1_R6 = IFGEZERO (tl6, Cl_6, 0)
C1_R7 = IFGEZERO (tl7, Cl_7, 0)
C1_R8 = IFGEZERO (tl8, Cl_8, 0)
C1_R9 = IFGEZERO (tl9, Cl_9, 0)
C1_R10= IFGEZERO (tllO, Cl_10, 0)
// Concentration contributed by source 1 as a whole
Cl_Sum=Cl_Rl+Cl_R2+Cl_R3+Cl_R4+Cl_R5+Cl_R6+Cl_R7+Cl_R8+Cl_R9+Cl_R10
// Source 2
A20 = 3.285
A2i = A20 / ni
M20=166262
M2_l = M20 * pet
M2_2=M20-M2_1
S2_start = 0.26
S2_end = 0.37
S2_app = S2_end - S2_start
S2_ti = S2_app / ni
// Source 2 - start time for each increment
S2_tl = S2_start + S2_ti / 2
S2_t2 = S2_tl + S2_ti
S2_t3 = S2_t2 + S2_ti
S2_t4 = S2_t3 + S2_ti
S2_t5 = S2_t4 + S2_ti
S2_t6 = S2_t5 + S2_ti
S2_t7 = S2_t6 + S2_ti
S2_t8 = S2_t7 + S2_ti
S2_t9 = S2_t8 + S2_ti
S2_tl0= S2_t9 + S2_ti
// elapsed time for increments of source 2
t21=t-S2_tl
t22=t-S2_t2
t23=t-S2_t3
t24=t-S2_t4
t25=t-S2_t5
t26=t-S2_t6
SPFI Methods Development
Page 198
-------
t27=t-S2_t7
t28=t-S2_t8
t29=t-S2_t9
t210=t-S2_tl0
// Source 2 - Ql, Q2 are the first part of the exact solution for concentration for each increment
Ql=A2i*M2_l*kl/V/(N-kl)
Q2=A2i*M2_2* k2/V/(N -k2)
C2_l=Ql*(exp(-kl*t21)-exp(-N*t21))+Q2*(exp(-k2*t21)-exp(-N*t21))
C2_2=Ql*(exp(-kl*t22)-exp(-N*t22))+Q2*(exp(-k2*t22)-exp(-N*t22))
C2_3=Ql*(exp(-kl*t23)-exp(-N*t23))+Q2*(exp(-k2*t23)-exp(-N*t23))
C2_4=Ql*(exp(-kl*t24)-exp(-N*t24))+Q2*(exp(-k2*t24)-exp(-N*t24))
C2_5=Ql*(exp(-kl*t25)-exp(-N*t25))+Q2*(exp(-k2*t25)-exp(-N*t25))
C2_6=Ql*(exp(-kl*t26)-exp(-N*t26))+Q2*(exp(-k2*t26)-exp(-N*t26))
C2_7=Ql*(exp(-kl*t27)-exp(-N*t27))+Q2*(exp(-k2*t27)-exp(-N*t27))
C2_8=Ql*(exp(-kl*t28)-exp(-N*t28))+Q2*(exp(-k2*t28)-exp(-N*t28))
C2_9=Ql*(exp(-kl*t29)-exp(-N*t29))+Q2*(exp(-k2*t29)-exp(-N*t29))
C2_10=Ql*(exp(-kl*t210)-exp(-N*t210))+Q2*(exp(-k2*t210)-exp(-N*t210))
// Source 2 - The source is in effect only after the increment is applied to surface
C2_R1 = IFGEZERO (t21, C2_l, 0)
C2_R2 = IFGEZERO (t22, C2_2, 0)
C2_R3 = IFGEZERO (t23, C2_3, 0)
C2_R4 = IFGEZERO (t24, C2_4, 0)
C2_R5 = IFGEZERO (t25, C2_5, 0)
C2_R6 = IFGEZERO (t26, C2_6, 0)
C2_R7 = IFGEZERO (t27, C2_7, 0)
C2_R8 = IFGEZERO (t28, C2_8, 0)
C2_R9 = IFGEZERO (t29, C2_9, 0)
C2_R10= IFGEZERO (t210, C2_10, 0)
// Concentrations contributed by source 2
C2_sum=C2_Rl+C2_R2+C2_R3+C2_R4+C2_R5+C2_R6+C2_R7+C2_R8+C2_R9+C2_R10
// Mass balance model
CT=Cl_sum+C2_sum
E-1.2 Model Output File for HFC-134a
Plot of fitted model output using the calculated kl and k2 parameters versus the input concentration
time data (CT vs T).
