The Open Collections website will be unavailable July 27 from 2100-2200 PST ahead of planned usability and performance enhancements on July 28. More information here.

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Field-based evaluation of processes and models for soil vapour intrusion into buildings Hers, Ian 2004

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata


831-ubc_2004-994309.pdf [ 21.13MB ]
JSON: 831-1.0063482.json
JSON-LD: 831-1.0063482-ld.json
RDF/XML (Pretty): 831-1.0063482-rdf.xml
RDF/JSON: 831-1.0063482-rdf.json
Turtle: 831-1.0063482-turtle.txt
N-Triples: 831-1.0063482-rdf-ntriples.txt
Original Record: 831-1.0063482-source.json
Full Text

Full Text

FIELD-BASED EVALUATION OF PROCESSES AND MODELS FOR SOIL VAPOUR INTRUSION INTO BUILDINGS By IAN HERS B.A.Sc, The University of British Columbia, 1986 M.A.Sc, The University of British Columbia, 1989 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY In THE FACULTY OF GRADUATE STUDIES Department of Civil Engineering We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA August, 2004 © Ian Hers, 2004 11 ABSTRACT This thesis presents the results of research on the validation of models used to evaluate the intrusion of subsurface volatile organic compounds (VOCs) into buildings, often referred to as the "soil vapour transport to indoor air pathway". Evaluation of this exposure pathway is important in the context of risk-based corrective action for contaminated sites. The study scope addresses both the fate and transport of VOCs in the unsaturated zone, and intrusion of VOCs through the subsurface building envelope (i.e., foundation). The primary approach adopted to investigate and validate models was to obtain extensive field monitoring data on V O C vapour transport and intrusion, obtained at a former petro-chemical plant site ("Chatterton" site). The Chatterton site, located near Vancouver, B.C., Canada, is contaminated with benzene, toluene and xylenes (BTX). To facilitate measurement of B T X intrusion, a small building (greenhouse) with controlled properties was constructed. A number of experiments were conducted to evaluate vadose zone processes and vapour intrusion for different greenhouse depressurization conditions. When the greenhouse was not subject to sustained fan-induced depressurization, there was significant aerobic biodegradation of B T X vapours between approximately 0.4 m and 0.8 m depth below the greenhouse, and subslab B T X vapour concentrations were low. When the greenhouse was depressurized to -10 pascals (Pa), the subslab B T X vapour concentrations were elevated, and significant vapour intrusion was measured using both tracer and flux chamber techniques, which was inferred to be a result of an upward B T X vapour flux that exceeded biodegradation capacity based on oxygen availability. A comprehensive analysis of case studies providing information on soil vapour intrusion was completed. This analysis, together with the Chatterton site results, were evaluated for key trends and factors affecting soil vapour intrusion. The vapour attenuation factors estimated based on field case studies were compared to those predicted using several screening level models, including the Johnson and Ettinger Model, a widely used model for this exposure pathway. The validation of screening models using field data is complicated in that model predictions can vary greatly depending on the model and input parameters used. Further, soil vapour fate and transport and intrusion into buildings is a complex process affected by numerous factors (soil properties, house conditions, and environmental factors); therefore, vapour intrusion will vary significantly depending on site specific conditions. The results of the model comparisons to field data indicated that screening level models, typically used for this exposure pathway, were found to generally yield conservative results (i.e., overpredict vapour intrusion); however, there were a few important exceptions. A multi-dimensional numerical model for vapour transport, which incorporated diffusion, biodegradation, sorption and soil gas advection, was also developed. Good comparisons between model predicted and measured vapour attenuation were obtained based on conditions observed at the Chatterton site. TABLE OF CONTENTS Abstract ii Table of Contents i i i List of Tables viii List of Figures x Acknowledgements xv Chapter 1 Introduction 1 Key Issues and Research Relevancy 1 Conceptual Site Model 2 Predictive Models and Framework for Their Use 3 Model Validation 3 Data Gaps and Questions Left to Be Answered 4 Purpose, Objectives and Scope 5 Background Information and Overview of Research Program 6 Structure of this Thesis 7 References 9 Chapter 2 Evaluation of Soil Vapour Sampling and Analysis Techniques 11 Introduction 12 Background Information Review 12 Probe Design for Soil Gas Collection 12 Sampling Collection and Handling 13 Analytical Techniques 14 Field Screening Methods 14 Laboratory Analytical Techniques 14 Evaluation of Experimental Methods 15 Rationale 15 Overview of Soil Gas Sampling and Analysis Methods at Chatterton Site.... 16 Tedlar Bag Permeation Study 17 Methods 17 Results 17 Syringe Sorption Study ...17 Methods 17 Results 18 P V C Sorption Study 18 Methods < 18 Results 18 Light Gas Analysis Study 20 Methods 20 Results 20 iv Consistency of Soil Gas with Biogeochemical Processes 22 B T X Analytical Quality Control Study 23 Methods 23 Results .• 24 Discussion of Experimental Techniques and Lessons Learned 24 Probe Design 24 Sampling Collection and Analysis 25 Analytical Techniques 26 Field Screening Methods.. 26 Laboratory Analytical Techniques 26 Conclusions 28 References 29 Chapter 3 Measurement of In situ Gas-Phase Diffusion Coefficients 31 Introduction 32 Background 33 Theory 35 Methods 37 Results 40 Sensitivity Analyses 44 Conclusions 44 References 46 Chapter 4 Evaluation of Vadose Zone Biodegradation of BTX Vapours 47 Introduction 48 Intrinsic Biodegradation 48 Processes and Effect of Building 48 Biodegradation Models and Kinetics 49 Evaluation of Biodegradation Potential 50 B T E X Attenuation 50 Geochemical Indicators 51 Laboratory Studies 52 Field study 52 Overview 52 Site Description 56 Methods 53 Baseline Soil Testing Results 55 Soil Gas Monitoring Results 56 Estimation of Field Biodegradation Rates 59 Comparison to Other Case Studies 60 Numerical Simulation of Vadose Zone B T X Transport 62 Model Description 62 Comparison to Analytical Solutions 65 Comparison of Biodegradation Models No. 1, 2, 3 and 4 (1-D Diffusion and Biodecay) 66 Comparison to Chatterton Vapour Profiles (2-D Diffusion and Biodecay).... 68 V Sensitivity Analysis Adjacent to Building at Chatterton site (1-D Diffusion and Biodecay) 71 Evaluation of Advective Transport at Chatterton site (1-D Diffusion, Advection and Biodecay) 72 Conclusions 74 References 75 Chapter 5 Measurement of BTX Vapour Intrusion into an Experimental Building 77 Introduction 78 Background Review and Case Studies 78 Soil Gas Models 79 Analytical Models 79 Numerical Models 80 Emperical Methods 81 Experimental Methods 81 Research Site Description 81 Description of Greenhouse, Soil Gas Probes and Soil-Air Permeability Testing ...82 Soil Gas, Pressure, Temperature and Oxygen Monitoring Methods 83 Flux Chamber 84 Soil Gas Flow Rate and B T X Flux into Greenhouse Measured Using Tracer Test 85 Soil Gas Flow Rate into Greenhouse Measured Using Flux Chamber 86 B T X Flux Rates into Greenhouse Measured Using Flux Chamber 86 Experimental Results 87 Overview Soil Gas Intrusion Testing Program 87 Soil Air Permeability 87 Soil Gas Flow Rate and B T X Flux into Greenhouse Measured Using Tracer Test 88 Soil Gas Flow Rates Measured Using Flux Chamber 88 B T X Flux Rates into Building Measured Using Flux Chamber 93 Non Depressurized Condition 93 Subslab Oxygen, Pressure and Weather Monitoring 94 Depressurized Condition 94 Comparison to Model Predictions 95 Discussion 98 Conclusions 99 References 100 Chapter 6 Comparision, Validation and Use of Models for Predicting Indoor Air Quality From Soil and Groundwater Contamination 102 Introduction 103 Model Review 104 Introduction 104 Source Zone Model Attributes 104 Unsaturated Zone Model Attributes 106 vi Diffusion and Advection 106 Biodegradation 106 Building Foundation Model Attributes 108 Diffusion 108 Advection 108 Selection of Models for Validation 109 Model Validation - Effect of Transport Processes 110 Model Validation - Model Sensitivity and Effect of Parameter Uncertainty 112 Sensitivity Analysis 112 Effect of Parameter Uncertainty 113 Effect of Uncertainty in Soil Moisture Content and Bulk Density 115 Model Validation - Comparisons Based On Field Data 117 Description of Case Study Sites 118 Biodegradation Rates 119 Soil Gas Advection and Flow into Buildings 119 Soil Vapour Intrusion into Buildings 121 Chatterton Site Results 121 Comparisons of Measured and Predicted Vapour Attenuation Ratios.... 124 Proposed Tiered Framework, Discussion and Guidance 125 Proposed Tiered Framework 125 Advection Potential 127 Source Depletion 128 Biodegradation 129 Concluding Thoughts 129 References 130 Chapter 7 Evaluation of the Johnson and Ettinger Model for Prediction of Indoor Air Quality 132 Introduction 133 J&E Model Input Parameters, Sensitivity and Uncertainty 133 Estimation of Effective Diffusion Coefficient (Air-Filled and Total Porosity) 133 Estimation of Soil Gas Advection Rate (QSOii) 135 J&E Model Sensitivity for Key Input Parameters 137 Sensitivity of dp to Q S Oii 139 Sensitivity of Op to Crack Ratio 139 Sensitivity of Op to Air-Filled Porosity (Moisture Content) 140 J&E Model Uncertainty for Range of Values 140 Field-Based Methods For Evaluation Of Vapour Intrusion 143 Results And Discussion Of Field Studies And Model Predictions 144 Indoor V O C Method 144 Measured Vapour Attenuation Ratios at Petroleum Hydrocarbon Sites. 148 Comparison to Model Predictions for Petroleum Hydrocarbon Sites 149 Measured Vapour Attenuation Ratios at Chlorinated Solvent Sites 149 Comparison to Model Predictions for Chlorinated Solvent Sites 151 Tracer Method 151 Vll Flux Chamber Method 151 Regulatory Implications 152 Conclusions and Recommendations 153 References 155 Chapter 8 The Use Of Indoor Air Measurements to Evaluate Exposure And Risk from Subsurface VOCs 157 Introduction 158 Mechanisms for Soil Vapour Intrusion 158 Process Overview 158 Building Underpressurization and Ventilation 160 Indoor Air Quality 162 Issues for the Use of Indoor Air Measurements to Quantify Soil Vapour Sources 168 Significance Of Subsurface Relative To Indoor V O C Sources 171 Conclusions 176 References 176 Chapter 9 Conclusions 180 Evaluation of soil vapour sampling and analysis techniques 180 Measurement of in situ gas-phase diffusion coefficients 180 Evaluation of vadose zone biodegradation of B T X vapours 181 Measurement of B T X Vapour Intrusion into an Experimental Building 181 Comparison, validation and use of models for predicting indoor air quality from soil and groundwater contamination 182 Evaluation of the Johnson and Ettinger model for prediction of indoor air quality 183 The use of indoor air measurements to evaluate exposure and risk from subsurface VOCs 183 Significant research Contributions 184 Recommendations for Further research 184 Vll l LIST OF TABLES Chapter 2. Table 2.1 Summary of Quality Control Testing for B T X Analyses 24 Chapter 3. Table 3.1 Comparison of measured and empirical gas-phase tortuosity factor from laboratory diffusion studies 34 Table 3.2 Thermal conductivity of selected gases 38 Table 3.3 Effect of soil gas containing CO2 on Mark 9822 helium detector response 39 Table 3.4 Measured mass required and beta parameter for push-pull tests 42 Table 3.5 Measured in situ effective diffusion coefficients based on push-pull tests 43 Chapter 4. Table 4.1 Results of nitrogen, sulpher and S O M analyses in soil 56 Table 4.2 Measured aerobic biodegradation rates from field studies 61 Table 4.3 Input Parameters for Comparison of Biodegradation Models 67 Chapter 5. Table 5.1 Summary of Soil Gas Intrusion Test Cases 89 Table 5.2 Results of Soil Gas Intrusion Tests 90 Table 5.3 B T X Flux Chamber Concentrations and Flux into Greenhouse at Edge Cracks 93 Chapter 6. Table 6.1 Summary of Model Characteristics 105 Table 6.2 Default Generic Values for Input Parameters (Benzene) 110 Table 6.3 Model Comparisons for Varying Processes 111 Table 6.4 Sensitivity of Predicted Vapour Attenuation Ratio to Parameter Uncertainty for Hypothetical Site (diffusion & advection model case) 114 Table 6.5 Effect of Parameter Uncertainty on Tortuosity Factor Estimates... 115 Table 6.6 Field Data Summary for Model Validation 117 Table 6.7 Comparison of Measured and Predicted Soil Gas Flow Rates into Building, 120 Table 6.8 Comparison of Measured Soil Gas Flow Rates into Buildings 120 Table 6.9 Recommended Soil Vapour Model Attributes 126 Chapter 7. Table 7.1 Comparison of measured and model-predicted soil gas flow rates into buildings 138 Table 7.2 Qualitative summary of sensitive parameters for the J & E model ..138 Table 7.3 Uncertainty analysis for nomalized effective diffusion coefficient 142 Table 7.4 Measured and Model-Predicted Vapour Attenuation Ratios 147 Chapter 8. Table 8.1 Survey of Building Depressurization Relative to Atmospheric Pressure 161 Table 8.2 Survey of Building Ventilation Rates 163 ix Table 8.3 Dominant Sources of Selected VOCs (adapted from Davis and Otson, 1993; Otson and Fellin, 1992) 165 Table 8.4 Compilation of Indoor Air Quality Data 166 Table 8.5 Assumed Soil Vapour Concentrations at Contaminated Source used for Modeling 174 Table 8.6 Exposure and Risk Factors .". 174 X LIST OF FIGURES Chapter 1 Figure 1.1 Conceptual Site Model for Soil Vapour Intrusion into Buildings 2 Chapter 2. Figure 2.1 Soil Gas Probes and Sampling Train Used At Chatterton Site 16 Figure 2.2 Results of Tedlar Bag Permeation Study: O2 and N 2 Concentrations 16 Figure 2.3 Results of Syringe Desorption Study: Gas Chromatograph Benzene Area Counts 18 Figure 2.4 Results of P V C Sorption Study: Benzene, TCE and M-Xylene Concentrations in Soil Vapour Samples 19 Figure 2.5 Comparisons of O2, CO2 and C H 4 Concentrations Measured Using Landtec GA-90 and GC/TCD Analysis 21 Figure 2.6 Results of Light Gas Analysis Study: Light Gas Concentration Adjacent To Building During Dry Season Measured By Landtec GA-90 (Summer 1998) 21 Figure 2.7 Results of Light Gas Analysis Study: Light Gas Concentration Below Building Measured By Landtec GA-9023 Chapter 3. Figure 3.1 Predicted Spherical Helium Diffusion from Point and Finite Source 37 Figure 3.2 Possible Helium Concentration When CO2 replaces 0 2 39 Figure 3.3 Calibration of Mark 9822 helium detector 41 Figure 3.4 Site plan 41 Figure 3.5 Moisture Content Below Building 42 Figure 3.6 Sensitivity Analysis for Effective Diffusion Coefficient (Based on Equations 7 and 10) 45 Chapter 4. Figure 4.1 Simplified Vadose Zone Biogeochemical Model 51 Figure 4.2 Site Plan 54 Figure 4.3 Soil Moisture Content Below and Adjacent To Building (Approximate Total Porosity Values are Provided in Table 4.3) 55 Figure 4.4 Benzene Vapour Concentrations Below and adjacent To Building (MC = Moisture Content) 57 Figure 4.5 B T X Vapour and Light Gas Concentrations Below Building (MC = Moisture Content) 58 Figure 4.6 Oxygen Gas Concentrations Below and Adjacent to Building 58 Figure 4.7 Comparison of Field Degradation Rates to Published Data in DeVaull et al. (1997) (used with permission) 61 Figure 4.8 Model Domain and Boundary Conditions Used for 2-D Simulations 64 Figure 4.9 Comparison of Numerical and Analytical Comparison of Numerical and Analytical Solutions for Steady-State Conditions (input parameters: Cg(source) = 15 mg/L, 8w = 0.14, 0 = 0.36, H = 0.23, k w l = 1.2 hr"1, U = 5.6xl0"6 m/s, 0Ci =0.1 m, U = specific discharge, xi al = longitudinal dispersivity, Pe = Peclet number, all other symbols defined in text) 66 Figure 4.10 Comparison of Biodegradation Models for Uniform and Non-Uniform Soil Conditions Using 1-D Diffusion and Reaction Model (Steady-State Conditions, JJR = Instantaneous Reaction) 67 Figure 4.11 Moisture Content Profile Used in Numerical Simulations (see Table 4.3 for Porosity Values) 69 Figure 4.12 Predicted Vapour Concentrations Below Centre of Building for 2-D Diffusion With Reaction (Model No. 4) - wet (winter) conditions (Tl = 0.001, T d u s t = l ) 69 Figure 4.13 Comparison of model-predicted and measured vapour concentrations from Fig. 4 and 6 for 2-D diffusion and reaction (model no. 4) (h=0.001, T D U S T = 1.0) (approximate steady-state conditions at 3 months) 70 Figure 4.14 Effect of slab open area and tortuosity on predicted vapour concentrations below centre of building for 2-D diffusion with reaction (model no. 4) - wet (winter) conditions (approximately steady state conditions at 3 months) 71 Figure 4.15 1-D diffusion & reaction (Model No. 4) model sensitivity for (i) critical layer (0.88-0.92 m) moisture content (constant moderate biodegradation rate), (ii) biodegradation rate +/-60% of moderate rate (constant measured moisture content), & (iii) soil profile volumetric moisture constant +/-0.03 (constant biodegradation rate) (steady-state conditions) 73 Figure 4.16 Comparison of Measured and Model-Predicted Vapour Concentrations Below Building for 2-D Advection, Diffusion, and Reaction (model no. 4) (Moderate Biodegradation Rate, T|=0.001, tdust=1.0, k=10 darcy, AP=5 Pa. AL=1.4m, Initial Concentrations for Advection Case Are Steady-State Diffusion and Biodegradation Concentrations) 73 Chapter 5. Figure 5.1 Schematic of Concrete Slab and Soil Vapour and Flux Chamber Testing Locations 82 Figure 5.2 Cross-Section of Greenhouse (pressures measured December 1, 1997 during greenhouse depressurization to -40 Pa, Case 3 conditions) ..83 Figure 5.3 Schematic of Flux Chamber 85 Figure 5.4 Flux Chamber Soil Gas Flow Rates Through South Edge Crack 91 Figure 5.5 Flux Chamber Soil Gas Flow Rates Through Hair Line Cracks 92 Figure 5.6 Concrete Core Through Hair Line Crack 92 Figure 5.7 Subslab Oxygen below greenhouse - Fall 1997 (0 2 probe located near probe SG-BMW) 94 Figure 5.8 Subslab Oxygen Monitoring - Summer 1998. (0 2 Probe located near probe SG-BMW) 95 Figure 5.9 Results of Modflow Modeling a) Domain and Permeabilities, b) Flow Vectors and Equipotentials 97 Xll Chapter 6. Figure 6.1 Results of Sensitivity Analysis for Selected Parameters (solid lines are diffusion only models, dashed lines are for diffusion and advection model) 113 Figure 6.2 Results of Monte Carlo Probabilistic Analysis for Benzene Vapour Attenuation Rate (SVJJVI model, Table 6.4 inputs, triangular distributions, 5000 trials) 115 Figure 6.3 Measured Vapour Attenuation Ratio based on Hydrocarbon Intrusion Monitoring at Chatterton Site (no significant difference in indoor & outdoor concentrations obtained for natural cases) 121 Figure 6.4 Measured Benzene Vapour Profile Below Centre of Building at Chatterton Site (AP=depressurization, T=time from start of depressurization to sampling) 122 Figure 6.5 Comparison of Measured and Predicted Vapour Attenuation Ratio 123 Figure 6.6 Conceptual Framework for Use of Models for Soil Vapour to Indoor Air Pathway (Tier 1 = diffusion only; Tier 2 =, diffusion & advection) (Possible depth ranges: shallow<~l m, intermediate-1 to 5 m, deep>~5m) 126 Figure 6.7 Influence of Advective Transport on Vapour Attenuation Ratio (SVEVI model modified to use Flow to Perimeter Crack model, AP=5Pa, Table 6.2 parameters) 128 Chapter 7. Figure 7.1 Conceptual Simplification of Water Retention Curve for Purposes of Estimating Moisture Contents And Capillary Rise (9W,R. 9\V,FC, 9w,cz, ©w,s are the residual, field capacity, capillary zone and saturated water contents) 134 Figure 7.2 Sensitivity of Soil Gas Flow to Perimeter Crack Model (used in J&E model) to (a) soil-air permeability (ka), (b) depth to perimeter crack (Zcrack) and (c) crack ratio (r\). X c r a c k = perimeter crack length, Ab = subsurface foundation area) 136 Figure 7.3 Sensitivity of Vapour Attenuation Ratio (Benzene) to Soil Gas Flow Rate (Q) into Building using Perimeter Crack Model. Height = building height, Dry dust-filled concrete cracks with total porosity = 0.3, A C H = air exchanges per hour (other symbols previously defined Q = QSOiO 137 Figure 7.4 Sensitivity of Vapour Attenuation Ratio (Benzene) to Soil Gas Flow Rate (Q) using Perimeter Crack Model and Foundation Crack Ratio (n) (other symbols previously defined) 139 Figure 7.5 Sensitivity of Vapour Attenuation Ratio (Benzene) to Water-filled Porosity (Pa). Other symbols previously defined 140 Figure 7.6 Predicted Vapour Attenuation Ratio (benzene) for vapour Concentrations at Source and Indoor Air Using Johnson and Ettinger (1991) model. Figure adapted from Johnson et at., 1998. Dry dust-filled cracks: total porosity = 0.3; Moist dust-filled cracks: water-filled porosity = 0.1 and total porosity = 0.3 141 Xll l Figure 7.7a Comparison Between Measured and J&E Model-Predicted Vapour Attenuation Ratio (benzene). Upper