SPFI Methods Development
Page 199
-------
HFC-134a Data Fitting Plot
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 3
Elapsed Time (hours)
Figure E-l. Scientist plot of fit of dual first order decay model to hrc-134a time-concentration data.
E-1.3 Statistics Report for HFC-134a Model fit
Microma th Scientist Sta tistics Report Microma th Scientist Sta tistics Report Microma th Scientist
Sta tistics Report Microma th Scientist Sta tistics Report Microma th Scientist Sta tistics Report
Input Information Input Information Input Information Input Information Input
Information
Model: ExplicitSlnHFC134a_0.02start95%.eqn
Da ta Set: Spreadsheet 4
Parameter Set: ExplicitSlnHFC134a_0.02start95%.mmp
Report Options Report Options Report Options Report Options Report Options
Descriptive Sta tistics N
Goodness-of-Fit Statistics Y
Confidence Intervals Y
Variance-Covariance Ma trix Y
CORRELA TION MA TRIX Y
Rigorous Limits N
SPFI Methods Development
Page 200
-------
Residual Analysis Y
Confidence Interval 95% 50
Goodness-of-Fit Statistics Goodness-of-Fit Statistics Goodness-of-Fit Statistics
Goodness-of-Fit Statistics Goodness-of-Fit Statistics
Data Column Name: CT
Weighted Unweighted
Sum of squared observations:
Sum of squared deviations:
Standard deviation of data:
R-squared: 0.99199
Coefficient of determination:
Correlation: 0.99491
5.3121E009
4.2528E007
470.64 470.64
0.99199
0.9892 0.9892
0.99491
5.3121E009
4.2528E007
Data Set Name: Spreadsheet 4
Weighted Unwei
Sum of squared observations:
Sum of squared deviations:
Standard deviation of data:
R-squared: 0.99199
Coefficient of determination:
Correlation: 0.99491
Model Selection Criterion:
5.3121E009 5.3121E009
4.2528E007 4.2528E007
470.64 470.64
0.99199
0.9892 0.9892
0.99491
4.5078 4.5078
Confidence Intervals Confidence Intervals Confidence Intervals Confidence Intervals
Confidence Intervals
Parameter Name: K1
Estimated Value: 75.565
Standard Deviation: 6.6122
95% Range (Univariate): 62.523 88.607
95% Range (Support Plane): 59.253 91.877
Parameter Name: K2
Estimated Value: 0.62855
Standard Deviation: 0.26931
95% Range (Univariate): 0.097357 1.1597
95% Range (Support Plane): -0.035838 1.2929
Variance-Covariance Matrix Variance-Covariance Matrix Variance-Covariance Matrix
Variance-Covariance Matrix Variance-Covariance Matrix
43.722
0.5236 0.07253
Correlation Matrix Correlation Matrix Correlation Matrix Correlation Matrix
Correlation Matrix
1
0.29403 1
SPFI Methods Development
Page 201
-------
Residual Analysis Residual Analysis Residual Analysis Residual Analysis
Residual Analysis
Expected Value: The following are normalized parameters with an expected value of 0.0. Values
ARE IN UNITS OF STANDARD DEVIATIONS FROM THE EXPECTED VALUE. THE FOLLOWING ARE NORMALIZED
PARAMETERS WITH AN EXPECTED VALUE OF 0.0. VALUES ARE IN UNITS OF STANDARD DEVIATIONS FROM THE EXPECTED
value. The following are normalized parameters with an expected value of 0.0. Values are in units of
STANDARD DEVIATIONS FROM THE EXPECTED VALUE.
Serial Correlation: 9.4823 indicates a systematic, non-random trend in the residuals
INDICATES A SYSTEMATIC, NON-RANDOM TREND IN THE RESIDUALS
Skewness -128.49 INDICATES the likelihood of a few large negative residuals having an
UNDULY LARGE EFFECT ON THE FIT. INDICATES THE LIKELIHOOD OF A FEW LARGE NEGATIVE RESIDUALS HAVING AN
UNDULY LARGE EFFECT ON THE FIT.
Kurtosis 208.89 IS PROBABLY NOT SIGNIFICANT. is probably not significant.
Weighting Factor: 0
Heteroscedacticity: 1.6809
Optimal Weighting Factor: 1.6809
E-1.4 HFC-134a Model Files for IECCU
Model parameter data input page screen shot from IECCU for the application phase source
Model 23 for HFC-134a. This page is filled for each application phase chemical. Note that the
application event is divided into two events for each chemical due to the discontinuous
application. This page shows the model inputs for the first application starting at 0.02 hours and
ending at 0.15 hours. As seen in the screen shot of the page that includes all inputs, there is a
second page with application start and end times of 0.26 and 0.37 hours. In this simulation, 95%
of the HFC-134a mass attributed to "fast" emissions and 5% of the mass is attributed to "slow"
emissions. Also note that "Total area (A)" of application is smaller for the second application
since the final quadrant of the last frame was not sprayed.
SPFI Methods Development
Page 202
-------
Error! Not a valid bookmark self-reference.
B Emission during application phase
Application-phase source models
(23) Dual first-order dacy
Dual first-order model for application-phase simulation
Ri(t) = Ai {M1 k1 exp[-k1 (t - tO)] + M2 k2 exp[-k2 (t - tO)]}
where
Rift) = emission rate for an incremental area at timet (ug/h)
Ai = area of the incremental source (m2)
M1 = amount of chemical for rapid emission (ug/m2)
M2 = amount of chemical for slow emission (ug/m2)
k1 = first-order decay constant for rapid emission (1/h)
k2 = first-order decay constant for slow emission (1/h)
t = elapsed time and t >= tO (h)
tO = time when the incremental area if applied (h)
NOTES:
To simulate the application-phase emissions, the source is divided into 200
incremental areas and the emission rate for each incremental area is calculated
separately.
M1 + M2 = total emittable amount of chemical available for emission.
Model parameters
Chemical name
Source location
Application start time (h)
Application end time (h)
Total area (A) (m2)
HFC134a
Zonel
0.15
3.755
Emittable amount -- rapid (M1) (ug/m2) 157.32E06
1st-order decay constant -- rapid (k1) (1/h) 119
Emittable amount — slow (M2) (ug/m2) 3.28E06
1st-order decay constant -- slow (k2) (1/h) 0.25
X
Figure E-2IECCU model file dual first order decay model
SPFI Methods Development
Page 203
-------
V. IECCU (v 1.0) HFC134aRun 1 _4_21 _2020_C ase_1 a,b,c.lEC
File Model Run Tools Help
(1) Building & Environment (2) Sources (3) Sinks (4) Airborne PM (5) Settled Dust (6) Chemical reactions (7) Simulation conditions (8) Output
a) Empirical models b) Application-phase c) Diffusion model d) Temperature-dependent K & D
Application-phase simulation models
Source type
Chemical
Start time (h)
End time (h)
Parameter3
Parameter 4
Parameters
© Add
E Edit
G Delete