and lower bound curves from Figure 7.6 are included. Dashed lines indicate that cxm is upper bound value. Symbols are best estimate cxp values 145 Figure 7.7b Comparison Between Measured and J&E Model-Predicted Vapour Attenuation Ratio (benzene). Upper and lower bound curves from Figure 7.6 are included. Dashed lines indicate that (Xm is upper bound 146 Chapter 8. Figure 8.1 Possible Conceptual Model For Soil Vapour Intrusion Into Building 159 Figure 8.2 Predicted Vapour Attenuation Between Benzene Vapour Concentrations at Source and Indoor Air Using Johnson And Ettinger (1991) Model. Figure Adapted From Johnson et al., 1998. Shaded Region Considered To Encompass Majority Of Sites. Height = building mixing height, Qsoil = Advective flow rate into building for coarse-grained soils, Dry dust-filled cracks, total porosity = 0.3; moist dust-filled cracks, water-filled porosity = 0.1, total porosity = 0.3; T| = crack ratio 172 Figure 8.3 Model-Predicted Vapour-Derived Indoor Air Concentration for (a) benzene, (b) TCE and (c) TVOC. The Soil Vapour Component of the Indoor Air Concentration Was Predicted Using The Johnson and Ettinger (1991) model 173 Figure 8.4 Predicted Incremental Lifetime Cancer Risks Using Soil Vapour-Derived Indoor Air Concentrations In Figure 8.3 For (A) Benzene And (B) TCE 175 xiv LIST OF APPENDICES Appendix I - Soil Vapour and Indoor Air Analysis 186 Appendix II - Equation for Vadose Zone Chemical Transport 247 Appendix III - Analysis of Moisture Content Data 250 Appendix IV - 2-D Finite Difference Numerical Model Code 258 Appendix V - Evaluation of Building Parameters 284 Appendix VI - Background Information on Chatterton Research Site and Photographs 287 Appendix VII - In situ Respiration Test Data 303 X V ACKNOWLEDGEMENTS This research project was sponsored by the Science Council of B.C. and Golder Associates Ltd. (Golder). Funding was provided by twelve industry, governmental and consulting groups: • Alberta Environmental Protection • American Petroleum Institute • B.C. Environment • B.C. Science Council • Canadian Association of Petroleum Producers • Canada Mortgage and Housing Corporation • Canadian Petroleum Products Institute • Chatterton Petrochemical • Dow Chemical • Golder Associates Ltd. • National Science and Engineering Research Council • Ontario Ministry of Environment In-kind support (field laboratory) was provided by Analytical Laboratory Services (ALS) of Vancouver, BC. This thesis would not have been possible without the encouragement and guidance of Reidar Zapf-Gilje. The support of the research committee, Loretta L i (Supervisor), Jim Atwater (Co-supervisor), Leslie Smith and Reidar Zapf-Gilje, is gratefully acknowledged. I would also like to thank Rob Dellatolla, David Jacubak and Scott Hannam for their assistance with the field program. This thesis is dedicated to Leanne, Meghan, Nathaniel, Jonathan and Benjamin. I love you all. Special thanks are given to our parents for their on-going support. Chapter 1 1 CHAPTER 1 INTRODUCTION This thesis presents the results of research on the validation of models used to evaluate the intrusion of subsurface volatile organic compounds (VOCs) into buildings, often referred to as the "soil vapour transport to indoor air pathway". Vapour intrusion into occupied buildings, when significant, has potential implications for human health arising from inhalation exposure to subsurface vapours. Models are commonly used to predict indoor air concentrations from vapour intrusion. The primary approach followed to investigate and validate models was to collect extensive field monitoring data on V O C vapour transport and intrusion, obtained at the former "Chatterton" petro-chemical plant site. Soil and groundwater at this site, located near Vancouver, B.C., Canada, is contaminated with benzene, toluene and m&p-xylene (BTX). This thesis is presented as a series of papers together with an introduction and conclusion. Supplementary information is provided in appendices. The introduction begins with a summary of key issues and why research on this topic is relevant and needed. Next, the research purpose and objectives are stated, and an overview of the research program is provided. The introduction concludes with a brief description of the thesis papers, which provides a "road-map" in terms of content, and interconnection between various aspects of the research program. KEY ISSUES AND RESEARCH RELEVANCY Numerous sites across North America have been impacted by VOCs. Some are urban brownfield sites, which increasingly are being developed for commercial or residential use. At numerous sites, groundwater contaminated with VOCs has migrated below existing structures including several well-documented sites where contaminated groundwater has migrated below hundreds of homes (see Chapter 7) (Johnson, 2002). Soil vapour intrusion into occupied buildings is considered a potential exposure pathway at most sites with V O C contamination. Therefore, this pathway is now routinely addressed as part of many human health risk assessments. Until recently, there were questions as to whether subsurface V O C intrusion into buildings was a significant exposure pathway (except in obvious cases where contamination was very near or in direct contact with the building). There is now a conclusive body of case study evidence indicating that soil vapour intrusion is a complete or operable exposure pathway at a significant number of sites. Vapour intrusion has been documented for a relatively wide range in site conditions in terms of contamination depth, soil properties, and house types. The evidence for significant soil vapour intrusion is strongest for sites contaminated with chlorinated solvents, and less so for petroleum hydrocarbon contaminated sites (Fitzpatrick and Fitzgerald, 1996; Chapter 7 this thesis). Human health risks due to the inhalation of vapours can, in some cases, greatly exceed those traditionally considered for risk-based groundwater clean-up programs. In part, this can be attributed to the relatively large volume of air a person breathes per day (about 20 m3/day Chapter 1 2 according to USEPA guidance) relative to the amount of groundwater ingested (about 2 L/day). When there are potentially unacceptable health risks, there are implications in terms of risk communication, response by those affected, and vapour intrusion mitigation. Predictive modeling is often needed to assess the vapour intrusion pathway. While indoor air quality testing may be an option (for a plume below an existing building), it is subject to background interference from the same VOCs that are found in soil and groundwater. As the number of published case study sites increase, the use of empirical data to assess vapour intrusion is emerging as an alternative but complementary approach to predictive modeling. Conceptual Site Model The conceptual site model (CSM) for the vapour intrusion pathway is complex (Figure 1.1). The source of vapour contamination may reside in groundwater as a dissolved plume, and/or be present in unsaturated zone deposits either as a non-aqueous phase liquid (NAPL) or sorbed-phase contaminant. Vadose processes that can affect vapour transport toward a building include chemical diffusion, biodegradation or biotransformation, and sorption. Geologic deposits tend to be heterogeneous leading to variable transport rates. Closer to the building and ground surface, advective processes can become significant, driven by pressure gradients created by factors that lead to depressurization of the building, or fluctuations in atmospheric pressure. Cracks or openings in the building foundation and subsurface utilities are potential pathways for vapour migration into the building. There is a large variation in building types and foundation characteristics. Once inside the building, subsurface vapours are mixed and diluted through ventilation, and may interact with building surfaces. The above processes are subject to temporal changes caused by seasonal weather patterns. L E G E N D : — D i f f u s i o n • Advection Infiltration (surface water) C O M P A R T M E N T S : 1. Salurated zone and capillary fringe. 2. Vadcuc lone. 3. Building envelope. 4. Building. I Building Envelope i t > Building (negatively pressurized) Vadoze Zone Capillary Fringe Saturated Zone Figure 1.1. Conceptual Site Model for Soil Vapour Intrusion into Buildings Chapter 1 3 In light of the above C S M , it follows that accurate quantitative prediction of vapour intrusion is difficult. What also becomes clear is that depending on the model and input parameters used, predicted indoor air concentrations can vary over many orders-of-magnitude. A key challenge for model use is to develop a framework that provides for sufficient conservatism in terms of protection of public health, but that also has sufficient power or ability to adequately discriminate between sites (i.e., a model that predicts risk in virtually all instances is meaningless). Predictive Models and Framework for Their Use Over the past decade several predictive models for the vapour intrusion exposure pathway have been developed. Typically, screening level models incorporating analytical or semi-analytical solutions for steady-state one-dimensional vapour transport are used for risk assessment purposes. Most models follow the heurestic framework put forth by Johnson and Ettinger (1991) and involve a compartmental solution to chemical diffusion in soil below a building, coupled with diffusion and advection through the building foundation (i.e., subsurface building envelope). More recent modeling developments include consideration of biodegradation, sorption and transient vapour transport (Sanders and Stern, 1994; Jeng et al., 1996; Johnson et al., 1998). In addition, there have been several models developed in the U K with a greater emphasis on building properties (Ferguson et al., 1995). Multi-dimensional numerical models have also been adapted to simulate this pathway. In Canada and the U.S., the Johnson and Ettinger (1991) model (or variants thereof) has become the model of choice. This is reflected in federal regulatory guidance (Canadian Council of Ministers of the Environment Canada Wide Standards for Petroleum Hydrocarbon Compounds (CWS-PHC) (2000), and the USEPA Draft Subsurface Vapour Intrusion (VI) Guidance (2002). Several provinces and states have also adopted the Johnson and Ettinger model for regulatory purposes (e.g., Atlantic provinces in Canada, Michigan, Massachusetts). There is an emerging recognition that models must be used in the context of an appropriate framework for site assessment and decision making. This, for example, is reflected in the USEPA VI Guidance, which provides a three-tiered framework for initial qualitative screening of sites that is followed by a secondary screening and site-specific assessment tier. The VI Guidance addresses the applicability of mathematical modeling relative to the C S M , and selection or measurement of input parameters needed for modeling. There is also recognition by practioners that at some sites, the use of more advanced site assessment techniques, relative to soil and groundwater assessments, are warranted including, for example, subslab soil vapour sampling, soil-air permeability testing, and monitoring of gases such as oxygen and carbon dioxide (to evaluate biodegradation). Model Validation Model validation is a process that begins with the development of a C S M . Sensitivity analysis can yield important insights into model characteristics and can help identify important model inputs. Ultimately, the "litmus test" is a model's ability to adequately predict real-world conditions. For this reason, field-based studies are a critical component of the model validation process, although theoretical and laboratory studies can be used in a complementary manner to better understand certain processes. Ideally, model validation should be a process where the Chapter 1 4 predictive capabilities of a model are tested using multiple independent cases over the expected range of conditions for model use. Model "validation" for this research is considered a process by which the predictive capabilities of a model are evaluated. It does not mean that a model is fully validated under all circumstances, which would be an unrealistic goal. Other pertinant issues include include (i) model verification, which is the determination of whether a model solves equations correctly, (ii) practicality of model use, and (iii) model acceptance by the regulatory and scientific community. Field-based studies that that can be used to validate models vary significantly in purpose, scope and quality. There are only a few detailed research studies specifically addressing intrusion of VOCs such as chlorinated solvents or petroleum hydrocarbons into buildings (Kliest, 1987; Hogdson, 1992; Fischer et al, 1996; Laubacher et al, 1997; Olson and Corsi, 2001). These studies have provide insight on the importance of specific processes (e.g., soil gas advection, biodegradation), and in a few cases have enabled the estimation of vapour attenuation ratio (the ratio of indoor air to vapour concentration near the contamination source) and comparisons to model predictions. There are a larger number of sites where groundwater, soil vapour (in some cases) and indoor air quality data has been obtained enabling the estimation of vapour attenuation ratios. Some of the available information has been summarized in technical papers (e.g., Fitzpatrick and Fitzgerald, 1996; Johnson et al, 2002). While useful correlations have been developed using this data, site characterization information has generally been limited therefore making this data less useful for model validation and evaluation of vapour intrusion processes. Several field studies have indicated vadose zone biodegradation can result in significant attenuation of hydrocarbon vapours (Ostendorf and Kampbell, 1991; Ririe and Sweeney, 1995, DeVaull et al., 1997). However, there are only limited field studies of the interaction between buildings and subsurface processes that affect biodegradation. There is a relatively extensive body of radon research including several studies involving in-depth testing of radon intrusion into homes (Nazaroff et al., 1997; Garbesi et al., 1993). The radon studies are significant since they have contributed to an enhanced understanding of mechanisms for soil gas intrusion into a building, including the influence of building underpressurization and advective soil gas flow, experimental methods (e.g., tracer tests), and models for prediction of soil gas intrusion. DATA GAPS AND QUESTIONS TO BE ANSWERED The following data gaps and questions to be answered were identified: 1. There are only limited field-based research studies of V O C intrusion. Only a few have been integrated studies involving an evaluation of the complete vapour intrusion pathway including characterization of contamination source, vadose zone fate and transport, migration through the building foundation, and mixing of vapours in indoor air. 2. There is only limited understanding of the interaction between buildings and subsurface processes affecting vapour transport such as soil gas advection, subslab oxygen transport and biodegradation. Chapter 1 5 3. The influence of seasonal factors on soil vapour concentrations and vapour intrusion is not well understood. Longer term studies evaluating seasonal effects have been limited. 4. While some case studies have been summarized, there have been few, if any, published studies involving a rigorous and comprehensive evaluation of the available case study database, estimation of vapour attenuation ratios, and comparison to model predictions. 5. The reliability of screening level models, such as the Johnson and Ettinger (1991) model, for prediction of soil vapour intrusion while improving, is still not well understood. 6. Protocols and guidance for certain test methods pertaining to the vapour intrusion pathway (e.g., soil vapour, flux chamber testing) are relatively limited. 7. There has been only limited use of multi-dimensional numerical computer models for the prediction of vapour intrusion into buildings. The above data gaps were addressed through a comprehensive, multi-year research program. The research is significant in terms of its breadth, extensive monitoring, new experimental methods, in-depth validation of screening models, and development of new numerical models. PURPOSE, OBJECTIVES AND SCOPE The primary purpose of the research program is to contribute to the validation of existing soil vapour intrusion models through a comprehensive field-based study and review of available case studies. Secondary goals are to develop new or adapted models and experimental methods that could be used to evaluate the vapour intrusion pathway. Specific objectives for the research program were: 1) Evaluate the reliability of screening level models (e.g., Johnson and Ettinger model) for prediction of soil vapour intrusion. 2) Obtain data that would enable evaluation of the interaction between a building and subsurface processes such as advection, diffusion and biodegradation. 3) Investigate the influence of building depressurization on soil vapour intrusion. 4) Obtain longer-term data to evaluate the influence of seasonal factors. 5) Develop new or adapted field methods to evaluate specific vapour intrusion processes and provide insight on method practicality. 6) Develop a multi-dimensional numerical model for this pathway, and compare the results to screening-level models. 7) Develop a framework for model use and application that could assist stakeholders in decision making. Chapter 1 6 The above objectives were supported through a comprehensive and integrated field study evaluating both vadose zone processes and soil vapour intrusion into an experimental greenhouse constructed at the Chatterton research site. BACKGROUND INFORMATION AND OVERVIEW OF RESEARCH PROGRAM The individual chapters of this thesis provide a description of research program site, experimental scope and methods. Relevant background information and an overview of the research program is provided below. Additional background information and photographs are provided in Appendix VI. The Chatterton site, located in Delta, B.C. , Canada, is contaminated with benzene, toluene and xylenes (BTX). The plant was in operation until 1992. The source of B T X contamination was spills and leaks from above-ground storage tanks, piping and valves. The specific area of the site chosen for the research program testing was located south of the main plant and source area for B T X releases, and down-gradient with respect to groundwater flow. Prior to the research program being initiated, the site had been subjected to a relatively comprehensive field investigation program. The main characteristics of the research site were as follows: • flat, grass covered area; • soil stratigraphy consisting of a surface sandy silt layer up to 0.3 m thick, underlain by relatively uniform fine to medium sand fill to 3.5 m depth, underlain by native silt; • variable water table with historical fluctuations between about 1.5 m and 3 m depth below ground surface; • well-defined non-aqueous phase liquid (NAPL) smear zone between about 1.5 m and 3 m depth; • total benzene, toluene and xylene (BTX) concentrations generally in the several hundred to thousands parts per million range in soil samples collected between about 1.5 m and 3 m depth; • approximate proportions by weight for the B T X compounds were 70 percent toluene, 20 percent benzene and 10 percent xylene, and; • No separate-phase product was observed in monitoring wells in this area of the site. The rationale for why this site was chosen for the study included site availability, relatively homogeneous soil properties, and no shallow contamination sources (i.e., above 1.5 m depth). The research program consisted of an extensive field testing and monitoring program over a two-year period complemented by laboratory testing, analysis and modeling. The site program began with a baseline investigation of soil properties (porosity, bulk density, field capacity, air-entry tension, grain size, organic carbon, nutrient levels, microbial enumeration) and testing of B T X concentrations in groundwater, soil, and soil vapour. Following construction of the Chapter 1 7 greenhouse, extensive vertical profiling of soil vapour concentrations both below and adjacent to the greenhouse was conducted on a regular basis. In addition, numerous other monitoring data were obtained (e.g., oxygen, carbon dioxide, methane, pressure, temperature, barometric pressure, wind speed, rainfall, soil moisture) and tests were conducted (in situ biological respiration, push-pull tracer tests to measure in situ diffusion coefficient, soil-air permeability). Vapour intrusion into the greenhouse was measured for six different cases, corresponding to different conditions for greenhouse depressurization and foundation crack area. Two cases were