App status: Awaiting user input
Current page = (2) Sources / b) Application-phase
0. Close
Figure E-3. Screen shot of IECCU application phase simulation parameter inputs for the dual first order
DECAY MODEL,
SPFI Methods Development
Page 204
-------
E-2 Model Files for 1,4-dioxane
E-2.1 Micromath Scientist Model File for 1,4-Dioxane
// Micromath Scientist Model File
//1,4-Dioxane application phase
// independent variable, dependent variables, and parameters to be estimated
IndVars: t
DepVars: CT
Params: kl, k2
//4/16/2020 Trial Version 2: M0=2.11E04
// Chamber volume (m3) and ventilation rate (1/h)
V = 30
N = 4.1
// Number of incremental segments for each source
ni = 10
// Source 1 - A10 = area of source l(m2); Ali = area of each increment; M10 =initial mass applied (mg/m2)
A10 = 3.76
Ali = A10 / ni
M10=21096
// percent of mass applied that is available for rapid emissions
// this parameter can be estimated by least squares,
pet = 0.9
// initial mass for rapid emissions
Ml_l = M10 * pet
// initial mass for slow emissions
M1_2=M10-M1_1
Sl_start = 0.021
Sl_end = 0.15
Sl_app = Sl_end - Sl_start
Sl_ti = Sl_app / ni
// Source 1 - start time for each increment
Sl_tl = Sl_start + Sl_ti / 2
Sl_t2 = Sl_tl + Sl_ti
Sl_t3 = Sl_t2 + Sl_ti
Sl_t4 = Sl_t3 + Sl_ti
Sl_t5 = Sl_t4 + Sl_ti
Sl_t6 = Sl_t5 + Sl_ti
Sl_t7 = Sl_t6 + Sl_ti
Sl_t8 = Sl_t7 + Sl_ti
Sl_t9 = Sl_t8 + Sl_ti
Sl_tl0= Sl_t9 + Sl_ti
// elapsed time for each increment of source 1
tll=t-Sl_tl
tl2=t-Sl_t2
tl3=t-Sl_t3
tl4=t-Sl_t4
tl5=t-Sl_t5
tl6=t-Sl_t6
tl7=t-Sl_t7
tl8=t-Sl_t8
tl9=t-Sl_t9
tll0=t-Sl_tl0
// Source 1 - PI, P2 are the first part of the exact solution for concentration for each increment
SPFI Methods Development
Page 205
-------
Pl=Ali*Ml_l*kl/V/(N-kl)
P2=Ali*Ml_2*k2/V/(N-k2)
// concentration contributed by each source increment of source 1
Cl_l=Pl*(exp(-kl*tll)-exp(-N*tll))+P2*(exp(-k2*tll)-exp(-N*tll))
Cl_2=Pl*(exp(-kl*tl2)-exp(-N*tl2))+P2*(exp(-k2*tl2)-exp(-N*tl2))
Cl_3=Pl*(exp(-kl*tl3)-exp(-N*tl3))+P2*(exp(-k2*tl3)-exp(-N*tl3))
Cl_4=Pl*(exp(-kl*tl4)-exp(-N*tl4))+P2*(exp(-k2*tl4)-exp(-N*tl4))
Cl_5=Pl*(exp(-kl*tl5)-exp(-N*tl5))+P2*(exp(-k2*tl5)-exp(-N*tl5))
Cl_6=Pl*(exp(-kl*tl6)-exp(-N*tl6))+P2*(exp(-k2*tl6)-exp(-N*tl6))
Cl_7=Pl*(exp(-kl*tl7)-exp(-N*tl7))+P2*(exp(-k2*tl7)-exp(-N*tl7))
Cl_8=Pl*(exp(-kl*tl8)-exp(-N*tl8))+P2*(exp(-k2*tl8)-exp(-N*tl8))
Cl_9=Pl*(exp(-kl*tl9)-exp(-N*tl9))+P2*(exp(-k2*tl9)-exp(-N*tl9))
Cl_10=Pl*(exp(-kl*tll0)-exp(-N*tll0))+P2*(exp(-k2*tll0)-exp(-N*tll0))
// Source 1 - The source is in effect only after the increment is applied to surface
C1_R1 = IFGEZERO (til, Cl_l, 0)
C1_R2 = IFGEZERO (tl2, Cl_2, 0)
C1_R3 = IFGEZERO (tl3, Cl_3, 0)
C1_R4 = IFGEZERO (tl4, Cl_4, 0)
C1_R5 = IFGEZERO (tl5, Cl_5, 0)
C1_R6 = IFGEZERO (tl6, Cl_6, 0)
C1_R7 = IFGEZERO (tl7, Cl_7, 0)
C1_R8 = IFGEZERO (tl8, Cl_8, 0)
C1_R9 = IFGEZERO (tl9, Cl_9, 0)
C1_R10= IFGEZERO (tllO, Cl_10, 0)
// Concentration contributed by source 1 as a whole
Cl_Sum=Cl_Rl+Cl_R2+Cl_R3+Cl_R4+Cl_R5+Cl_R6+Cl_R7+Cl_R8+Cl_R9+Cl_R10
// Source 2
A20 = 3.28
A2i = A20 / ni
M20=21096
M2_l = M20 * pet
M2_2=M20-M2_1
S2_start = 0.26
S2_end = 0.37
S2_app = S2_end - S2_start
S2_ti = S2_app / ni
// Source 2 - start time for each increment
S2_tl = S2_start + S2_ti / 2
S2_t2 = S2_tl + S2_ti
S2_t3 = S2_t2 + S2_ti
S2_t4 = S2_t3 + S2_ti
S2_t5 = S2_t4 + S2_ti
S2_t6 = S2_t5 + S2_ti
S2_t7 = S2_t6 + S2_ti
S2_t8 = S2_t7 + S2_ti
S2_t9 = S2_t8 + S2_ti
S2_tl0= S2_t9 + S2_ti
// elapsed time for increments of source 2
t21=t-S2_tl
t22=t-S2_t2
t23=t-S2_t3
t24=t-S2_t4
t25=t-S2_t5
t26=t-S2_t6
SPFI Methods Development
Page 206
-------
t27=t-S2_t7
t28=t-S2_t8
t29=t-S2_t9
t210=t-S2_tl0
// Source 2 - Ql, Q2 are the first part of the exact solution for concentration for each increment
Ql=A2i*M2_l*kl/V/(N-kl)
Q2=A2i*M2_2* k2/V/(N -k2)
C2_l=Ql*(exp(-kl*t21)-exp(-N*t21))+Q2*(exp(-k2*t21)-exp(-N*t21))
C2_2=Ql*(exp(-kl*t22)-exp(-N*t22))+Q2*(exp(-k2*t22)-exp(-N*t22))
C2_3=Ql*(exp(-kl*t23)-exp(-N*t23))+Q2*(exp(-k2*t23)-exp(-N*t23))
C2_4=Ql*(exp(-kl*t24)-exp(-N*t24))+Q2*(exp(-k2*t24)-exp(-N*t24))
C2_5=Ql*(exp(-kl*t25)-exp(-N*t25))+Q2*(exp(-k2*t25)-exp(-N*t25))
C2_6=Ql*(exp(-kl*t26)-exp(-N*t26))+Q2*(exp(-k2*t26)-exp(-N*t26))
C2_7=Ql*(exp(-kl*t27)-exp(-N*t27))+Q2*(exp(-k2*t27)-exp(-N*t27))
C2_8=Ql*(exp(-kl*t28)-exp(-N*t28))+Q2*(exp(-k2*t28)-exp(-N*t28))
C2_9=Ql*(exp(-kl*t29)-exp(-N*t29))+Q2*(exp(-k2*t29)-exp(-N*t29))
C2_10=Ql*(exp(-kl*t210)-exp(-N*t210))+Q2*(exp(-k2*t210)-exp(-N*t210))
// Source 2 - The source is in effect only after the increment is applied to surface
C2_R1 = IFGEZERO (t21, C2_l, 0)
C2_R2 = IFGEZERO (t22, C2_2, 0)
C2_R3 = IFGEZERO (t23, C2_3, 0)
C2_R4 = IFGEZERO (t24, C2_4, 0)
C2_R5 = IFGEZERO (t25, C2_5, 0)
C2_R6 = IFGEZERO (t26, C2_6, 0)
C2_R7 = IFGEZERO (t27, C2_7, 0)
C2_R8 = IFGEZERO (t28, C2_8, 0)
C2_R9 = IFGEZERO (t29, C2_9, 0)
C2_R10= IFGEZERO (t210, C2_10, 0)
// Concentrations contributed by source 2
C2_sum=C2_Rl+C2_R2+C2_R3+C2_R4+C2_R5+C2_R6+C2_R7+C2_R8+C2_R9+C2_R10
// Mass balance model
CT=Cl_sum+C2_sum
E-2.2 1,4-Dioxane Data Input File
1,4-dioxane time - concentration input file (CSV file) for estimating kl and k2 by nonlinear least-squares
fitting in a commercial software system. Elapsed time (t) is in hours, and chamber concentration (Ct) is in
Hg nr3.
t
Ct
0
0
0.07
602.6
0.17
1804.6
0.27
1389.6
0.41
2159.8
0.58
1129.4
0.75
594.1
0.94
239.9
SPFI Methods Development
Page 207
-------
1.10 191.6
1.27 114.6
1.48 71.8
1.97 36.1
3.06 25.5
E-2.3 1,4-Dioxane Plot
1,4-Dioxane ki and k2 determined by nonlinear least square fit of Equation 8 to time - concentration
data. The figure below shows the time-concentration data plotted with model prediction (Ct) calculated
using ki (70.6) and k2 (0.57).