for a non-mechanically depressurized condition where the time-averaged pressure difference between indoor and outdoor air was approximately zero. For three cases, a fan was used to depressurize the greenhouse with tests conducted at -2.5 pascals (Pa), -10 Pa, and -30 Pa. For all six cases, indoor air concentrations (and other data) were obtained obtained enabling estimation of vapour intrusion rates. For selected cases, flux chambers were used to both measure soil gas flow and B T X flux into the greenhouse. Extensive quality control (QC) testing, and several "mini-studies" to evaluate QC issues were also performed. The project modeling scope included (i) evaluation of available screening-level models and development of new screening model for vapour intrusion (SVEVI model, Chapter 6) ; (ii) development of a two-dimensional finite difference model for hydrocarbon vapour diffusion, advection, dispersion and biodegradation (Chapter 4), and (iii) evaluation of analytical models and adaption of M O D F L O W for simulation of soil gas flow into a building (Chapter 5). STRUCTURE OF THIS THESIS The results of the research program are documented in seven technical papers and associated appendices. Technical papers allowed for rapid dissemination of research results to the consortium sponsoring this project, and was also important since vapour intrusion is a rapidly developing field. Significant contributions are made in the areas of field methods, new models for this pathway (conceptual and quantitative), comprehensive information on model validation, and a protocol for the application of vapour intrusion models. The purpose and scope of each paper is summarized below: Chapter 2 - Evaluation of soil vapour sampling and analysis techniques - The paper provides both a general discussion of soil vapour sampling and analysis issues, and describes and presents result for the program implemented at the Chatterton site. The paper includes several studies conducted to evaluate testing and quality control issues such as cross-contamination, soil vapour probe materials and use of field screening instruments for gas analysis. The appendix to this paper includes standard operating procedures, raw data for the B T X analyses, and results of the baseline field investigation not reported elsewhere. Chapter 3 - Measurement of in situ gas-phase diffusion coefficients - The purpose of this study was to use push pull tests, conducted using an inert tracer, to predict in situ diffusion coefficients, and to compare the results to empirical predictions. Theoretical equations used to calculate diffusion coefficients, field methods and quality control issues are discussed. The measured in situ diffusion coefficients were subsequently used as input for biodegradation modeling described below. Chapter 1 8 Chapter 4 - Evaluation of vadose zone biodegradation of BTX vapours - This paper focusses on field monitoring and modeling used to evaluate the biodegradation of B T X vapours below and adjacent to the greenhouse. A review of biodegradation studies and rate constants is initially provided. A two-dimensional finite difference numerical model is used to simulate diffusion, advection, sorption and biodegradation. Through comparisons of model-predicted to measured B T X vapour profiles, first-order biodegradation rate constants for B T X vapour are estimated, and insight into key factors (oxygen availability, advection) affecting biodegradation rates are obtained. The appendix to this paper includes additional discussion and modeling investigating the relationship between moisture content and biodegradation. Chapter 5 - Measurement of BTX Vapour Intrusion into an Experimental Building -This paper describes the testing program conducted to measure B T X vapour intrusion and soil gas flow into the greenhouse. A number of tracer and flux chamber experiments were designed and implemented. The measured rates were then compared to model predicted results. Both an analytical model and groundwater numerical model (MODFLOW), adapted to simulate soil gas flow, were used to predict soil gas flow rates into the greenhouse. Although the focus of this study were conditions near to the building foundation, the biodegradation of vapours, investigated in detail in Chapter 4, had an important effect on B T X vapour intrusion. Chapter 6 - Comparision, validation and use of models for predicting indoor air quality from soil and groundwater contamination - The purpose of this paper is to provide information and guidance on the use of available screening level models to predict indoor air quality from soil and groundwater contamination. An extensive review of model attributes followed by evaluation of model sensitivity and uncertainty for different models is initially provided. Measured V O C intrusion and soil gas flow rates (primarily those derived from the Chatterton study) are compared to model-predicted rates to assist in the validation process. The paper concludes with a protocol that provides guidance on the application of models within a tiered risk-based framework. Chapter 7 - Evaluation of the Johnson and Ettinger model for prediction of indoor air quality - This paper evaluates the use of the Johnson and Ettinger model for prediction of vapour intrusion and indoor air quality. The paper consists of three main parts: (i) detailed evaluation of measured vapour attenuation factors for chlorinated solvent and petroleum hydrocarbon sites, including several recent unpublished studies, (ii) estimation of input parameters for the Johnson and Ettinger model, and (iii) comparison of measured vapour attenuation rates to those predicted using the Johnson and Ettinger model. The paper concludes with a assessment of the reliability of the Johnson and Ettinger model for predictive purposes. Chapter 8 - The use of indoor air measurements to evaluate exposure and risk from subsurface VOCs - The purpose of this paper is to identify and evaluate issues for the use of indoor air quality (IAQ) data to evaluate exposure and risk from soil vapour intrusion. A key point is that there are numerous sources of background VOCs, which can be problematic when the objective is to quantify the sub-surface derived vapour component. To provide added perspective, the risks arising from subsurface VOCs, predicted using standard models equations for soil vapour transport and intrusion, are compared to those associated with background indoor sources. Also included is an extensive compilation of building depressurization values and Chapter 1 9 ventilation rates, which represent a valuable database of information for vapour intrusion modeling. The thesis ends with the key conclusions drawn from the technical papers and recommendations addressing remaining data gaps and research needs. REFERENCES American Society for Testing and Materials. 1995. Standard Guide for Risk-Based Corrective Action Applied at Petroleum Release Sites (E-1739-95). Canadian Council of Ministers of the Environment (CCME), 2000. Canada Wide Standards for Petroleum Hydrocarbon Compounds, June. DeVaull, G., R.A. Ettinger, J.P. Salanitro, and J. Gustafson. 1997. Benzene, toluene, ethylbenzene and xylenes degradation in vadose zone soils during vapour transport: first-order rate constants. Proceedings of 1997 Petroleum Hydrocarbons and Organic Chemicals in Ground Water, API/NGWA, Houston, Texas, November, 365-379. Ferguson, C.C., V.V. Krylov, and P.T. McGrath. 1995. Contamination of Indoor Air by Toxic Soil Vapours: a Screening Risk Assessment Model. Building and Environment. 30(3): 375-383. Fischer, M.L., A.J. Bentley, K.A. Dunkin, A.T. Hodgson, W.W. Nazaroff, R.G. Sextro, and J.M. Daisay. 1996. Factors affecting indoor air concentrations of volatile organic compounds .at a site of subsurface gasoline contamination.. Environ. Sci. Technol., 30: 2948-2957. Fitzpatrick, N.A. and J.J. Fitzgerald. 1996. An Evaluatoin of Vapour Intrusion into Buildings Through a Study of Field. Presented at the 11th Annual Conference on Contaminated Soils. University of Massachusetts (Amherst). October. Garbesi, K., R.G. Sextro, W.J. Fisk, M.P. Modera, and K.L. Revzan. 1993. Soil-Gas Entry into an Experimental Basement: Model Measurement Comparisons and Seasonal Effects. Environ. Sci. Technol. 27(3): 466-473. Hodgson, A.T., 1992. Soil-Gas Contamination and Entry of Volatile Organic Compounds into a House Near a Landfill, Journal of the Air and Waste Management Association, 29: 277-283, March. Jeng, C.Y, V.J. Kremesec, Jr., and H.S. Primack. 1996. Models of hydrocarbon vapour diffusion through soil and transport into buildings. Proceedings of 1996 Petroleum Hydrocarbon and Organic Chemicals in Ground Water, API/NGWA, Houston, Texas, November 13-15, 319-338. Johnson, P.C. and R. Ettinger. 1991. Heuristic Model for Predicting the Intrusion Rate of Contaminant Vapours into Buildings. Environ. Sci. Technol. 25 (8): 1445-1452. Johnson, P.C, W. Kemblowski, and R.L. Johnson. 1998. Assessing the Significance of Subsurface Contaminant Vapour Migration to Enclosed Spaces - site specific alternatives to generic estimates. API Publication. Johnson, P.C. 2002. Migration of Soil Gas Vapours to Indoor Air. Determining Vapour Attenuation Factors Using Screening-Level Model and Field Data from the CDOT MTL, API Bulletin 16, April. Jury, W.A., D. Russo, G. Streile, H. El Abd. 1990. Evaluation of volatilization by organic chemicals residing below the soil surface. Water Resources Research, 26(1): 13-20. Kliest, J., T. Fast, J.S.M. Bolij, H. van de Wiel and H. Bloemen. 1989. The Relationship Between Soil Contaminated with Volatile Organic Compounds and Indoor Air Pollution. Environment International, 15, 419-425. Laubacher, R.C., P. Bartholomae, P. Velasco, and H.J. Reisinger. 1997. An evaluation of the vapour profile in the vadose zone above a gasoline plume. Proceedings of 1997- Petroleum Hydrocarbon and Organic Chemicals in Ground Water, Houston, Texas, November, 396-409. Nazaroff, W.W., S.R. Lewis, S.M. Doyle, B.A. Moed, and A.V. Nero. 1987. "Experiments on Pollutant Transport from Soil into Residential Basements by Pressure Driven Airflow", Environmental Science and Technology, 21:459-466. Olson, D.A. and R. Corsi. 2001. Characterizing exposure to chemicals from soil vapour intrusion using two-compartment model. Atmospheric Environment 35:4201-4209. Ostendorf, D.W. and D. H. Kampbell. 1991. Biodegradation of hydrocarbon vapours in the unsaturated zone. Water Resources Research, 27 (4):453-462. Ririe, T. and B. Sweeney. 1995. API Conference. Fate and Transport of Volatile Hydrocarbons in the Vadose Zone. In Proceedings of the 1995 Petroleum Hydrocarbon and Organic Chemicals in Groundwater: Prevention, Detection and Remediation Conference, Houston, Texas, November 29 to December 1:529-542. Chapter 1 10 Sanders, P. and A. H. Stern. 1994. Calculation of Soil Cleanup Criteria for Carcinogenic Volatile Organic Compounds as Controlled by the Soil-To-Indoor Air Exposure Pathway. Environmental Toxicology and Chemistry, 13 (8): 1367-1373. USEPA, 2002. Draft Subsurface Vapour Intrusion Guidance. OWSER. Chapter 2 11 CHAPTER 2 EVALUATION OF SOIL GAS SAMPLING AND ANALYSIS TECHNIQUES AT A FORMER PETROCHEMICAL PLANT SITE This chapter was accepted for publication in Environmental Technology1 in October 2003. ABSTRACT Methods for soil gas sampling and analysis are evaluated as part of a research study on soil vapour intrusion into buildings, conducted at a former petro-chemical plant site ("Chatterton site"). The evaluation process was designed to provide information on reliability and selection of appropriate methods for soil gas sampling and analysis, and was based on literature review of data and methods, and experiments completed as part of the research study. The broader context of this work is that soil gas characterization is increasingly being used for input into risk assessment for contaminated sites, particularly when evaluating the potential intrusion of soil vapour into buildings. There are only a limited number of research studies and protocols addressing soil gas sampling and analysis. There is significant variability in soil gas probe design and sample collection and analysis methods used by practitioners. The experimental studies conducted to evaluate soil gas methods address the permeation or leakage of gases from Tedlar bags, time-dependent sorption of VOC-vapours onto probe surfaces and sampling devices, and analytical and quality control issues for light gas and V O C analyses. Through this work, common techniques for soil gas collection and analysis are described together with implications for data quality arising from the different methods used. Some of the potential pitfalls that can affect soil gas testing are identified, and recommendations and guidance for improved protocols are provided. KEY WORDS: soil vapour, diffusion, biodegradation, first-order decay, numerical model, B T X 1 Hers, I., Hannam, S. and Li, L. Chapter 2 12 INTRODUCTION This paper evaluates methods for soil gas sampling and analysis as part of a research study on soil vapour intrusion into buildings conducted at a former petro-chemical plant site ("Chatterton site") (Hers and Zapf-Gilje, 1998, Hers et al, 2000). The evaluation process was designed to provide information on reliability and selection of appropriate methods for soil gas sampling and analysis. To meet this objective, test data and methods cited in the literature were reviewed, and several experiments designed to address data gaps pertaining to soil gas techniques were completed. The analytical context for this work are both VOCs associated with contaminants, and light gases such as oxygen (O2), carbon dioxide (CO2) and methane (CH4), which are often used to assess biotransformation processes. The broader context for this work is the collection and analysis of soil gas samples is often required as part of human health risk assessments of contaminated sites or when implementing a risk-based corrective action approach (ASTM, 1995; Johnson et al., 1998). The early application of soil gas data was primarily to provide an indirect indication of potential volatile organic compound (VOC) contamination in soil or groundwater, and thus enhance the site characterization process (Marrin and Kerfoot, 1988). The more recent use of soil gas monitoring for risk assessment is different in purpose, and sampling and analysis requirements are generally more stringent since soil gas data is used to predict possible exposure concentrations in indoor or outdoor air. In contrast to soil and groundwater media, the collection and analysis of soil gas has received less in-depth scrutiny, and the few published protocols or guidance that are available (ASTM, 1992; San Diego County, 2002) are, in the authors opinion, not widely followed. There is significant variability in soil gas sampling probe design, sample collection and analysis methods and relatively few research studies where different soil gas sampling and analysis techniques have been evaluated. Based on the results of the experiments conducted as part of the Chatterton study and the literature review, practical recommendations and guidance for soil gas collection and analysis are provided. The benefits of this paper are new data and insights into important factors affecting soil gas quality, and improved protocols for the collection of high quality, representative data. B A C K G R O U N D INFORMATION REVIEW Soil gas testing can be divided into three steps consisting of probe installation, sample collection and handling, and analysis. The methods commonly used by practitioners and laboratories, and published studies evaluating soil gas characterization are reviewed below. Only active soil gas sampling methods are considered since passive methods do not provide for concentration data (i.e., information usually needed for risk assessment). Probe Design for Soil Gas Collection One of four basic probe types are used to collect soil gas: (i) driven probes, (ii) probes installed using direct-push methods, (iii) probes installed in drilled boreholes, and (iv) monitoring wells screened across the water table. Driven probes in their simplest form are hollow steel bars with an internal diameter typically ranging between 12.5 and 25 mm (often referred to as ground probes). The bars include a loosely fitting conical tip that is pushed a short Chapter 2 13 distance further into the formation using an inner rod, once the probe is driven to its desired depth. Several holes may also be drilled near the base of the probe to increase the area over which soil gas can be drawn into the probe. More sophisticated driven probes include a retractable drive tip connected to a system of threaded steel rods. The retractable tip is pulled back to expose a relatively short screened section allowing multiple soil gas samples to be collected from one probe hole, as the probe is driven. Screened sections of the probe are often constructed using stainless steel. In some cases, small diameter tubing constructed of stainless steel, Teflon (i.e., Polytetrafluoroethylene (PTFE)), or polyethylene is situated inside the outer rods to reduce the purge volume and minimize the potential for cross-contamination at rod joints. More recently, direct push technology has been used to install permanent soil gas implants. One proprietary design is a 6.25 to 12.5 mm diameter stainless steel screen that is 150 mm long. For some systems, implants are installed post-run after the desired depth is reached by lowering the implant down the hollow rods and attaching it to a detachable anchor drive point. The design of probes that are installed in boreholes varies widely and include probes constructed of steel (3.2 mm ID) (Schroedl and Kerfoot, 1990), Teflon (6.3 mm ID) (Yeates and Nielson, 1988) and P V C (30 mm ID, No. 30 slot screen) connected to narrow diameter vinyl tubing (Van Sciver, 1992). The authors are also aware of numerous instances where soil gas samples have been collected from conventional 50 mm diameter groundwater monitoring wells that have screens that extend above the water table. Sampling Collection and Handling Soil gas samples are typically collected by applying a vacuum to the distal end of the probe. Sample collection devices can include gas-tight syringes, glass cylinders, Tedlar bags, sorbent tubes, and Summa evacuated steel canisters. Tedlar is a proprietary plastic with properties similar to Teflon. The selection of a collection device is influenced by analytical requirements, discussed in a separate section below. Purge flow rates reported in the literature vary widely and range from 20 ml min 4 (Smith et al, 1996) to 2 to 3 1 min"1 (Rizvi and Fleischacker, 1992). Purge volumes also vary with some practitioners advocating only minimal purging corresponding to about 1 to 1 V2 probe volumes (Ririe et al, 1998). Cody (2003) reported purge volumes on the basis of a differential equation for the sequential and complete mixing of VOCs over each time step within the entire volume under consideration (probe and tubing). On the basis of this equation, the estimated concentration within the probe volume reaches 90 % of the input concentration after purging about three volumes. Christy and Spradlin (1992) evaluated the effect of purging by repeated purging, and then sampling in succession from a single probe and found that toluene concentrations decreased by a factor of 2 to 4 from the beginning to end of the purging process, which involved the removal of about 5 to 8 probe volumes. Ririe and Sweeney (1998) measured a slight decrease in CO2, and increase in O2 with increasing purge volume at a hydrocarbon-contaminated site. Chapter 2 14 After purging has been completed, the air-sampling pump is typically turned off and the vacuum is allowed to dissipate. In high permeability deposits this occurs almost instantaneously; however, in low permeability deposits this can take several minutes or longer (Christy and Spradlin, 1992). The vacuum generated during purging, and length of time for the vacuum to dissipate can provide a qualitative indication of the soil permeability and whether the probe is plugged. If a vacuum persists for a long period of time, it may be difficult to obtain a representative soil gas sample. When Tedlar bags are used to collect soil gas samples, a vacuum chamber can be used to avoid passing the gas sample through a pump. Since Surnma canisters are under sub-atmospheric pressure, an option for this method is to collect the soil gas sample by directly connecting the canister to the probe, thus avoiding drawing the gas through the sampling pump. There are only a few published studies of the effect of sample device and storage time on sample quality. Wang et al. (1996) found that for samples collected in Tedlar bags, between 16.5 and 25.6 % of the toluene, dichloroethylene (DCE) and trichloroethylene (TCE) vapours were lost over the first 24 hours after sampling. The loss was due to sorption and bag leakage. Andiro and Butler (1991) report that