Elapsed Time (hours)
Figure E-4. Scientist plot of fit of dual first order decay model to 1,4-dioxane time - concentration data.
E-2.4 1,4-dioxane files for IECCU
The first order decay constants for "fast" and "slow" emissions are entered in Application Phase
Model 23 of the simulation program IECCU as Parameter 3 and Parameter 5. Parameters 2 and 4
SPFI Methods Development Page 208
-------
' IECCU (v 1.0) 1,4dioxaneRun1 _4_24_2020_Case_1 _S5IEC.I EC
¦ile Model Run Tools Help
(1) Building & Environment (2) Sources (3) sinks (4) Airborne PM (5) Settled Dust (6) Chemical reactions (7) Simulation conditions (8) Output
a) Empirical models b) Application-phase c) Diffusion model d) Temperature-dependent K & D
Application-phase simulation models
1
2
3
4
5
6
7
8
9
10
Source type
23
23
Chemical
1,4-dioxane
1,4-dioxane
©Add
Zone ID
1
1
Start time (h)
0.02
0.26
End time (h)
0.15
0.37
m Edit
Parameter 1
3.755
3.2S5
Parameter 2
17932
17932
Parameter 3
74.3
74.3
O Delete
Parameter 4
3164.5
3164.5
Parameter5
0.609
0.609
<
>
App status: Awaiting user input Current page = (2) Sources / b) Application-phase
Figure E-5 IECCU input page for 1,4-dioxane
y. Close
are the mass available for "fast" and "slow" emissions determined as 85% and 15% of Mo,
respectively.
SPI I Methods Development
Page 209
-------
¦ Emission during application phase
Application-phase source models
(23) Dual first-order dacy
Model parameters
Chemical name
1,4-dioxane
Dual first-order model for application-phase simulation
Ri(t) = Ai {M1 k1 exp[-k1 (t - tO)] + M2 k2 exp[-k2 (t - tO)]}
where
Ri(t) = emission rate for an incremental area at timet (ug/h)
Ai = area of the incremental source (m2)
M1 = amount of chemical for rapid emission (ug/m2)
M2 = amount of chemical for slow emission (ug/m2)
k1 = first-order decay constant for rapid emission (1/h)
k2 = first-order decay constant for slow emission (1/h)
t = elapsed time and t > = tO (h)
tO = time when the incremental area if applied (h)
NOTES:
To simulate the application-phase emissions, the source is divided into 200
incremental areas and the emission rate for each incremental area is calculated
separately.
M1 + M2 = total emittable amount of chemical available for emission.
Source location
Application start time (h)
Application end time (h)
Total area (A) (m2)
Zonel
0.37
3.285
Emittable amount -- rapid (M1) (ug/m2) 17932
1st-order decay constant — rapid (k1) (1/h) 74.3
Emittable amount -- slow (M2) (ug/m2) 3164.5
1st-order decay constant — slow (k2) (1/h) 0.609
OK
X Cancel
Figure E-6IECCU model parameter entry page for the dual first order decay model.
E-3 Model files for PMDETA
b-3.1 Micromath Scientist Model File for PMDbTA
// Micromath Scientist Model File
//PMDETA application phase
// independent variable, dependent variables, and parameters to be estimated
IndVars: t
DepVars: CT
Params: kl, k2
//6/21/2019 Version 5: M0=137870 ug/m2, PCT 0.90 chamber concentrations used to calculate M0, apparent outlier at 0.79 h
removed
// Chamber volume (m3) and ventilation rate (1/h)
V = 30
N =4.102
// Number of incremental segments for each source
ni = 10
// Source 1 -- A10 = area of source l(m2); Ali = area of each increment; M10 =initial mass applied (mg/m2)
A10 = 3.76
Ali = A10 / ni
M10=137870
// percent of mass applied that is available for rapid emissions
//This parameter estimated by least squares,
pet = 0.90
SPFI Methods Development
Page 210
-------
// initial mass for rapid emissions
Ml_l = M10 * pet
// initial mass for slow emissions
M1_2=M10-M1_1
//start time set to zero to capture spray in bucket
Sl_start = 0.00
Sl_end = 0.15
Sl_app = Sl_end - Sl_start
Sl_ti = Sl_app / ni
// Source 1 - start time for each increment
Sl_tl = Sl_start + Sl_ti / 2
Sl_t2 = Sl_tl + Sl_ti
Sl_t3 = Sl_t2 + Sl_ti
Sl_t4 = Sl_t3 + Sl_ti
Sl_t5 = Sl_t4 + Sl_ti
Sl_t6 = Sl_t5 + Sl_ti
Sl_t7 = Sl_t6 + Sl_ti
Sl_t8 = Sl_t7 + Sl_ti
Sl_t9 = Sl_t8 + Sl_ti
Sl_tl0= Sl_t9 + Sl_ti
// elapsed time for each increment of source 1
tll=t-Sl_tl
tl2=t-Sl_t2
tl3=t-Sl_t3
tl4=t-Sl_t4
tl5=t-Sl_t5
tl6=t-Sl_t6
tl7=t-Sl_t7
tl8=t-Sl_t8
tl9=t-Sl_t9
tllO=t-Sl_tlO
// Source 1 - PI, P2 are the first part of the exact solution for concentration for each increment
Pl=Ali*Ml_l*kl/V/(N-kl)
P2=Ali*Ml_2*k2/V/(N-k2)
// concentration contributed by each source increment of source 1
Cl_l=Pl*(exp(-kl*tll)-exp(-N*tll))+P2*(exp(-k2*tll)-exp(-N*tll))
Cl_2=Pl*(exp(-kl*tl2)-exp(-N*tl2))+P2*(exp(-k2*tl2)-exp(-N*tl2))
Cl_3=Pl*(exp(-kl*tl3)-exp(-N*tl3))+P2*(exp(-k2*tl3)-exp(-N*tl3))
Cl_4=Pl*(exp(-kl*tl4)-exp(-N*tl4))+P2*(exp(-k2*tl4)-exp(-N*tl4))
Cl_5=Pl*(exp(-kl*tl5)-exp(-N*tl5))+P2*(exp(-k2*tl5)-exp(-N*tl5))
Cl_6=Pl*(exp(-kl*tl6)-exp(-N*tl6))+P2*(exp(-k2*tl6)-exp(-N*tl6))
Cl_7=Pl*(exp(-kl*tl7)-exp(-N*tl7))+P2*(exp(-k2*tl7)-exp(-N*tl7))
Cl_8=Pl*(exp(-kl*tl8)-exp(-N*tl8))+P2*(exp(-k2*tl8)-exp(-N*tl8))
Cl_9=Pl*(exp(-kl*tl9)-exp(-N*tl9))+P2*(exp(-k2*tl9)-exp(-N*tl9))
Cl_10=Pl*(exp(-kl*tll0)-exp(-N*tll0))+P2*(exp(-k2*tll0)-exp(-N*tll0))
// Source 1 - The source is in effect only after the increment is applied to surface
C1_R1 = IFGEZERO (til, Cl_l, 0)
C1_R2 = IFGEZERO (tl2, Cl_2, 0)
C1_R3 = IFGEZERO (tl3, Cl_3, 0)
C1_R4 = IFGEZERO (tl4, Cl_4, 0)
C1_R5 = IFGEZERO (tl5, Cl_5, 0)
C1_R6 = IFGEZERO (tl6, Cl_6, 0)
C1_R7 = IFGEZERO (tl7, Cl_7, 0)
C1_R8 = IFGEZERO (tl8, Cl_8, 0)
C1_R9 = IFGEZERO (tl9, Cl_9, 0)