approximately 16 % of the methanol vapour concentration in 60-litre Tedlar bags was lost over a 24-hour period, and that the apparent vapour loss increased as the water vapour content increased. Smith et al. (1996) compared TCE vapour concentrations for samples collected with absorbent tubes (Supelco Inc. Carbotrap 300) and glass cylinders. They found that concentration differences were less than 5 %, and concluded that high quality samples could be obtained using both methods. Analytical Techniques Field Screening Methods Soil gas surveys commonly utilize field instruments such as photoionization detectors (PIDs) and multi-gas detectors for light gases (e.g., CO2, O2, CH4) as an initial screening step. Aelion et al. (1996) tested duplicate soil vapour samples using field instruments and by gas chromatographic (GC) methods and found a poor correlation (r2=0.45) between volatile hydrocarbons, measured using a platinum catalyst type sensor, and B T E X , measured by a GC/flame ionization detector (FID). The correlation between CO2, measured using a non-dispersive infrared analyzer and measured by a GC/thermal conductivity detector (TCD) was fair (r2=0.83). Possible reasons given for the poor correlation were variation in relative humidity, and a wide range in B T E X vapour concentrations. Robbins et al. (1990) demonstrated that several factors including flow rate restrictions, relative humidity, C 0 2 concentrations, and detector non-linearity have a significant effect on total PID and FID concentrations, when used to measure hydrocarbon vapours. Laboratory Analytical Techniques Selection of laboratory analytical techniques for quantification of individual VOCs will depend on project and data quality objectives. An initial investigation of site contamination may involve the collection of syringe or glass bomb samples followed by direct injection and quantification using a GC/PID or FID. For risk assessment studies, low detection limits and more rigorous quality control requirements often require that samples be collected using either sorbent tubes or Surnma stainless steel canisters and quantification by GC/MS (e.g., EPA Chapter 2 15 Methods TO-2, TO-14, TO-15, TO-17). In a comprehensive review, Harrington et al. (1998) point out that while air methods are similar to purge and trap methods for groundwater, air methods are less prescriptive, and there are also fundamental differences in terms of calibration standards, methods and materials used to clean sampling equipment and prevent cross-contamination, and the training required to ensure proper quality controls are followed. Sorbents commonly used consist of charcoal, polymeric and/or carbonaceous resins which are either thermally desorbed or extracted with a solvent such as carbon disulphide. Some tubes contain multiple compartments containing materials with differing sorptive properties (e.g., graphitized carbon black, carbon molecular sieve), designed to optimize the collection process. There are wide variations in sorbent properties. Since soil vapour typically has a relative humidity of close to 100 %, hydrophobic sorbents are preferred since sorbed water reduces the retention of VOCs, and because water vapour can affect the GC analysis (Harper, 1994). Polar V O C compounds can also partition into the water phase reducing recovery. Other issues for sorbent sampling include sorbent pore size and uniformity, possible reactions between the sorbent and adsorbed molecules, and slow breakdown of certain polymeric sorbents and release of aromatic hydrocarbons (Harper, 1994). Thermal desorption involves rapidly heating the sorbent to desorb the V O C , while passing an inert carrier gas through the tube. The VOCs are carried by the gas and concentrated on a smaller downstream trapr which usually is cryogenically cooled. For thermal desorption, the whole sample is analyzed at one time without the possibility of replicate analyses. In contrast, replicate analyses can performed on the extract for sorbent tubes that are chemically desorbed. While methods based on chemical extraction are typically not as sensitive as thermal desorption, reduced sensitivity may not be an issue for quantification of VOCs in soil vapour (i.e., as opposed to ambient air). Positive aspects of stainless steel canister sampling and GC/MS analysis include collection of relatively large "whole air" samples (e.g., typically six litres), low sensitivity and ability to take multiple aliquots from canister samples. Potential disadvantages for canister samples with elevated water vapour are problems caused during cryogenic focussing prior to analysis, although an alternate method of multiple focussing using non-cooled sorbent tubes can reduce problems associated with water vapour. EVALUATION OF EXPERIMENTAL METHODS Rationale The background information review identified several issues relating to sampling protocols and quality control. In part to address these issues, five soil gas studies were conducted during the Chatterton research program. These studies helped to evaluate method reliability and selection. The studies were an evaluation of time-dependent change in light gas concentrations in Tedlar bags ("Tedlar bag permeation study"), evaluation of B T X sorption on syringes ("syringe sorption study"), evaluation of V O C sorption onto P V C probes ("PVC sorption study"), comparison between light gas concentrations measured using a field detector and gas chromatographic (GC) analyses ("light gas analysis study"), and evaluation of analytical quality control for B T X vapour testing ("BTX analytical quality control study"). Chapter 2 16 Overview of Soil Gas Sampling and Analysis Methods at Chatterton Site The research program at the Chatterton site has included extensive monitoring of B T X vapour and light gas concentrations (Hers and Zapf-Gilje, 1998). Near-surface soil at the site consists of a thin surface layer up to 0.3 m thick consisting of sandy silt to fine sand with trace organic matter, underlain by dredged river sand with trace silt that extends to about 3.5 m depth. Soil gas samples were obtained from either 6.25 mm or 12 mm internal diameter hollow steel probes with 50 mm long perforated or slotted tips. Probes were equipped with dedicated miniature inert 3-way valves, one port had a 30 cm by 0.75 mm stainless steel tube connected (inserted in probe), while the other two ports had Teflon luer lock male connectors (sample collection and probe purging) (Figure 2.1). Probes were driven to the desired depths (between 0.15 m and 1.8 m depth). : [HI •< 3-Way Teflon Valve •<——— Swage-lock Fittings Rubber Stopper 50mm 0:75mm x 300mm Steel Tube K 12mm die. Steel Probe Detachable Steel Tip 24 36 48 T i m e (hours) Figure 2.1. Soil gas probes and Figure 2.2. Results of Tedlar bag sampling train used at Chatterton site. permeation study: 0 2 and N 2 The sampling procedure was as follows: (i) the probe was purged by removing 1 V2 probe-air volumes; (ii) the air-sampling pump was turned off and the vacuum was allowed to dissipate, and (iii) a soil gas sample was collected (Figure 2.1). The purging and sampling rate was 150 ml min"1. Vacuums dissipated within one second. Soil gas samples for B T X analyses were collected in a 5 ml syringe (described in detail below) or Supelco Inc. Carbotrap 300 sorbent tubes. Samples for light gas analyses were collected in 500 ml S K C Inc. Tedlar bags (Series 232). Samples were analyzed for B T X using a SRI 8610 GC/PID. Samples were analyzed for Chapter 2 17 light gases using a HP 5880 GC with TCD and FID detector, or using a Landtec Control Technologies GA-90 field portable gas detector. Samples for B T X analysis were analyzed within four hours of collection, while samples for light gas analysis were analyzed within 48 hours of collection. Tedlar Bag Permeation Study Methods The purpose of the Tedlar bag diffusion study was to measure the possible intrusion of atmospheric air into bags over time. The study was conducted by filling one-litre SKC Inc. Series 232 Tedlar bags to approximately 80 % capacity with ultra-high purity (UHP) helium. At five minutes and three, five, 24, 48 and 72 hours after filling, sub-samples were collected from the bag septum using a 5 ml VICI Series A-2 gas syringe and analyzed for nitrogen (N2) and O2 using a HP 5890 Gas Chromatograph equipped with thermal conductivity detector (TCD), flame ionization detector (FID) , and build-in gas sampling and back flush valves. Separate bags were used for tests at each time period {i.e., consecutive sub-samples were not taken from one bag). Three Tedlar bags were tested for each time period (Figure 2.2). Results Nitrogen and 0 2 concentrations remained relatively constant over the first five hours of the experiment, and were on average about 0.15 % for N2 and 0.05 % for 0 2 . After five hours, N2 and O2 concentrations began to increase with the rate of increase showing a slight exponential rise in time. The rate of increase was slightly greater for 0 2 than N2. At 72 hours, the average N2 and O2 concentrations were 0.63 and 0.55 %, respectively. The permeation of atmospheric air into the Tedlar bags tested was negligible over the first 24 hours but began to approach significant levels at 72 hours. The increase in O2 would be of concern when accurate light gas data are required to evaluate biodegradation conditions. The implication is that Tedlar bags used for light gas analyses should be analyzed within about 24 to 48 hours. Syringe Sorption Study Methods Five milliliter soil gas/headspace syringes constructed by SGE (Australia) (Part No. 5 M A X -HASV) with glass syringe barrels, Teflon coated gas-tight plunger, push-button valve, with Teflon luer lock male connector were used to collect soil gas samples from the Chatterton site. After prolonged use of the syringes in high B T X concentration areas, elevated concentrations started to be measured in blanks {i.e., syringes filled with helium). The time between sample collection and analysis, and blank collection and analysis was normally approximately the same. The elevated concentrations in blanks initiated an evaluation of the time-dependent desorption of benzene from the syringes (Figure 2.3). This was conducted by filling the syringe barrel with UHP helium and then waiting varying lengths of time before analysis using a SRI 8610 GC/PID. Chapter 2 18 Time (min) Figure 2.3. Results of syringe desorption study: Gas chromatograph benzene area counts. Results A progressive increase in benzene concentrations over time was measured indicating that significant adsorption of B T X had occurred, and that desorption was a time-dependent process (Figure 2.3). The implication is that blanks testing after only a few minutes would erroneously indicate a non-contaminated sampling device. PVC Sorption Study Methods A laboratory study was conducted to evaluate whether sorption onto the walls of a PVC probe has a significant effect on V O C vapour concentrations. P V C probes are often used to collect soil gas samples. The study was performed using three P V C probes, each installed in a sand-filled bucket. Initially, two probes were placed in buckets containing sand contaminated with benzene, m-xylene, and TCE ("Probes No. 1 and 2"). One control probe was placed in a non-contaminated bucket ("Probe No. 3"). Probe sorption was evaluated for the two soil gas probes in contact with V O C vapours. After four weeks, one of the probes was removed and placed in a non-contaminated environment and desorption was evaluated over a 6-week period. The progression from a contaminated to non-contaminated environment could conceivably approximate field conditions where vapour concentrations are reduced through remediation of the contamination source and/or rise in the water table (i.e., submergence of a residual contamination source). The set-up for probes No. 1 and 2 was as follows: (i) 60 ml each of benzene, m-xylene and TCE was added to 1.19 litres of water, (ii) the water and contaminants were briefly mixed with 7.92 litres of sand, (iii) the sand was placed in a clean 20-litre food-grade plastic pail (corresponding height in pail was 0.15 m), (iii) a 100 mm diameter hole was created in the sand in the pail, (iv) a 0.6 m long soil gas probe constructed of pre-washed environmental quality Chapter 2 19 50 mm diameter P V C with a 0.15 m long No. 20 slot screened section and bottom cap was placed in the hole, (iv) the space surrounding the probe and the remainder of the pail was filled with moist clean silica sand, (v) the probe top was completed with a cap and VA inch diameter polyethylene tube assembly, and (vi) the top of the plastic pail was covered with plastic. The water-filled and total soil porosity were approximately 0.15 and 0.40. Probe No. 3 was identical to Probes No. 1 and 2, except that no contaminants were added. One pail just containing non-contaminated silica sand was also set up. After four weeks, Probe No. 2 was pulled from the contaminated pail and inserted into the pail that just contained non-contaminated sand. Soil gas samples were collected from the probes on a weekly, or more frequent, basis. After purging approximately one probe volume, a sample was collected using either a 1 or 5 ml gas-tight syringe. Soil gas samples were analyzed using a SRI 8610 GC/PID. Results The vapour testing results indicate that during the first four weeks of the study, V O C concentrations for Probes #1 and #2 were highly elevated, and, as expected, concentrations in the blank (Probes #3) were near to the detection limit (Figure 2.4). The first four weeks of testing indicated a slight reduction in vapour concentrations for Probe #2, which may have been a result of contaminant mass depletion through purging and volatilization. 1.E+03 Benzene Probe #1 T C E Probe #1 m-Xylene Probe #1 Benzene Probe #2 T C E Probe #2 m-Xylene Probe #2 Benzene Probe #3 T C E Probe #3 • A • • m-Xylene Probe #3 S a m p l i n g Date Figure 2.4. Results of P V C sorption study: Benzene, TCE and m-Xylene concentrations in soil vapour samples. Vapour concentrations for Probe #2 decreased steadily after it was placed in the non-contaminated pail and after six weeks there was a 25-fold to 800-fold reduction in concentration relative to baseline concentrations. However, vapour concentrations after six weeks remained much higher than background concentrations in the non-contaminated pail indicating a significant mass of chemicals was undergoing time-dependent desorption. The concentration Chapter 2 20 decrease over time was greatest for benzene, which we infer to be due to a smaller mass of chemical initially sorbed to the P V C and a faster desorption rate. The study indicates that the mass of V O C vapours sorbing in or onto P V C can be significant, and that sorption and desorption processes could bias vapour testing results. Steel probes were used for the Chatterton study. Light Gas Analysis Study Methods To investigate the differences between field and laboratory methods, two samples taken consecutively from the same probe were analyzed. The first sample was tested in the field using a Landtec GA-90 gas detector, which utilizes a galvanic cell for O2 and infrared detector for CO2 and CH4. The Landtec was calibrated to O2 using atmospheric air (i.e., O2 equal to 20.9 %) and to C 0 2 and CH4 using a mixture of 15 % C 0 2 and 15 % CH4 (balance N 2 ) . The 0 2 calibration was periodically checked using 4 % 0 2 (balance N 2 ) . The Landtec 0 2 readings for this standard consistently ranged between 3.5 and 3.8 %. The second sample collected was analyzed using a HP 5880 GC with TCD and FID detector. Two capillary columns were used to separate the compounds within the GC/TCD-FID consisting of a (i) HP Plot molecular sieve 5A 30 m by 0.53 mm column or (ii) JW GSQ 30 m by 0.53 mm column. These columns were connected in-series using a gas valve. Results A relatively good comparison was obtained between the Landtec and GC results for CO2 (r2 = 0.95) (Figure 2.5). A good correlation was also obtained between the Landtec and G C results for O2 (r2=0.99). However, O2 concentrations by GC methodology were consistently higher than the Landtec concentrations. Possible reasons for this difference include a (i) downward bias in the Landtec concentrations due to non-linear response (Landtec was always calibrated to an O2 concentration of 20.9 %), (ii) upward bias in GC concentrations due to permeation of 0 2 into Tedlar bags, and (iii) slight increase in O2 concentrations during purging since the G C sample was always collected after the field sample. In total, these factors could potentially contribute to an 0 2 difference as high as 1.5 % and therefore do not fully account for the difference observed, which was up to 3 %. While there is no readily apparent reason for this difference, the lower Landtec O2 concentrations are consistent with the vadose zone model biogeochemical model for the Chatterton site as elaborated below. Chapter 2 21 (a) (b) (c) 20 15 u o 10 1 CM O 5 0 O c t o b e r & N o v e m b e r 1997 Data 1 = 1.204-Ix - 3.681 1 QRR4 5 — 5 10 15 02 GC/TCD (%) 20 O c t o b e r & N o v e m b e r 1997 Data y = 1.052 9x + 0. 5065 H' = 0.979f 3 6 9 12 C02 GC/TCD (%) O c t o b e r & N o v e m b e r 1997 25 3? 20 ? 15 10 Data y - 2 2 B08x - 0 -1491 R2 = 0.987 2 5 10 15 20 Methane GC/TCD (%) 25 Figure 2.5. Comparisons of 0 2 , C 0 2 and C H 4 Concentrations Measured Using Landtec GA-90 and GC/TCD Analysis A good correlation was also obtained between Landtec and GC results for C H 4 (r2=0.99). However, there was a significant upward bias in the Landtec CH4 concentrations. The Landtec GA-90 "methane" infrared detector detects hydrocarbons with a stretching (vibrational) frequency between 3.2 and 3.45 pLiri. Benzene and toluene also have stretching values within this range, and therefore, the elevated Landtec methane concentrations are likely due to benzene and toluene vapours. When B T X concentrations were high, the Landtec concentrations were corrected using the correlation in Figure 2.5 (Corrected CH4 = (Landtec C H 4 - 0.15) / 2.75)). The corrected CH4 concentrations are approximate values that are only accurate to a few %. The use of the Landtec field detector was considered appropriate for the Chatterton study. 16 12 u O u T O L U E N E 1 /IINERALIZAT ION C 7 H 8 + 9 0 2 - 7 C 0 2 + 4 H S 0 • C 0 2 c o n c e n t r a t i o n s • C H 4 c o n c e n t r a t i o n s — — L i n e a r ( C 0 2 c o n c e n t r a t i o n s ) y = - 0 . 8 . S 4 7 x + 1 7 . 8 4 9 R 2 = 0 . 9 9 8 4 • n nffa ran a 10 15 0 2 ( % ) 20 25 Figure 2.6. Results of light gas analysis study: Light gas concentration adjacent to building during dry season measured by Landtec GA-90 (summer 1998). Chapter 2 22 Consistency of Soil Gas with Biogeochemical Processes Comprehensive monitoring at the Chatterton site indicates an aerobic zone, where biodegradation of B T X vapours and oxidation of CH4 is occurring, and an anoxic or anaerobic zone (Hers and Zapf-Gilje, 1998; Hers et al., 2000). Under anoxic or anaerobic conditions, low O2 concentrations (less than 2 %) would be expected as indicated by monitoring at other sites (Jeng et al., 1996; Ririe and Sweeney, 1995; L i , 1995). Soil gas samples with Landtec O2 concentrations below 1 % were generally from the anoxic or anaerobic zone and therefore at levels consistent with inferred biogeochemical processes. The internal consistency of the light gas levels, relative to biodegradation processes at the Chatterton site, was checked by evaluating soil gas concentrations measured below a building (greenhouse) and an adjacent grass-covered area. Soil gas concentrations between 0.15 m and 1.5 m depth were measured at 0.15 m to 0.3 m intervals in both areas. Additional information on soil gas monitoring and biodegradation modeling are provided in Hers and Zapf-Gilje (1998) and Hers et al. (2000). Below the grass-covered area, the soil moisture contents were relatively low during the relatively dry summer season. As a result, atmospheric O2 readily diffused into soil resulting in aerobic biodegradation. A plot of the summer O2 versus C 0 2 concentrations indicated a linear relationship (r2=0.998), which is consistent with aerobic biodegradation of B T X (Figure 2.6). The slope of the CO2 and O2 relationship (0.83) is close to the stochiometric relationship for the complete aerobic mineralization of toluene, which states that 0.88 moles of CO2 is produced for each mole of O2 consumed. Toluene comprises over 80 % of the hydrocarbon mass present at the Chatterton site. Below the building, the biogeochemical model appears to be more complex since both C H 4 and CO2 concentrations are elevated when 0 2 concentrations are low, suggesting both aerobic and anaerobic processes are operational below the building (Figure 2.7). The concrete slab of the building is a partial barrier to atmospheric O2 diffusion into subsoil. As a result, anaerobic processes appear to be more significant below the building compared to adjacent to the building. In the presence of O2, CH4 will be oxidized generating CO2. Under anaerobic conditions, the generation of methane through methanogenesis could result, and if hydrogen gas were present, methane could be generated through CO2 reduction. The slope of the CO2 and O2 relationship decreases at low 0 2 and elevated C H 4 concentrations (Figure 2.7). The reduced slope is approximately consistant with the stoichiometry for methane oxidation where 0.5 moles of C 0 2 is produced for every mole of O2 consumed. Chapter 2 23 16 4 • C 0 2 Co • CH4 Co ncentrations ncentrations V Q 4 . "V • • • • • • BP a • IP D • < t : D °fi • HD ]flBPD 0 5 10 15 20 25 02(%) Figure 2.7. Results of light gas analysis study: Light gas concentration below building measured by Landtec GA-90. BTX Analytical Quality Control Study Methods Soil gas samples were analyzed for B T X vapours using a SRI 8610 GC/PID. The GC/PID capillary column was a DB-624 30 m by 0.53 mm column with a methylated phenoxy stationary phase (1.8 u,m thickness). The C G carrier gas was ultra-high purity (UHP) helium and the flow rate was 8 ml min"1. Syringes samples were directly injected into the GC. Sorbent tubes were initially desorbed onto a focusing tube using a thermal tube desorber (Dynatherm Analytical Instruments, Inc. Model 850), which in turn was desorbed and passed through the GC. The G C oven temperature ramp implemented for B T X analysis was a constant 50°C for two minutes followed by 5°C/minute increase to 85°C followed by a two minute hold at the upper temperature. Calibration standards consisted of 1 \xl of 100, 250, 500 and 1,000 ng ul"1 benzene, toluene and xylene (individual concentrations) in pentane. Since the boiling point of pentane is 36°C, the pentane based calibration standards would have readily volatilized in the injection port (at 50°C). Standards were directly injected in the GC using a 10 ui syringe while vapour samples were injected using a 5 ml syringe. A 0.2 ml vapour sample was the minimum volume that could be accurately injected into the GC (using 5 ml syringes) while a 5 ml sample was the maximum volume that could be run without excessive broadening of the chromatogram peaks. In part, the maximum sample volume is controlled by the relatively small volume of the GC column (about 6.6 ml). Replicate soil gas samples obtained from the same probe were analyzed for B T X to evaluate sampling and analysis variability. Replicates consisted of either two or three samples obtained within about 15 to 60 minutes. Prior to collecting each sample, the probe was purged. The B T X recovery was evaluated by injecting a known mass of 3-fluorotoluene (surrogate) into the sorbent tube (typically 5 (0.1 of 50 ng ui"1 3-fluorotoluene), and then measuring the response. Chapter 2 24 Results The replicate precision was assessed using the relative standard deviation (RSD) (standard deviation divided by mean), since at some locations more than two replicates were analyzed. The relative percent difference (RPD) is more commonly used, and when two samples are analyzed, is 1.41 X greater than the RSD. The recovery was calculated as a percentage of the initial injected mass (Table 2.1). The median RSD of 14 to 29 %, which corresponds to a RPD between 20 and 40 %, is greater than the regulatory acceptable levels for duplicate laboratory water analyses (RPD of about 20 %). However, for this study, the RSD incorporates the combined variability associated with laboratory analysis, and sampling and short-term variation in soil vapour concentrations since replicate samples were concurrently collected in separate syringes. The analytical precision (RSD) and bias (recovery) was considered acceptable for purposes of this research project Table 2.1: Summary of Quality Control Testing for BTX Analyses QC Measure Compound No. QC Samples Minimum Maximum Arithmetic Mean Median Precision (% RSD) Benzene 22 0.52 108 28 14 Precision (% RSD) Toluene 22 0.86 119 33 26 Precision (% RSD) M&p-Xylene 18 1.9 68 30 29 Recovery (%) 3-fluoro-toluene 26 19 138 80 85 DISCUSSION OF EXPERIMENTAL TECHNIQUES AND LESSONS LEARNED Experimental techniques, lessons learned and recommended practice for soil gas sampling and analysis, based on both the results of the experimental studies and broader literature review, are discussed below. Probe Design Site geology and depth requirements can dictate the type of probe installed. Driven probes have the potential advantage of less disturbance compared to methods involving drilling, and are generally less expensive than probes installed in boreholes. However, our experience with driven ground probes is that the side perforations in some instances become smeared with fine-grained soil during installation making it difficult to draw a soil gas sample. There is greater flexibility in the design of probes installed in boreholes. Factors that potentially affect the quality of soil gas samples include the probe material, internal area of probe in contact with the sample, the screened length, cross-sectional area of the screened or open portion of the probe, the probe volume and opportunity for leaks or short-circuiting from the surface. The probe material combined with the internal area of the probe in contact with the sample potentially has implications in terms of sorption of VOCs and possible time-dependent desorption of VOCs into a clean air stream. Chapter 2 25 There is little quantitative information on the best type of sample tubing to use. The results of the P V C sorption study suggest that V O C sorption into P V C can be significant. While further testing of other material types is required to evaluate relative sorption characteristics, P V C may not be an appropriate probe and tubing material for vapour sampling. Teflon is sometimes cited as the plastic of choice, but others indicate that Teflon is porous and a poor choice of tubing material for vapour sampling (Kreamer, 2001). Although Teflon has also been the material of choice for sampling containers and tubing used by environmental laboratories, it's selectivity in removing organics from samples and standards has increasingly been noted over the past few years as being problematic. This has resulted in the introduction and use new relatively inert tubing manufactured from PEEK, Tefzel (ETFE-polymer), fused-silica glass, and HDPE. However, most of these materials are not readily available for use in soil gas surveys. Based on currently available materials, Teflon or high density polyethylene (HDPE) appear to have reasonable sorption characteristics compared to other types of plastic. To obtain representative soil gas samples, the relative open area over the perforated or screened section of the probe should be maximized and sufficiently large to provide for unimpeded flow of soil gas into the probe. However, the probe volume and filter pack volume should be kept relatively small to minimize purge volumes and the surface area over which sorption could occur. Smaller probe volumes also reduce the potential for non-representative samples due to over-purging and disturbance of equilibrium soil vapour concentrations, and possible short-circuiting. Especially for deeper probes, consideration should be given to systems that employ an implant connected to smaller diameter tubing. Suggested probe diameter for most applications is 25 mm or less. Since an overly large filter pack may alter the local vapour concentrations, installation of small diameter probes in large diameter boreholes should be avoided. The screened length of the probe is important since usually the investigation objective is to measure concentrations representative of a discrete point in the unsaturated zone. Several studies have indicated that hydrocarbon vapour concentrations can change several orders-of-magnitude over short vertical distances (i.e., on the order of 0.3 m) (Fischer et al, 1996; Hers and Zapf-Gilje, 1998). For this reason, soil gas probes with relatively short screens (i.e., 300 mm or less) are recommended for most applications. Soil vapour sampling from groundwater monitoring wells is not recommended except to provide approximate screening information. Short-circuiting of atmospheric air to the probe can result between the probe and soil, and leakage of soil gas and/or atmospheric air can occur at probe joints. Prevention of short-circuiting and leaks is particularly important for low permeability soil deposits. The surface seal integrity can be tested by introducing a tracer gas (e.g., propane, butane) around the probe at the contact with the ground surface and then analyzing the collected soil gas samples for the tracer gas (Hartman, 2002). Sampling Collection and Analysis Critical factors affecting soil gas sample quality include sampling collection methodology, sample losses and cross-contamination. The purge volume and rate can affect soil vapour concentrations over time. Sample losses can occur as a result of contaminant condensation within the sampling device or tubing, solution into condensed water, sorption onto sampling Chapter 2 26 device materials, leakage out of the sampling container and chemical change. Cross-contamination can result from intrusion of atmospheric air into the container, incomplete decontamination of the sampling device or sampling train. Several studies have suggested that sorption to and leakage from Tedlar bags can result in relatively large V O C losses over short time periods. When used for light gases, the Chatterton research program permeation study suggests that Tedlar bags can provide acceptable results when analyzed within 24 to 48 hours. Longer holding times for Tedlar bags could result in oxygen concentrations with an upward bias. Sorption of B T X onto sampling equipment (Teflon coated gas-tight syringe barrel) can be significant as demonstrated by the syringe study, highlighting the importance of decontamination of sampling devices. Typically, soil vapour sampling protocols due not include probe development prior to purging and sampling (i.e., as conducted for groundwater sampling). It some cases, it may also be prudent to develop or "condition" probes prior to purging and sampling, particularly if the probe surface area is large or if drilling fluids (air or water) are used during the drilling process. We are aware of one project where soil vapour probes were installed in deep boreholes advanced using an air rotary drill rig, and where probe development, consisting of the removal of air introduced into the formation, was needed to obtain representative results. The purging volume and rate should be based on a number of factors including considerations relating to the mixing of VOCs in the probe, and the desired representative soil gas volume to be tested. While based on a complete sequential mixing equation upwards of three probe volumes should be purged (Cody, 2003), for narrow diameter tubing fewer purge volumes are likely needed to obtain a representative sample due to reduced mixing resulting from more of a "plug flow" phenomena. If local vapour concentrations adjacent to the probe tip are required, it is recommended that purge volumes be minimized (about 1 Vz volumes). It is recommended that the vacuum be measured using a manometer and that vacuums generated be allowed to fully dissipate before collecting samples. Relatively low flow sampling rates should generally be employed (i.e., on order of 100 to 200 ml min-1) to minimize short-circuiting. The fittings between tubing and pump should be tight. Consistency in sampling is important to obtain repeatable results. Sampling devices should be consistent with analytes and data quality objectives (e.g., Tedlar bags would generally be inappropriate for VOCs). Quality control procedures must be carefully adhered to including the use of low sorptive materials for sampling trains dedicated to individual probes, thorough decontamination procedures, and appropriate holding times. Tape should never be used for sampling train connections. It is recommended that syringe samples be analyzed within a few hours of collection. While there is little guidance in terms of appropriate sample storage procedures, a reasonable approach is to keep samples dark to avoid photooxidation losses and to prevent excessive changes in temperature relative to the in situ soil gas temperature. Analytical Techniques Field Screening Methods The use of hand-held detectors for measuring total hydrocarbon and light gas concentrations at contaminated sites has become common. While these types of detectors are potentially valuable for site screening, the limitations associated with these instruments, including non-Chapter 2 27 specificity to compounds of possible interest, and effect of environmental factors and sampling methods, should be clearly understood (e.g., Robbins et al, 1990). The light gas study demonstrated that while a relatively good correlation between the Landtec GA-90 field detector and GC methods was obtained for CO2, the Landtec C H 4 results appeared to be affected by non-specificity to CH4 and elevated B T X vapour concentrations. The use of field detectors such as the Landtec GA-90 is considered appropriate provided that instrument limitations are recognized and data is corrected. It is also noted that detector technology continues to evolve and, in the case of the Landtec, an optical sensor that is highly specific to methane (wavelength=3.322 fim) is now available. The light gas study also showed how an understanding of biogeochemical processes can be used to evaluate the quality of gas measurements. Oxygen concentrations near atmospheric levels within hydrocarbon source zones, or lack of an inversely proportional O2 versus C 0 2 relationship (for soil zones where aerobic biodegradation is occurring) may be indicative of short-circuiting of air to the probe or equipment problems. Laboratory Analytical Techniques There are a wide range of options for sampling containers and laboratory analytical methods for quantification of individual VOCs. Method selection will typically depend on the target VOCs, data quality objectives and cost considerations. For the Chatterton site research study, the collection of samples using syringes and sorbents, and analysis by GC/PID methodology was found to be effective. For risk assessment studies, it is important that quality control requirements and detection limits be carefully considered. The authors have reviewed data for relatively numerous projects where the soil gas data quality was suspect. Problems have included for sorbent tubes the selective breakthrough of certain compounds and poor sample recovery due to excessive moisture, for Summa canisters the improper cleaning of canisters and contaminant carry-over, and elevated levels of non-target compounds associated with poor probe design (i.e., use of tape) and cross-contamination from nearby above-ground use of petroleum products. Some measures that can be taken to improve data quality are discussed below. Detection Limits: For risk assessments, the measured vapour concentrations are often used to predict indoor air concentrations. Appropriate detection limits can be back-calculated using risk-based target indoor air concentrations combined with expected dilution factors between soil vapour and indoor air. Guidance on dilution factors can be found in US EPA (2002), Hers et al. (2002) and Johnson et al. (1998). Sorbent Selection: Should be based on V O C type, desired detection limit, and data quality objectives. Due to elevated humidity in soil vapour, some sorbents used for air sampling are not appropriate for soil vapour. For low level analysis of soil vapour, older sorbent materials such as coconut shell charcoal have been largely replaced by newer sorbents such as processed synthetic carbon, or molecular sieve materials. Sorbent Sampling Volume: Should be carefully determined through consideration of the expected V O C concentration and mass, the sorption capacity and required detection limits. For highly volatile compounds such as vinyl chloride the use of sorbent tubes can be challenging due to limited capacity of the sorbent to retain vinyl chloride. Field screening of V O C concentrations Chapter 2 28 using a PID can be used to guide the calculation of the sample volume. It is recommended that two tubes in series be analyzed to evaluate possible breakthrough, for at least of portion of the samples collected. Blanks: For all methods, it is recommended that the analytical laboratory demonstrate, for at least a portion of the sample containers, that the sampling device (canister, sorbent tube) is clean to below the required detection limits, prior to sampling and analysis. Method blank and spike samples should be analyzed, and depending on the project, transport blanks and background samples may also be warranted. A field method blank can be collected by drawing atmospheric air or inert gas through the sampling train and probe to be sampled, prior to installation. CONCLUSIONS Soil gas sampling and analysis was an important component of the Chatteron research study. Acquiring good quality, representative soil gas data requires careful attention to probe design, sampling containers and methods, and analytical methods. For this reason, several experiments were conducted to evaluate methods for soil gas testing. The broader context for the studies and recommendations provided in this paper is that there is significant variability in soil gas probe design and sample collection and analysis methods used by practitioners, and there is a need for improved protocols for soil gas sampling and analysis. The results of the research program experiments and our background review indicate that significant issues for soil gas testing sampling include probe design and possible short-circuiting of atmospheric air, sorption of VOCs onto probe materials, contamination of sampling containers, and possible gas leakage from Tedlar bags. Some of the key issues that need to be addressed for soil gas program design include the probe size, material and installation method, purging volume and rate, the sampling and analysis method (sorbent tube versus "whole air" canister samples), and quality control procedures and checks. Although not the focus of this paper, soil properties and environmental factors should also be considered in the design of a soil gas sampling program. Soil gas samples from shallow depths can be affected by infiltration of rain water, fluctuations in temperature and barometric pressure. In some cases, there can be significant concentration attenuation of hydrocarbon vapours due to fine-grained high moisture content layers (representing a diffusive barrier) and/or biodegradation of hydrocarbon vapours (Hers et al., 2000). As a result of concentration variability, the use of closely spaced vertical network of probes may be warranted. Chapter 2 29 REFERENCES Aelion, C M . , J.N. Shaw, R.P. Ray, M.A. Widdowson, and H.W. Reeves. 1996. Simplified Methods for Monitoring Petroleum-Contaminated Groundwater and Soil Vapour. J. of Soil Contamination. 5(3): 225-241. American Society for Testing and Materials (ASTM), 1992. Standard Guide for Soil Gas Monitoring in the Vadose Zone. D-5314-92. American Society for Testing and Materials (ASTM), 1995. Standard Guide for Risk-Based Corrective Action Applied at Petroleum Release Sites. E-1739-95. Andiro, J.M. and J.W. Butler. 1991. A study of the stability of methanol-fueled vehicle emissions in Tedlar bags. Environ. Sci. Technol. 25: 1644-1646. Christy, T.M. and S.C. Spradlin. 1992. The Use of Small Diameter Probing Equipment for Contaminated Site Investigation. In: Proc. of the Sixth National Outdoor Action Conference. May 11-13, Las Vegas, NV. 87-101. Cody, R. 2003. Soil Vapour Sampling and Analysis: Sources of Random and Systematic Error in the Characterization of Low Level Organohalide Sources. In: US EPA Seminar on Indoor Air Intrusion, January 14-15, 2003. Fairmont Hotel, Dallas, Tx. Fischer, M.L., A.J. Bentley, K.A. Dunkin, A.T. Hodgson, W.W. Nazaroff, R.G. Sextro, and J.M. Daisey. 1996. Factors Affecting Indoor Air Concentrations of Volatile Organic Compounds at a Site of Subsurface Gasoline Contamination. Environ. Sci. Technol., 30 (10): 2948-2957. Harper, M., 1994. Novel Sorbents for Sampling Organic Vapours. Analyst, January, Vol. 119. Harrington, D., Tekmar-Dohrman, Introduction to Air Toxics Analyses, Rev3.10-98 Tekmar-Dohrman (1998). Hartman, B., 2002. How to Collect Reliable Soil-Gas Data for Risk-Based Applications, Part 1: Active Soil-Gas Method. LUSTLine Bulletin, 42, October Hers, I. and R. Zapf-Gilje. 1998 Canadian consortium research project - field validation of soil gas transport to indoor air pathway. In: Proc. 1998. Petrol. Hydro, and Org. Chem. in Ground Water, API/NGWA, Houston, Tx., Nov. 11-13:251-266. Hers, I., J. Atwater, L. Li, and R. Zapf-Gilje. 2000. Evaluation of vadose zone biodegradation of BTX vapours. Journal of Contaminant Hydrology, 46: 233-264. Hers, I., D. Evans, R. Zapf-Gilje, and L. Li. 2002. Comparison, Validation and Use of Models for Predicting Indoor Air Quality from Soil and Groundwater Contamination. J. of Soil and Sediment Contamination. 