C1_R10= IFGEZERO (tllO, Cl_10, 0)
SPFI Methods Development
Page 211
-------
// Concentration contributed by source 1 as a whole
Cl_Sum=Cl_Rl+Cl_R2+Cl_R3+Cl_R4+Cl_R5+Cl_R6+Cl_R7+Cl_R8+Cl_R9+Cl_R10
// Source 2
A20 = 3.28
A2i = A20 / ni
M20=137870
M2_l = M20 * pet
M2_2=M20-M2_1
S2_start = 0.26
S2_end = 0.37
S2_app = S2_end - S2_start
S2_ti = S2_app / ni
// Source 2 - start time for each increment
S2_tl = S2_start + S2_ti / 2
S2_t2 = S2_tl + S2_ti
S2_t3 = S2_t2 + S2_ti
S2_t4 = S2_t3 + S2_ti
S2_t5 = S2_t4 + S2_ti
S2_t6 = S2_t5 + S2_ti
S2_t7 = S2_t6 + S2_ti
S2_t8 = S2_t7 + S2_ti
S2_t9 = S2_t8 + S2_ti
S2_tl0= S2_t9 + S2_ti
// elapsed time for increments of source 2
t21=t-S2_tl
t22=t-S2_t2
t23=t-S2_t3
t24=t-S2_t4
t25=t-S2_t5
t26=t-S2_t6
t27=t-S2_t7
t28=t-S2_t8
t29=t-S2_t9
t210=t-S2_tl0
// Source 2 - Ql, Q2 are the first part of the exact solution for concentration for each increment
Ql=A2i*M2_l*kl/V/(N-kl)
Q2=A2i*M2_2* k2/V/(N -k2)
C2_l=Ql*(exp(-kl*t21)-exp(-N*t21))+Q2*(exp(-k2*t21)-exp(-N*t21))
C2_2=Ql*(exp(-kl*t22)-exp(-N*t22))+Q2*(exp(-k2*t22)-exp(-N*t22))
C2_3=Ql*(exp(-kl*t23)-exp(-N*t23))+Q2*(exp(-k2*t23)-exp(-N*t23))
C2_4=Ql*(exp(-kl*t24)-exp(-N*t24))+Q2*(exp(-k2*t24)-exp(-N*t24))
C2_5=Ql*(exp(-kl*t25)-exp(-N*t25))+Q2*(exp(-k2*t25)-exp(-N*t25))
C2_6=Ql*(exp(-kl*t26)-exp(-N*t26))+Q2*(exp(-k2*t26)-exp(-N*t26))
C2_7=Ql*(exp(-kl*t27)-exp(-N*t27))+Q2*(exp(-k2*t27)-exp(-N*t27))
C2_8=Ql*(exp(-kl*t28)-exp(-N*t28))+Q2*(exp(-k2*t28)-exp(-N*t28))
C2_9=Ql*(exp(-kl*t29)-exp(-N*t29))+Q2*(exp(-k2*t29)-exp(-N*t29))
C2_10=Ql*(exp(-kl*t210)-exp(-N*t210))+Q2*(exp(-k2*t210)-exp(-N*t210))
// Source 2 - The source is in effect only after the increment is applied to surface
C2_R1 = IFGEZERO (t21, C2_l, 0)
C2_R2 = IFGEZERO (t22, C2_2, 0)
C2_R3 = IFGEZERO (t23, C2_3, 0)
C2_R4 = IFGEZERO (t24, C2_4, 0)
C2_R5 = IFGEZERO (t25, C2_5, 0)
SPFI Methods Development
Page 212
-------
C2_R6 = IFGEZERO (t26, C2_6, 0)
C2_R7 = IFGEZERO (t27, C2_7, 0)
C2_R8 = IFGEZERO (t28, C2_8, 0)
C2_R9 = IFGEZERO (t29, C2_9, 0)
C2_R10= IFGEZERO (t210, C2_10, 0)
// Concentrations contributed by source 2
C2_sum=C2_Rl+C2_R2+C2_R3+C2_R4+C2_R5+C2_R6+C2_R7+C2_R8+C2_R9+C2_R10
// Mass balance model
CT=Cl_sum+C2_sum
E-3.2 Data Input File
CSV data input file for the model
t
Ct
0
0
0.07
3423
0.25
8897
0.26
9872
0.42
13097
0.44
13691
0.73
4684
1.09
903
1.44
546
1.76
428
2.8
170
E-3.3 Plot of Model Output for PMDETA
PMDETA dual first order decay model fit to concentration-time data using Scientist® to generate decay coefficients kl and k2.
SPFI Methods Development
Page 213
-------
¦ Ct vs t
— CT Calc vs I
—
1 2
Elapsed Time (hours)
Figure E-7. Scientist plot of fit of dual first order decay model to PMDETA time-concentration data.
SPI I Methods Development
Page 214
------- |