11 (4): 491- : 527. Jeng, C.Y, V.J. Kremesec, Jr., and H.S. Primack. 1996. Models of Hydrocarbon Vapour Diffusion through Soil and Transport into Buildings. In: Proc. 1996 Petrol. Hydro, and Org. Chem. in Ground Water, API/NGWA, Houston, Tx., November 13-15, 319-338. Johnson, P.C., W. Kemblowski, and R.L. Johnson. 1998. Assessing the Significance of Subsurface Contaminant Vapour Migration to Enclosed Spaces - Site Specific Alternatives to Generic Estimates. API Publication 4674, December. Kreamer, D., 2001. Field Innovation Forum: Down the Rabbit Hole with Alice - Sucking Soil Gas All the Way. Ground Water Monitoring and Remediation. Fall. Li, D.X., 1995. Bioventing feasibility assessment and system design using subsurface oxygen sensors. J. of Air & Waste Management Association. 45: 762-769. Marrin, D.L. and H.B. Kerfoot. 1988. Soil-gas surveying techniques. Environ. Sci. Technol. 22(7): 740-745. Robbins, G.A., B.G. Deyo, M.R. Temple, J.D. Stuart, and M.J. Lacy. 1990. Soil-Gas Surveying for Subsurface Gasoline Contamination Using Total Organic Vapour Detection Instruments Part I. Theory and Laboratory Experimentation. GWMR. Summer. Ririe, T. and R. Sweeney. 1995. Fate and Transport of Volatile Hydrocarbons in the Vadose Zone. In: Proc. 1995 Petrol. Hydro, and Org. Chem. in Groundwater. API/NGWA, Houston, Tx., Nov. 29 to Dec. 1, 529-542. Ririe, T., R. Sweeney, S. Daughery, and P. Peuron. 1998. A Vapour Transport Model That is Consistent with Field and Laboratory Data. In: Proc. 1998 Petrol. Hydro, and Org. Chem. in Groundwater, API/NGWA, Houston, Tx., Nov. 11 to 13, 299-308. Ririe, T. and R. Sweeney. 1998. Sampling/Analysis Protocols for Vapour-Phase Investigations. Petroleum Environmental Research Forum Soil Vapour to Indoor Air Workshop, Feb. 6-7. Schroedl, S.R. and H.B. Kerfoot. 1990. Long-term Soil-Gas Monitoring of Underground Storage Tanks. In: Proc. 1990 Petrol. Hydro. & Org. Chem. in Ground Water, API/NGWA, Houston, Tx., Nov. 15-17. San Diego County, 2002. San Diego County Site Assessment Manual health/lwq/sam/pdf files/presentations/soilvapour guide.pdf. Chapter 2 30 Smith, J.A., C T . Chiou, J.A. Kammer, and D.E. Kile. 1990. Environ. Sci. Technol., 24: 676-683. Smith, J.A., A.M. Tisdale, and H.J. Cho. 1996. Quantification of Natural Vapour Fluxes of Trichloroethene in the Unsaturated Zone at Picatinny Arsenal, New Jersey. Environ. Sci. Technol. 30: 2243-2250. U.S. EPA, 2002. Draft Guidance for Evaluating the Vapour Intrusion to Indoor Air Pathway from Groundwater and Soils. December. Van Sciver, C. 1992. Monitoring a Methane Plume at a Sanitary Landfill Utilizing Cost Efficient Soil and Gas Techniques. Hazardous Materials Control. September/October. Wang, Y., T.S. Raihala, A.P. Jackman, and R. St. John. 1996. Use of Tedlar Bags in VOC Testing and Storage: Evidence of Significant VOC Losses., 30: 3115-3117. Yeates, G.L. and D.M. Nielsen. 1998. Design and Implementation of an Effective Soil Gas Monitoring Program for Four-dimensional Monitoring of Volatile Organics in the Subsurface. In: Proc. 1998 Petrol. Hydro, and Org. Chem. in Ground Water, API/NGWA, Houston, Tx., Nov. 14-16. Chapter 3 31 CHAPTER 3 MEASUREMENT OF IN SITU GAS-PHASE DIFFUSION COEFFICIENTS This chapter was published in Environmental Technology. 21: 631-640 (2000)2. ABSTRACT Vadose zone in situ, diffusion coefficients were measured at a former petro-chemical plant ("Chatterton" research site) using a push-pull test and helium tracer. The test is relatively simple to perform, utilizes relatively inexpensive materials and field equipment, and requires only a few hours to complete each series of tests. In situ diffusion coefficients are derived using an analytical solution for spherical diffusion from a point source, assuming instantaneous injection of the tracer. The results of push-pull tests conducted at the Chatterton research site indicate that measured effective diffusion coefficients in sand fill with moisture content between 5 and 12% (by weight) ranged from about 0.01 to 0.07 cm2sec"1. A good comparison was obtained between the measured gas-phase tortuosity factor and that predicted using a common empirical relationship with measured tortuosity factors consistently about twice the predicted values. While further comparisons need to be conducted for various moisture contents and different soil types, the results of this study suggest that the push-pull test is an effective tool for estimating diffusion coefficients and can be used to validate empirical relationships for diffusion coefficient. KEY WORDS: soil gas, diffusion coefficient, tracer, helium, vadose zone. 2 Hers, I., Zapf-Gilje, R., Li, L. and Atwater, J. Chapter 3 32 INTRODUCTION Gas-phase diffusion is an important vadose zone transport mechanism for volatile chemicals. Prediction of gas-phase diffusion is integral for human health risk assessment at contaminated sites, where estimation of potential exposure for the soil gas transport to indoor and/or outdoor air pathways is often required (ASTM, 1995). In addition, quantification of diffusion coefficients may be important for in situ remediation projects involving soil vapour extraction and bioventing, where diffusion from low permeability soil zones may represent a rate-limiting process. Diffusion occurs as a result of the movement of chemicals as influenced by their kinetic energy. In most cases, predictive relationships assume a Fickian model where diffusive transport through a porous medium is a function of the concentration gradient, temperature, fluid viscosity and tortuosity. Diffusive transport rates in air are several orders-of-magnitude higher than those in water, and therefore consideration of liquid-phase diffusive transport is generally only important when moisture contents are very close to saturation levels. The soil tortuosity is dependent on the porosity, moisture content, pore geometry, horizontal and vertical soil layering and other potential macroscopic features such as dessication cracking (Batterman, 1995). It is noted that in some cases, transport may not be adequately described by a Fickian model. These include surface-hindered or Knudson diffusion which may be important for fine-grained soils where molecular collisions with pore walls prevail over collisions between molecules (Lin, 1994) and when multi-component gas-phase concentrations are very high. In this case, gas-phase diffusion is intrinsically coupled with viscous gas-flow arising from varying partial pressures of the gas mixture (Thorstenson and Pollock, 1989). The effective diffusion coefficient is typically estimated using the chemical-specific diffusion coefficient in air and water and a factor to account for porous medium tortuosity. In many cases, relatively simple empirical correlations based on relationships between the tortuosity factor, moisture content and soil porosity are used (Millington and Quirk, 1961). The Millington and Quirk model is typically used when simulating diffusive vapour transport (Johnson and Ettinger, 1991; A S T M , 1995; Lowell and Ekland, 2004). In situ or laboratory diffusion coefficients are rarely measured since to date such tests have not been straightforward, and perhaps due to an implicit assumption that empirically derived diffusion coefficients are sufficiently accurate for their intended purpose. Higher accuracy estimates of in situ diffusion coefficient may be warranted for risk assessment where accurate estimates of chemical flux are desired, or when there is a need to separate diffusive transport from other transport processes (e.g., biodegradation). This paper presents in situ diffusion coefficients estimated using a relatively simple push-pull test using a non-reactive tracer (helium) (Johnson et al., 1998). Diffusion testing was conducted at the former "Chatterton" petro-chemical plant site located near Vancouver, B.C. , Canada, a site contaminated with benzene, toluene and xylene, and represents one component of a comprehensive research program investigating the soil gas transport to indoor air pathway. The contribution of this work is to provide a simple and inexpensive technique for in situ measurement of diffusion coefficients, and to provide a comparison between measured and predicted diffusion coefficients. Chapter 3 33 BACKGROUND Determination of diffusion coefficients has primarily been conducted using laboratory tests where steady-state chemical flux is measured in soil cores or packed soil columns. Field tests have generally involved evaluation of the transient variation in gas concentrations through sequential (in time) extraction of small quantities of gas, relative to the initial volume of tracer injected (Lai et al, 1976; Roltson and Brown, 1977; Jellick and Schable, 1986). For these studies, gas samples were obtained using hyperdermic type needles and analyzed using gas chromatographic methods. For the transient method, the diffusion coefficient is determined using a least squares or similar fitting technique that minimizes the difference in predicted and measured gas concentrations based on an analytical solution for transient diffusion. The primary advantage of the push-pull test is a decreased sensitivity to potential errors since a much larger gas sample is extracted and analyzed resulting in a diffusion coefficient that represents transport integrated over a spherical volume. The relative merits of transient and push-pull type tests are further described in a recent publication (Johnson et al, 1996). The results of several laboratory-based studies where diffusion coefficients and tortuosity factors (see definition below) were estimated and are compared to empirically-predicted values in Table 3.1. The rationale for comparing tortuosity factors, as opposed to effective diffusion coefficients, is that the tortuosity factor is theoretically a function on soil properties only, whereas the effective diffusion coefficient is a function of chemical and soil properties and therefore subject to somewhat greater variability. Empirical tortuosity factors (Tg) that have been proposed include: • Tg = (pgw/3/^ (Millington, R.J. and Quirk, J .M., 1961) (henceforth referred to as the "Millington and Quirk" relationship); • Tg = (f)g4/^n (Currie, 1970) ("Currie relationship"), and; • Tg = O.660g (Penman, 1940) ("Penman relationship"), where dg and 6 are the gas-phase and total porosity (L L ), respectively. The comparisons indicate that under laboratory conditions the common Millington and Quirk relationship generally underpredicts the tortuosity factor, and that the deviation between measured and predicted values increases with moisture content. At higher moisture contents, the difference between predicted and measured tortuosity factor may be as much as one order-of-magnitude. Under certain conditions it appears that the Penman relationship provides improved predictive capabilities. Few field studies have included comparisons of measured to predicted diffusivities or tortuosity factors using different empirical models (Lahvis and Baehr, 1996; Lai et al, 1976). One field study involved estimation of in situ diffusivity by model calibration with transient carbon dioxide production (Lahvis and Baehr, 1996). Comparison of in situ diffusivities through model calibration with those estimated using the Millington and Quirk relationship indicated that the empirical values were about three-times higher than the field-derived values for sand but two times less for clay (Lahvis and Baehr, 1996). In a second field study, a tortuosity factor of 1.66(f»g provided a reasonable approximation of measured field and laboratory values for small-scale oxygen diffusion (Lai etal, 1976). Chapter 3 34 Table 3.1. Comparison of measured and empirical gas-phase tortuosity factor from laboratory diffusion studies. Experi- Calculated Gas-Phase Tortuosity Factor Total Air-filled mental Penman Millington & Currie Soil Type Chemical Porosity Porosity Tortuosity (1940) Quirk (1961) (1970) Fine Silty Loam 1 Methane 0.40 0.040 0.0025 0.027 0.00014 0.000025 Pachappa Loam 2 Toluene 0.50 0.36 0.22 0.24 0.14 0.099 Pachappa Loam Freon 0.50 0.26 0.11 0.17 0.04 0.024 Pachappa Loam Freon 0.50 0.33 0.20 0.22 0.10 0.065 Pachappa Loam Freon 0.50 0.43 0.28 0.28 0.23 0.18 Fine Sand3 TCE 0.38 0.38 0.34/0.37 0.26 0.28 0.24 Fine Sand TCE 0.38 0.31 0.21/0.20 0.20 0.13 0.10 Fine Sand TCE 0.38 0.23 0.15/0.091 0.15 0.051 0.03 Fine Sand TCE 0.38 0.15 0.085/0:052 0.10 0.013 0.0061 Fine Sand TCE 0.38 0.08 0.032/0.034 0.051 0.0013 0.00038 Concrete Sand4 Hexane 0.26 0.26 0.18 0.17 0.17 0.13 Concrete Sand Benzene 0.26 0.26 0.17 0.18 0.17 0.14 Concrete Sand Iso-octane 0.26 0.26 0.17 0.17 0.17 0.13 Silt Loam 5 Toluene 0.48 0.44 0.23 0.29 0.28 0.23 Silt Loam Toluene 0.52 0.40 0.32 0.27 0.17 0.13 Silt Loam Toluene 0.51 0.33 0.24 0.22 0.10 0.06 Silt Loam TCE 0.40 0.38 0.31 0.25 0.25 0.21 Silt Loam TCE 0.39 0.26 0.17 0.17 0.07 0.05 Silt Loam TCE 0.39 0.20 0.11 0.13 0.03 . 0.02 References: 1 Johnson and Perrott (1991);2 Jin et al. (1994);3 Batterman et al. (1995); 4 Baehr and Bruell (1990);5 Arands et al. (1997) Note: For study 5, assumed free-air diffusion coefficients used for calculation were Dair toluene=0.078 cm2sec_1, Dair TCE=0.083 cm2sec Chapter 3 35 THEORY One-dimensional transient gas-phase diffusion (i.e., Fick's second law) of a non-reactive chemical in a homogenous porous media can be expressed as: 0 R 8- = — 8 3, dZ f D ^ V H J BC g dz (1) where, R is the retardation factor (dimensionless), Cg is the soil gas concentration (mol L'3), t is the time (T) and Dg and Dw are the gas- and aqueous-phase diffusion coefficients (L2T]) (Appendix I). Assuming local equilibrium partitioning between the solid (sorbed) and soil water, and soil water and soil gas phases, and linear isotherms, the retardation factor can be expressed as: R = \ + JLU.+ P>£± (2) 6GH' 6GH' where (L. is the water-filled porosity (L3L3), pfy is the bulk density of the soil (ML3), is the distribution coefficient (L3M~l) and W is Henry's Law Constant (dimensionless). In many cases, an effective gas-phase diffusion coefficient (Deff) is defined that includes both the gas-phase porosity and diffusion coefficient. The effective diffusion coefficient is dependent on the tortuosity factor, and free-air and free-water diffusion coefficients, as follows: Df = 6gDg = TgDair (3) ®w = ® w "777 = TwDwater (A) tl where tg and are factors to account for tortuosity (dimensionless) and Da;> and Dwater are the free-air and free-water diffusion coefficients (L2T!). The free-air diffusion coefficient for helium in air can be estimated using an equation for binary diffusion coefficient (Fuller et al, 1966): _10-3TL75[(MA+MB)/MAMB]V2 where DAB is the binary diffusion coefficient of gas A in gas B , T is temperature (K), P is pressure (atmospheres), MA and MB are molecular weights of A and B, and vA and VB are atomic diffusion volumes. Equations below exclude diffusion through soil water from the transport formulation since free-water diffusion coefficients are about four orders-of-magnitude less than free-air diffusion coefficients. Aqueous-phase diffusion should be included for moisture contents close to saturation levels. Analytical solutions for several initial and boundary conditions have been derived (Crank, 1956). The analytical solution for spherical gas-phase diffusion, assuming a point source of mass M0 (M), released at t = 0 in a porous medium with uniform and homogenous properties, is: Chapter 3 36 Cv{r,t) = M0/[eg+ew/H'+QbKd/H] „ 3/2 3/2n3/2 8 ; r f Dsorp AD t sorp (6) D eff D sorp eg+ejH'+PbKdiH' (7) where Cv(r,t) is the concentration at any distance r (L) from the point source, and Dsorp is the "sorption-corrected" effective diffusion coefficient (l}T~l). In the case where a small volume of tracer gas is injected at a point source, the measured fraction of the initial mass of tracer recovered (if) at time ts is set equal to the predicted fraction obtained by integrating equation 6 over a known spherical volume (corresponding to volume of gas withdrawn). This manipulation enables calculation of Dsorp using the following equations, which were obtained from Johnson et al. (1998): n = M{ts)/M0=(Cv)Vs/cv°V0 (8) n = erf\B X 2 B/ 2 • 4n - B (9) B =• ADsorph X (10) where M(ts) is the mass recovered, M„ is the mass injected, C v is the concentration in gas recovered, Vs is the volume of gas recovered, C° is the concentration in gas injected, and V0 is the volume of gas injected. The point source assumption is reasonably approximated for Vr/Vs < 0.1 and 7] < 0.1. Using equations 7 and 10, the effective diffusivity can be estimated provided that soil porosity and moisture content are known or can be approximated. The equation for one-dimensional gas-phase diffusion can also be solved analytically for an initial finite source consisting of radius R0, assuming that the injected tracer uniformly displaces soil gas (Johnson et al, 1998): Cv(r,t) = ^-\ C° V r D t sorp erf R0 + r + erf exp 4Dsorpt -exp AD t sorp J (11) Chapter 3 37 Diffusion characteristics of helium were evaluated for initial concentrations and volumes corresponding to the test conditions used for the push-pull test for both a point and finite source i.e., equations 6 and 11). The results, presented in Figure 3.1, indicate that predicted helium concentrations decrease rapidly with distance from source and time. The comparison of the point and finite source solutions indicates that the differences in concentrations were greatest for short time periods and distances from source. As time increased, the two solutions converged to almost identical concentrations. 1000000.0 100000.0 10000.0 > E a . a 6 c o o a Notes: Moisture Cbntent 1000.0 100.0 10.0 1.0 0.1 Injected (1 = 10%| Density=1. p,300 ug or OJ Porosity=0. g/cm3 1 litre) Bulk - H e -- • — H e -- A — H e -- • — H e - K — H e -- He • 4 min - point 10 min - point 20 min - point -100 min - point 500 min - point 4 min - finite 10 min - finite 20 min - finite • 100 min - finite 500 min - finite 0.2 0.4 0.6 0.8 Distance (m) 1.2 Figure 3.1. Predicted spherical helium diffusion from point and finite source. Methods In situ diffusion coefficients were measured at probes designed to collect soil gas samples using a previously published push-pull test protocol with several slight modifications (Johnson et al., 1998). Soil gas probes consisted of 13 mm I.D. hollow-steel probes with an inner 6 mm polyethylene tubing. Once driven to the desired depth, the conical probe tip was rammed a further 10 mm below the bottom of the probe using an inner rod creating a small air cavity. Chapter 3 38 Helium concentrations were measured using a field portable Mark Products Inc. Model 9822 detector. The Mark 9822 detector uses a thermal conductivity detector with a carbon graphite filter located upstream of the detector. According to the manufacturer, the filter was designed primarily to remove methane and heavier organic compounds. The detector utilizes the contrast in thermal conductivity between nitrogen (N2) and oxygen (O2), the primary components in air, and helium which has a much higher thermal conductivity. The thermal conductivities of several gases are provided in Table 3.2. Table 3.2. Thermal conductivity of selected gases. Gas Thermal Conductivity (cal °C cm"3 sec1))xlO 6 Air 60.34 C 0 2 37.61 CO 57.86 C2H6 47.94 He 352.1 H 2 433.92 0 2 61.58 N 2 60.34 Source: Thermal conductivities are from CRC Handbook of Chemistry and Physics, 66th Ed., CRC Press, 1985-86 Of potential concern is the possible interference of carbon dioxide (CO2) and other light gases with helium quantification. For example, if CO2 were to replace O2 in the gas mixture, helium concentrations would be negatively biased, assuming CO2 was not filtered out by the carbon graphite. The manufacturer indicates that no testing has been conducted to determine whether the filter would also remove CO2. Gas-phase adsorption is affected by several factors including properties of the chemical and adsorbent material. For a non-polar molecule such as C 0 2 , molecular attraction and adsorption likely occurs primarily through Van der Waals force. The possible effect of carbon dioxide on detector response was theoretically evaluated assuming that CO2 was not filtered out, and that CO2 replaced O2 in the air and helium mixture. Assuming a linear response relative to the thermal conductivities provided in Table 3.2, the hypothetical measured helium detector response is compared to the "true" response (i.e., for a gas mixture excluding CO2) in Figure 3.2. As shown, relatively significant deviations occur for helium concentrations below about 1%. Chapter 3 39 100 True Helium Cone. (%) Figure 3.2. Possible helium concentration when CO2 replaces O2. The effect of CO2 on helium response was further evaluated by comparing measured helium concentrations for a known standard, after calibrating the detector to air, to measured helium concentrations in the same standard after calibrating the detector using soil gas containing elevated CO2. The results, presented in Table 3.3, indicate that CC^may have affected the helium response for these tests. When calibrated using soil gas, the CO2 concentrations were higher, a result of a decreased detector response for baseline conditions. Table 3.3. Effect of soil gas containing CO2 on Mark 9822 helium detector response. Helium Standard Cone. (Air and He) (%) Helium Concentration: Calibrated to Soil Gas (%) Helium Concentration: Calibrated to Air (%) 1 1.9 0.76 2 3.7 2.3 5 6.6 5.3 20 24 23 Notes: 1) Standard comprised of helium and air. 2) Soil gas comprised of C 0 2 = 16.8%, C H 4 = 3.9%, 0 2 = 0.5%. Other potential gases (N2, hydrocarbons) were not quantified. As indicated below, soil gas near the probe tip was "purged" through injection of air. As a result, CO2 concentrations were likely low, and therefore no significant effect on helium detector Chapter 3 40 response was anticipated. Evaluation of the effect of CO2 on helium detector response should be conducted in cases where CO2 concentrations are elevated. The protocol used for the push-pull tests consisted of the following: 1. Purge soil gas in region of soil gas probe tip by injecting a minimum of 15 litres of air. 2. Collect one-litre soil gas sample in one-litre Tedlar™ bag (i.e., "blank") and measure helium concentration. Inject additional air if helium concentration is above instrument detection limit (0.01%). 3. Partially fill an evacuated one-litre Tedlar™ bag with industrial (i.e., welding) grade helium. The helium volume placed into bag should be 0.1 litre plus the volume of the probe. Inject helium into soil. 4. Immediately extract from the same probe a 1-litre soil gas sample into a bag and measure helium concentration. 5. Repeat Steps #1 through #4 except wait for periods of 2, 5, 10 and 15 minutes before withdrawing a soil gas sample. Purging of soil gas was conducted using an air flow rate of 0.5 to 1 litre min"1 while gas injection and extraction was conducted using a flow rate of 0.5 litre min"1. Gas injection and extraction was conducted using calibrated battery-powered air sampling pumps. Each round of tests at a particular depth took approximately two hours to complete. Before conducting the push-pull tests, the helium detector calibration was checked using standards comprised of air and helium mixtures prepared in Tedlar™ bags. As shown in Figure 3.3, a relatively good comparison was obtained between the measured concentrations and prepared standards. After the initial comprehensive calibration check, instrument response was periodically checked prior to use. RESULTS In situ diffusion coefficients were measured at five soil gas probes installed at varying depths below the approximate centre of a research greenhouse constructed at the Chatterton site, and at one nearby probe installed in an open lightly vegetated area (Figure 3.4). Soils at the site consist of a relatively uniform dredged river sand fill with less than 5 percent silt content. During testing, the depth to the water table was approximately 2.5 m below ground surface. Testing was conducted in August 1998 during sunny and warm weather (i.e., daytime high temperatures of 22 to 27°C). Chapter 3 41 BHSO BH97-20 Paved Road BH-17 ': Groundwater Collection Tank ~—... BH97-19 BH97-10 + • +" BH97-8x\ KSG-A I SITE A ^BH97-6 ' MW97-2 Concrete Foundation Research Greenhouse (Constructed in August 1997) A BH97-7 'MW97-1 BH97-31 LEGEND: Borehole or Monitoring BH97-•: T Well Location V. y : A BH97-30 4-BH97-2 £\. • •: Soil Gas Probe Location . 0/'••.•' .''.'5'. ;:'.;:'lO'. ' J ~>ui I SITE B ^BH97-4 BH-15 BH97-3 " v T V » r Figure 3.4. Site plan. Chapter 3 42 The measured mass recovered, effective diffusivity and gas-phase tortuosity factors are presented in Tables 3.4 and 3.5. Tortuosity factors were calculated using equation 3. The elapsed time between injection and extraction (i.e., t in equation 10) was the difference between the mid-points of the extraction and injection time. The moisture contents used to calculate the effective diffusivity were obtained from gravimetric moisture content determinations for several soil cores obtained within 2 m of the soil gas probe clusters. The approximate mean moisture content for each depth based on the data shown in Figure 3.5 was used in the calculations. Based on the range of measured moisture contents, the measurement scale for the diffusion coefficient is a sphere with radius on the order of 9 to 11 cm. Table 3.4. Measured mass required and beta parameter for push-pull tests. Time= 1.24 min Time= :3.24 min Time: =6.24 min Time= 11.2 min Time= 16.2 min Location Depth (m) P P P P T| P SG-A 0.60 0.46 1.08 0.24 0.59 0.08 0.25 0.046 0.16 0.031 0.13 SG-BC 0.3 0.30 0.71 0.15 0.40 0.09 0.26 0.029 0.12 0.023 0.10 SG-BC 0.45 0.50 1.18 0.32 0.76 0.24 0.59 0.17 0.44 0.073 0.23 SG-BC 0.58 0.57 1.38 0.46 1.08 0.23 0.57 0.15 0.40 0.086 0.26 SG-BC 0.75 0.96 4.15 0.61 1.51 0.46 1.08 0.30 0.71 0.21 0.52 SG-BC 0.9 0.48 1.13 0.30 0.71 0.16 0.42 0.11 0.31 - -Moisture Content (% wet wt.) 5 10 15 20 Soil Cores from - 2 m from centre of greenhouse Figure 3.5. Moisture content below building. Chapter 3 43 Table 3.5. Measured in situ effective diffusion coefficients based on push-pull tests.1 Measured Eff. Diffusivity - Varying Extraction Times Mean2 Mean Predicted Time= Time= Time= Time= Time= Effective Measured Tortuosity Factor4 Location Depth M C 1.24 min 3.24 min 6.24 min 11.2 min 16.2 min Diffusivity Tortuosity Factor (M&Q, '61) (m) (% wt.) (cm2sec-1) (cm2sec4) (cm2sec_1) (cm2sec1) (cm2sec_1) (cn^sec1) SG-A 0.60 6.9 0.24 0.075 0.053 0.063 0.056 0.049 0.055 0.17 0.069 SG-BC 0.3 5 0.27 0.12 0.081 0.064 0.078 0.063 0.071 0.22 0.10 SG-BC 0.45 8 0.22 0.067 0.040 0.027 0.020 0.026 0.028 0.085 0.053 SG-BC 0.58 10 0.19 0.054 0.027 0.026 0.021 0.022 0.024 0.072 0.031 SG-BC 0.75 12 0.16 0.017 0.018 0.013 0.011 0.010 0.013 0.039 0.016 SG-BC 0.9 7 0.24 0.072 0.043 0.038 0.028 - 0.037 0.11 0.067 1 Physical/chemical properties for Helium: D a j r = 0.33 cm2sec~', H (dimensionless) = 121; Kj = 0.002 2 Arithmetic mean based on results for time equal to 3.24, 6.24, 11.2 and 16.2 minutes 3 MC = moisture content, MC are approximate values based on data in Figure 3.5, total porosity = 0.356. 4 Predicted tortuosity factor based on Millington and Quirk relationship. Chapter 3 44 The measured effective diffusivities for each depth were relatively consistent for elapsed extraction times equal to or greater than 3.24 minutes. The effective diffusivity for the 1.24 minute test was generally higher than those for longer times possibly due to experimentally inaccuracies compounded by the short duration over which the test was conducted, and deviations from the point source assumption incorporated in the analytical solution to the gas diffusion equation. The test precision for extraction times equal to or greater than 3.24 minutes, as characterized by the relative standard deviation (RSD), was on the order of 20%. As expected, the effective diffusion coefficients decreased with increasing moisture. In addition, a good comparison was obtained between the measured gas-phase tortuosity factor and that predicted using the Millington and Quirk relationship with measured tortuosity factors consistently about twice the predicted values. SENSITIVITY ANALYSES A sensitivity analysis was conducted for key input parameters by individual varying inputs while keeping remaining parameters constant. The baseline values were rf equal to 0.1, ts equal to 500 seconds, 0g equal to 0.19, H' equal to 121, and Kd equal to 0.002. The input values were varied +/- 40% of the baseline values. The results, presented in Figure 3.6, indicate that the diffusion coefficient is most sensitive to time, followed by fraction of mass recovered. The results were relatively insensitive to gas-filled porosity for the input values chosen with a 40% variation in gas-filled porosity resulting in only a 25% variation in the effective diffusion coefficient. The relatively low sensitivity of effective diffusion coefficient to variation in gas-filled porosity for the push-pull test enables a reasonably accurate effective diffusion coefficient to be estimated even when direct moisture content measurements are not available (Johnson et al, 1998). CONCLUSIONS Vadose zone in situ diffusion coefficients were measured at the Chatterton research site using a push-pull test and helium tracer. The test is relatively simple to perform, utilizes relatively inexpensive materials and equipment, and requires only a few hours to complete each series of tests. Appropriate pre-cautions and care must be taken during the test to account for possible effects of soil gas on detector response (i.e., through purging or calibration), and gas volumes and time must be measured as accurately as possible. In situ diffusion coefficients were obtained using an analytical solution for spherical diffusion from a point source, assuming instantaneous injection of the tracer. The results of several push-pull tests conducted at the Chatterton research site indicate that measured effective diffusivity coefficients in sand ranged from about 0.01 to 0.07 cm2sec_ 1. A good comparison was obtained between the measured gas-phase tortuosity factor and that predicted using the Millington and Quirk relationship (i.e., zg = <pgim/$) with measured tortuosity factors consistently about twice the predicted values. The results indicate the Millington and Quirk relationship, which is commonly used to account for tortuosity, is appropriate over the range of moisture contents evaluated. Chapter 3 45 0.07 £ 0 06 8~ Q IS g 0 05 CM E 0.04 L 0.03 0.02 300 400 500 600 700 Elapsed Time (sec) <y o o 0.07 0.06 8^ c o 3 I* 2 8 0.05 |;o.o4 •2 0.03 |S 0.02 J 0.11 0.15 0.19 0.23 0.27 Gas-filled Porosity 0.06 0.08 0.1 0.12 0.14 Fraction Mass Recovered n 0.07 c . <u § 0.06 5 c ^ 005 O j/) tn CN £ g 004 $ •5 0.03. u i 0.02 70 95 120 145 170 Henry's Law Constant He (dimensionless) Figure 3:6. Sensitivity analysis for effective diffusion coefficient (based on equations 7 and 10). Measurement of in situ diffusion coefficient may be warranted in cases where accurate estimates of diffusion coefficient are required for risk assessment purposes, or when evaluating vadose zone natural attenuation (e.g., biodegradation rates). In this context, they represent a valuable component of the research currently being conducted to evaluate the soil gas transport to air pathway at the Chatterton site. While further comparisons need to be conducted for various moisture contents and different soil types, the results of this study suggest that use of the Millington and Quirk empirical relationship may be reasonable in many situations, in light of the experimental accuracy of the push-pull test and relative effect of other factors affecting V O C fate and transport. Chapter 3 46 REFERENCES American Society for Testing and Materials (ASTM). 1995. Standard Guide for Risk-Based Corrective Action Applied at Petroleum Release Sites, E1739-95. Batterman, S. 1995. Hydrocarbon vapour transport in low moisture soils. Environmental Science and Technology, 29(1): 171-180. Crank, J. 1956. The mathematics of diffusion. Oxford Clarendon Press. Currie, J.A. 1970. Sorption and transport processes. SCI Monogr. 27: 152-171. Fuller, E.N., P.D. Schettler, and J.C. Giddings. 1966. Ind. Eng. Chem. 58(5): 18. Jellick, G.J. and R.R. Schabel. 1986. Evaluation of a field method for determining the gas diffusion coefficient in soils. Soil Sci. Soc. Am. J, Vol. 50. Johnson, P.C., C. Bruce, R.L. Johnson, and M.W. Kemblowski. 1998. In situ measurement of effective vapour-phase porous media diffusion coefficients. Accepted for Publication, Environ. Sci Technol. Lahvis, M.A. and A.L. Baehr. 1996. Estimation of rates of aerobic hydrocarbon biodegradation by simulation of gas transport in the unsaturated zone. Water Resources Research. 32(7): 2231-2249, July. Lai, S., J.M. Tiedje, and E. Erickson. 1976. In situ measurements of gas diffusion coefficients in soils. Soil. Sci. Soc. Amer. J,. 40. Lin, T. 1994. Transport and sorption of volatile organic compounds and water vapour within dry soil drains. Environmental Science and Technology. 28: 322-330. Lowell, P.S. and B.E. Eklund. 2004. VOC Emission Fluxes as a Function of Lateral Distance from the Source. Environmental Progress, 25 (1): 52-58. Millington, R.J. and J.M. Quirk. 1961. Permeability of porous solids. Trans. Farady Soc:, 1200-1207. Penman, H.L. 1940. Gas and vapour movements in the soil. I. The diffusion of vapours through porous solids. J. Agric. Sci. (Cambridge), 30:437-462. Roltson, D.E. and B.D. Brown. 1977. Measurement of soil gaseous diffusion coefficients by a transient-site methods with time-dependent surface condition. Soil Sc. Soc. Am. J. 41. Thorstenson, D.C., and D.W. Pollock. 1989. Gas transport in unsaturated porous media: The adequacy of Fick's Law. Review of Geophysics. 27(1), February. Chapter 4 47 CHAPTER 4 EVALUATION OF VADOSE ZONE BIODEGRADATION OF BTX VAPOURS This chapter was published in J. of Contaminant Hydro geology. 46:233-264. 2000.3 ABSTRACT Soil vapour transport to indoor air is an important potential exposure pathway at many sites impacted by subsurface volatile organic compounds (VOCs). The inclusion of biodegradation in vadose zone transport models for benzene, toluene and xylene (BTX) and fuel hydrocarbons has been proposed; however, there is still significant uncertainty regarding biodegradation rates, and the local effects of buildings or ground surface cover on fate and transport processes. The objective of this study was to evaluate biodegradation processes through comprehensive monitoring at a site contaminated with B T X and model simulation. Study methods included extensive vertical profiling of B T X vapour and light gas (oxygen and carbon dioxide) concentrations and moisture content, and semi-continuous monitoring of oxygen and pressure below a building floor slab. Significant vadose zone biodegradation over a relatively small depth interval was observed. Based on the observed soil vapour profile, first-order biodegradation rates were estimated by fitting an analytical solution for diffusion and biodecay to the data. Degradation rates were found to compare well to other reported laboratory and field data. A two-dimensional numerical model incorporating vapour-phase diffusion, advection, sorption and biodegradation was used to simulate the effect of a building floor slab on transport processes. Model results demonstrate the sensitivity of vapour-phase B T X and oxygen transport to partial barriers to diffusion (e.g., building foundation) and highlight the importance of using a model that ties biodecay to oxygen availability. In addition, depressurization within a building and advective transport are shown to have a potentially significant effect on B T X fate in soil below the building. KEY WORDS: soil vapour, diffusion, biodegradation, first-order decay, numerical model, B T X 3 Hers, I., Atwater, J., Li, L. and Zapf-Gilje, R. Chapter 4 48 INTRODUCTION Soil vapour transport to indoor air is an important exposure pathway at many sites impacted by volatile organic compounds (VOCs). Relatively simple screening-level models are often used to quantify potential exposure and risk (ASTM, 1995); however, there is significant uncertainty surrounding processes and factors affecting this pathway, and the accuracy of models used. To address this limitation, a comprehensive field-based research program is being conducted at a former petro-chemical plant site impacted by benzene, toluene and m&p-xylene (BTX) releases, located near Vancouver, B.C. ("Chatterton" research site). As shown in this paper, monitoring at the Chatterton site indicates that intrinsic biodegradation is an important vadose zone process. Evidence for vadose zone biodegradation of B T X or fuel hydrocarbons has been seen at a limited number of field sites (e.g., Ostendorf and Kampbell, 1991; Ririe and Sweeney, 1995; Franzmann et al, 1999). Several analytical models for hydrocarbon vapour transport that include biodegradation have been proposed. These include a model incorporating vapour-phase and aqueous-phase diffusion and aqueous-phase advection, subject to sorption and first-order decay (Jury et al, 1990), and a model for steady-state vapour-phase diffusion, subject to first-order decay (Johnson et al, 1988; DeVaull et al, 2002). Incorporating biodecay in models simulating soil vapour transport to indoor air, in some instances, reduces predicted exposure concentrations by several orders-of-magnitude. Therefore, sound evaluation of biodegradation potential is essential. There remains considerable uncertainty in terms of vadose zone biodegradation rates for organic chemicals. There are only a few comprehensive field-based assessments of biodegradation potential for petroleum hydrocarbons. Further, there has been little direct evaluation of the local effect of a building or ground surface cover on soil gas transport and rates of hydrocarbon vapour biodegradation. The focus of this work is to evaluate B T X biodegradation processes and rates through comprehensive analysis of field data combined with model simulation, in the context of the exposure pathway of soil vapour transport to indoor air. The objective is to gain new insight on biodegradation processes and kinetics, and to contribute to improved modeling methods. The paper is presented in three parts; (i) overview of intrinsic biodegradation processes, kinetics and case studies, (ii) presentation of monitoring data from the Chatterton site and derivation of field biodegradation rates, and (iii) description of a numerical model developed for this project and comparison of model-predicted to measured soil vapour transport. INTRINSIC BIODEGRADATION Processes and Effect of Building Diffusion, sorption and biodegradation, for non-recalcitrant VOCs, are generally thought to have the most significant effect on V O C fate and transport within the vadose zone. Biodegradation is an important natural attenuation mechanism since it is the only process whereby there is a reduction in total hydrocarbon mass. More than 200 species of bacteria, yeast and fungi capable of degrading petroleum hydrocarbons have been identified, with Pseudomonas spp. and Corynebacterium spp. thought to be two major bacterial agents (Fan and Krishnamurthy, 1995). Aerobic biodegradation of petroleum hydrocarbons will occur in the vadose zone providing there is sufficient O2, indigenous microbes that produce enzymes capable Chapter 4 49 of degrading the compound of interest, soil moisture, nutrients, and appropriate pH, temperature and salinity conditions, and no inhibiting materials. DeVaull (1997) suggests that O2 levels above 4 percent (as gas) and 1 to 2 mg/L nitrate (as nitrogen in pore water) will sustain aerobic biodegradation. Anoxic and/or anaerobic biodegradation can also potentially occur in the vadose zone depending on types of electron acceptors present, pH conditions and oxidation-reduction potential. Conceptually, there are several ways in which a building (or other similar surface barrier) could affect intrinsic biodegradation. For aerobic degradation of VOCs, the ability of O2 to diffuse, or be transported through advection to below the building is of critical importance. Oxygen replenishment will be a function of the diffusivity and permeability of near surface soils and the subsurface building structure, and gradients driving these processes. Advective transport of soil gas through shallow soil can occur as a result of changes in atmospheric pressure and temperature gradients (Massman and Farrier, 1992). A building or other low permeability structure will also eliminate surface water infiltration below the structure footprint and potentially promote long-term drying of the soil. While there is little direct study of the effect of soil moisture on intrinsic biodegradation, bioventing studies suggest that biodegradation rates are significantly reduced for low moisture contents (Zwick et al., 1995). A laboratory study indicated that biodegradation of toluene by Pseudomonas putida was affected by matric potential with about 35 to 60 % reduction in cell growth and first-order biodegradation rate when the matric suction was increased from 0 to 1.5 MPa (Holden et al., 1997). The permanent wilting point is the soil moisture content at a matric suction of about 1.5 MPa. Biodegradation Models and Kinetics Biodegradation kinetics for petroleum hydrocarbons can be simulated using several models including (i) first-order or zero-order decay models, (ii) models based on Monod-kinetics, and (iii) instantaneous reaction models. A first-order (i.e., exponential) decay model can be represented as follows: where S is the substrate concentration (M L3), k is the first-order rate constant (V), and tin. is the half-life (J). A first-order decay constant assumes that both the O2 and hydrocarbon-degrading microbes are available in excess, and that only the hydrocarbon substrate is rate limiting. A zero-order model assumes a constant biodegradation rate, and that the O2 and/or hydrocarbon-degrading microbes are rate-limiting. In some cases, kinetics based on equations proposed by Monod (1949) provide for a more accurate representation of biodegradation. The Monod model describes the growth of a pure culture of microorganisms suspended in liquid through the utilization of a single rate-limiting substrate (Bekins et al., 1998). One simplified expression ("Monod-no-growth") given by Simpkins and Alexander (1984) is: dS/dt = kS tm = In (2)/k (1) dS/dt = kj S/(KS + S) kl = Umax B /Y (2) Chapter 4 50 where ki is the substrate utilization rate (ML3T'), Ks is the half-saturation constant or hydrocarbon concentration at which the biodegradation rate is half its maximum value (ML3), Umax is the maximum specific growth rate of the biomass (T1), B is the biomass concentration (ML3), and Y is the biomass yield (M biomass M'1 substrate). When hydrocarbon substrate concentrations are high (i.e., S » Ks), equation 2 approximates a zero-order expression. Conversely, when hydrocarbon substrate concentrations are low (i.e,. S«Ks), equation 2 approximates a first-order expression. A dual-Monod equation can be used to represent substrate utilization as a function of both biomass concentration and electron acceptor concentration (Kissel, 1985). The dual-Monod equation can be expressed as: dS/dt = k!S/(Ks + S) (A/(KA + A)) (3) where A is the concentration of the electron acceptor (M L3) and KA is the electron acceptor concentration at which the utilization rate is half the maximum (M L3). Coupled transport equations for the primary substrate and electron acceptor can be linked through the Monod expressions and stochiometric ratio of the electron acceptor to substrate consumed (MacQuarrie et al., 1990). A practical disadvantage of the Monod-kinetics model is that input parameters can be difficult to obtain. The instantaneous-reaction model couples hydrocarbon vapour transport with O2 or alternative electron acceptor concentrations. Two factors that control the biodegradation rate for the instantaneous-reaction model are: (i) the electron acceptor utilization factor for hydrocarbon mineralization (estimated from stochiometric relationships), and (ii) the electron acceptor transport rate from source areas and mixing with the hydrocarbon-impacted soil vapour. Additional considerations for biodegradation modeling are multi-component effects and appropriate use of reaction models where there is competing utilization of O2. Evaluation of Biodegradation Potential Multiple lines of evidence should be considered when evaluating vadose zone intrinsic biodegradation, as typically conducted for dissolved hydrocarbon fate in groundwater. Lines of evidence include: (i) field measurements indicating that hydrocarbon vapour concentrations are being attenuated beyond levels that would be expected for other non-destructive mechanisms, (ii) geochemical data showing the depletion of electron acceptors and generation of metabolic by-products, and (iii) laboratory studies (e.g., microcosm, column studies). BTEX Attenuation There have been several field studies where vertical profiling has enabled evaluation of B T E X attenuation. Fischer et al. (1996) reported that hydrocarbon vapour concentrations below an at-grade building decreased sharply over a small vertical interval (0.1 to 0.7 m depth). The authors suggested that a partial physical barrier to vertical transport (i.e., high moisture content zone) in combination with biodegradation accounted for the steep gradient. Contrasting results were presented by Laubacher et al. (1997) where vapour profiling was performed below and adjacent to a house with a basement. Testing directly below the basement floor slab indicated elevated B T E X vapour concentrations and low O2 concentrations (less than one percent). In contrast, B T E X vapour concentrations adjacent to the house (i.e., at the same depth) were two Chapter 4 51 orders-of-magnitude lower, and O2 levels were about 14 percent. The Laubacher et al. (1997) study is significant since it suggests that hydrocarbon vapour can accumulate below a building. Several studies have involved monitoring at sites not covered by buildings. Ririe and Sweeney (1995) present data showing that B T E X vapour concentrations decreased sharply with decreasing depth. Complimentary geochemical data was obtained to demonstrate biodegradation was occurring. Ostendorf and Kampbell (1991) present similar data for a site contaminated with aviation fuel and derive kinetic biodegradation rate constants using a coupled diffusive hydrocarbon and O2 transport model calibrated using field data. Geochemical Indicators Geochemical data are often good indicators of biodecay. A simplified vadose zone biogeochemical model is presented in Figure 4.1. For each process, the main reactants and final metabolic end products are shown. Aerobic biodegradation or oxidation of hydrocarbons will result in the consumption of O2 and generation of CO2. Oxygen concentrations in hydrocarbon source zones are generally below 1 to 2 % (Jeng et al, 1996; Ririe and Sweeney, 1995; L i , 1995). While there is ample evidence for aerobic processes, anoxic and anaerobic processes are not as well understood. Hydrogen gas has been measured near the interface between the oxic and anoxic zones at hydrocarbon-contaminated sites (Ririe and Sweeney, 1995). Hydrogen gas may form through the breakdown of.volatile fatty acids (VFAs), and in turn may be oxidized in the sub-oxic zone by knall-gas bacteria (Burlage et al, 1998). Hydrocarbon mineralization through denitrifying, sulphate-reducing or iron-reducing bacteria could also occur, although there is little, if any, field data to verify this process for the vadose zone. Under anaerobic conditions, the generation of methane through methanogenesis could also result, as documented for groundwater studies (Wiedemeier et ah, 1995). O X I C Z O N E BTX,0 2 C H 4 , 0 2 O X I D A T I O N > C 0 2 , H 2 0 eg C 7 H 8 + 9 0 2 -> 7 C 0 2 + 4H 2 0 > C 0 2 , H 2 0 eg C H 4 + 2 0 2 -> 2H 2 0 + C 0 2 DENITRIFICATION ? S U B - O X I C Z O N E H 2 OXIDATION ? 0 2 , C 0 2 , H 2 - > H 2 0 . \ (SULPHATE REDUCTION ? \ « H 2 FORMATION ? M E T H A N O G E N E S I S S VFA - co2f'rt2 C 0 2 R e d u c t i o n : C 0 2 , H 2 - > C H 4 , H 2 0 F e r m e n t a t i o n : V F A ( a c e t a t e ) - > C H 4 , C 0 2 M e t h a n o g e n e s i s : B T X , H 2 0 - > C H 4 ; G p 2 : A N O X I C Z O N E B T X S O U R C E Figure 4.1. Simplified vadose zone biogeochemical model (VFA=Volatile fatty acids) Chapter 4 52 The interpretation of geochemical data can be problematic in that there are natural sources of CO2 and methane. Natural sources of CO2 include organic matter, plant root respiration and marine carbonates. Natural sources of methane include biogenic generation through fermentation of acetate and thermogenic sources. Several researchers have proposed the use of stable isotope ratios for carbon and radiocarbon ( 1 4C) content of metabolic endproducts to evaluate the source of these gases (Van der Velde et al, 1995). Laboratory Studies DeVaull et al. (1997) present a detailed analysis of B T E X biodegradation rate constants based on a review of seven laboratory studies and one field study designed to simulate unsaturated soil conditions. Significant variability in first-order rate constants was observed; for example, k based on aqueous concentrations ranged from 0.002 to 35 h r T h e use of Monod-type kinetics (equation 2) provided an improved fit to the experimental data (ki = 0.9 mg/L-hr, Ks = 0.2 mg/L) suggesting that a first-order model is inappropriate when hydrocarbon concentrations increase beyond a certain point. Bekins et al (1998) concluded that when toluene concentrations exceed Ks, it may be better to use a zero-order, as opposed to first-order rate expression to model intrinsic biodegradation. As described by DeVaull et al. (1997), insight into biodegradation kinetics can be gained by considering O2 saturation limits in water, and diffusion rate limitations for micro-scale transfer of dissolved O2 through the cell walls of hydrocarbon-degrading bacteria. To further evaluate the effect of O2 limitations, the use of a qualitative indicator termed biodegradation capacity (BC, dimensionless) is proposed and is defined as: BC = DO/(SR*S) (4) where DO is the dissolved O2 concentration in soil water (M L'3), and SR is the stochiometric ratio for complete mineralization (by use of O2) of hydrocarbon to CO2 and water (M-O2 M-hydrocarbon1). When hydrocarbon concentrations are sufficiently low, the available O2 will be high relative to consumptive requirements. As hydrocarbon concentrations increase, BC will decrease with rate limitations becoming more important as BC approaches unity. An example calculation is provided below for a single-chemical substrate consisting of benzene, which has a SR of about 3 mg-CVmg-benzene. Equilibrium partitioning between phases and a temperature of 20°C yields a DO concentration of about 8 mg/L in soil moisture, in equilibrium with approximately 20 % O2 by volume in soil gas. The benzene concentration in soil water, corresponding to a BC equal to unity, is about 2.7 mg/L (concentration in vapour is 0.48 mg/L). This calculation suggests that first-order biodegradation kinetics may not be applicable for B T E X concentrations in excess of a few milligrams per litre since O2 is no longer available in excess relative to the hydrocarbon concentrations. FIELD STUDY Overview Biodegradation processes were evaluated through a comprehensive multi-year testing program conducted at the Chatterton research site. The testing scope has included baseline analyses of soil samples, soil gas monitoring below and adjacent to a building (greenhouse), Chapter 4 53 semi-continuous monitoring of O2 and pressure below the building slab, monitoring of environmental conditions (e.g., temperature, barometric pressure) and in situ respiration tests. Pertinent methods and results are described below with additional information provided in Hers and Zapf-Gilje (1998). Site Description The former Chatterton petrochemical plant is located in Delta, B.C. Near-surface soil in the plant area consists of fill (about 3.5 m thick) underlain by native silt. In most areas, the fill consists of a thin surface layer up to 0.3 m thick consisting of sandy silt to fine sand with trace organic matter, underlain by dredged river sand. The surface unit is henceforth referred to as the "crust". The depth to the water table generally ranges from 1.5 to 2.5 m below ground surface. Releases of B T X from sources removed from the study site and lateral migration on the water table have resulted in an extensive zone of residual non-aqueous phase liquid (NAPL). The N A P L is distributed vertically over an approximate 1 m interval, which corresponds to the approximate water table fluctuation. No other hydrocarbon compounds are present at the plant site. Some testing at a nearby non-contaminated reference site with similar soils was also conducted for comparative purposes. Average annual precipitation at a nearby weather station (i.e., 7 km from site), considered to approximate site conditions, is 1240 mm (Richmond Nature Park). ... Methods Continuous soil cores were initially obtained in March 1997 from 12 boreholes drilled using a Geoprobe™ sampler (Figure 4.2). Multiple soil samples from each borehole, obtained at 0.15 to 0.3 m intervals, were generally analyzed for B T X , grain size, moisture content, porosity and soil organic matter (SOM). Soil samples from the B T X contaminated area and from the reference site were also analyzed for various species of nitrogen (nitrate, nitrite, and total Kjeldahl nitrogen (TKN)) and sulphur. Soil samples representing a broad range of B T X concentrations were enumerated for total heterotrophic bacteria (THB) and BTX-degrading bacteria (BTXB) using the most-probable number (MPN) method (Haines et al, 1996). A small building (greenhouse) was subsequently constructed for research purposes in August 1997 (Figure 4.2). The building was fastened to a 6.1 by 9.3 m at-grade concrete slab of 125 mm nominal thickness. The slab was constructed with a 2-mm wide crack located approximately 0.5 m from the edge of the slab. Steel inserts were subsequently used to vary the open area of the edge crack. The crust below the building was replaced with sand fill and the foundation slab during construction. Multiple soil gas monitoring events, and testing of soil moisture content was conducted between March 1997 and February 1999 to assess the effect of seasonal changes on soil vapour fate and transport (Figure 4.3). Soil gas probes consisted of either 6.25 mm or 12.5 mm internal diameter hollow steel rods with 50 mm long perforated or slotted tips driven to the desired depth. Multiple soil gas probes were installed at 0.15 to 0.3 m intervals below the approximate centre of the building and adjacent to the building. Soil gas samples obtained for B T X analyses were collected using gas-tight glass and Teflon syringes while gas samples obtained for light gas Chapter 4 54 analyses (0 2 , C 0 2 and CH 4 ) were collected using syringes or 500 ml TedlarTM bags. Samples were analyzed for B T X using a SRI 8610 gas chromatograph with photoionization detector (GC/PID). The GC/PID capillary column was a DB-624 30 m by 0.53 mm column with a methylated phenoxy stationary phase (1.8 (xm thickness). Samples were analyzed for light gases using a HP 5880 GC with thermal conductivity or flame ionization detector (TCD or FID), or using a Landtec Control Technologies GA-90 field portable gas detector. One of two capillary columns were used for the GC/TCD consisting of a (i) HP Plot molecular sieve 5A 30 m by 0.53 mm column or (ii) JW GSQ 30 m by 0.53 mm column. BH-50 T B H 9 7 - 2 0 Paved Road BH-17 Groundwater Collection Tank BH-1S BH97-10 BH97-19 • BH97-8± I £± 97-19 x • •••/• / • •'••V WW97-2 Concrete Foundation I SITE A ^BH97-6 Research Greenhouse (Constructed in August 1997) BH97-7 \ T M W 9 7 - 1 LEGEND. Borehole or Monitoring BH97-T Well Location BH97-31l A J SG-BR A BH97-S0 J ' BH97-2 BH97-3 A Soil Gas Probe Location 0 : • . 5 10 15m i SITE B ~BH97-4 Figure 4.2. Site plan. A Datawrite Research Co. Model XT252 electrochemical cell and datalogger was used to measure 0 2 concentrations below the building slab near location SG-BC (Figure 4.2). Differential pressures between the building, atmosphere and various soil gas probes (i.e., representative of soil pressures adjacent to probe openings) were measured using Setra Systems Model 264 differential pressure transducers with a full-scale range of 63 Pa, and accuracy equal to approximately 1 percent of the full-scale range. To further investigate biodegradation processes, an in situ respiration test was conducted north of the building at probe cluster SG-A (Figure 4.2, Appendix VII). The test was conducted by injecting air with helium tracer over a 24-hour period (U.S. EPA, 1995). After injection was stopped, the 0 2 , C 0 2 and helium concentrations were monitored over multiple depth levels for a Chapter 4 55 monitored over multiple depth levels for a two-day period. Helium concentrations remained relatively constant indicating O2 depletion was likely due to biological processes, as opposed to O2 transport away from the test location. Baseline Soil Testing Results Elevated B T X concentrations, indicative of residual N A P L , were measured below 1.4 to 1.5 m depth below ground surface. The residual N A P L layer is at least 1 m thick, and at least 0.5 m was exposed above the water table during the testing program. Above 1.4 to 1.5 m depth, the B T X concentrations in soil decreased sharply. The medium sand fill is relatively uniform with a silt (less than 0.074 mm, U.S.S classification) content ranging from 1.7 to 5.5 % for samples composited over 0.15 m intervals. Visual examination of continuous soil cores indicated a thin silty fine sand layer (i.e., few millimeters thick) at about 0.92 m depth near vapour probes (SG-BR) adjacent to the building (Figure 4.2). Multiple cores indicate that this fine layer is locally continuous on the scale of a few meters, but was not observed below the building. The fine layer has implications for B T X vapour transport as discussed in the modeling section of this paper. Soil moisture contents below the building were at moderate levels and correspond to residual saturations (relative to soil pore volume) between 25 and 58 % (Figure 4.3). On average, moisture contents below the building appear to be highest between about 0.6 and 0.9 m depth. Greater variability in soil moisture content is observed in boreholes adjacent to the building. BELOW BUILDING NATURAL SOIL COVER Moisture Content (% wet wt.) Moisture Content (% wet wt.) 0 5 10 15 20 25 0 5 10 15 20 25 Figure 4.3. Soil moisture content below and adjacent to building (approximate total porosity values are provided in Table 3). The nitrogen and sulphur testing indicated that the total Kjeldahl nitrogen (TKN) concentrations were elevated within the zone of inferred greatest biological activity based on O2 depletion, relative to shallower and deeper zones and the reference site. In addition, nitrate was depleted relative to the reference site (Table 4.1). Sulphide concentrations were, in contrast, Chapter 4 56 anaerobic reactions. The soil organic matter content (SOM) was relatively consistent and ranged from 0.55 to 0.70 %, except for one sample from 0.88 to 0.92 m depth where the S O M was 0.97 %. This sample was collected immediately above the thin fine-grained soil layer and corresponds to the depth where a significant decrease in O2 concentrations was observed. The T K N concentrations were highest in the sample from 0.9 to 1.05 m depth. Although much more sophisticated methods are available for biomass characterisation (e.g., Franzmann et al., 1999), the elevated T K N and S O M levels near 0.9 m depth may provide a crude indication of biomass formation and where biological activity is occurring. Table 4.1. Results of nitrogen, sulpher and SOM analyses in soil Depth Nitrate Nitrite T K N 1 Total N^ Sulphate Sulphide SOM (m) (mg/kg) (mg/kg) (mg/kg) (mg/kg) (mg/kg) (mg/kg) (%) Adjacent Buildin2J 0.3-0.45 N/A N/A N/A N/A N/A N/A 0.57 0.45-0.6 0.06 0.01 3.2 3.3 <10 1.3 0.60 0.6-0.75 <0.05 0.01 3.1 3.1 <10 1.0 0.71 0.75-0.9 <0.05 0.02 6.7 6.7 <10 1.1 0.55 0.88-0.92 N/A N/A N/A N/A N/A N/A 0.97 0.9-1.05 <0.05 0.01 21.9 21.9 <10 1.3 0.66 1.05-1.2 <0.05 <0.01 10.3 10.3 <10 1.1 0.48 1.2-1.35 0.06 0.01 ' 8.6 8.7 <10 0.5 ' 0.70 1.35-1.5 " N/A N/A " N/A ' N/A N/A N/A 0.55 Reference Site Location 1 0.45-0.9 0.24 <0.02 <0.5 0.24 <10 0.2 N/A Location 2 0.45-0.9 0.24 <0.02 0.6 0.30 <10 0.3 N/A TKN = Total Kjeldahl Nitrogen Total N ~ TKN + nitrate + nitrite; Organic N ~ TKN - ammonia Samples collected 1 m north of SG-BR probe cluster on June 2, 1999 (from one soil core). The THB counts ranged from 4.9E+02 CFU/g to 1.7E+09 CFU/g (geometric mean of 6.4E+05) while the B T X B counts ranged from no growth (less than 50 CFU/g) to 1.1E+06 (geometric mean of 1.3E+03). There was no correlation between THB or B T X B counts with B T X concentrations, or inferred zones of biodegradation based on soil vapour monitoring and in situ respiration testing. Franzmann et al. (1999) likewise found a poor correlation between microbial numbers, calculated from the phospholipid content, and B T E X contamination within an aquifer. Soil Gas Monitoring Results Soil vapour benzene concentrations at probes (SG-BC) below the building (greenhouse), and at probes (SG-BR) not covered by a building (i.e., below the present building before it was constructed, or adjacent to the building) are presented in Figure 4.4. The benzene concentration profiles below the centre of the greenhouse were relatively consistent over time and are characterized by high concentrations at depth, significant attenuation (about 3 orders-of magnitude) between about 0.4 to 0.8 m depth, and lower but variable concentrations within 0.4 m depth below slab surface. The measured vapour concentrations below 1 m depth are within a factor of two and one half of those predicted assuming equilibrium partitioning between the N A P L and vapour phases, and vapour pressure adjusting using the mole fraction and Raoult's Law. The benzene concentration profiles for the no building case were somewhat more Chapter 4 57 variable than those below the building. Postulated causes include variation in surface water infiltration and moisture content, and advective soil gas pumping caused by barometric pressure and/or temperature fluctuations. Similar results were obtained for toluene and m&/?-xylene below and adjacent to the building; results for one typical monitoring round are presented in Figure 4.5. The O2 concentrations below the centre of the building, and at probes not covered by the building are presented in Figure 4.6. The O2 concentrations below the building were relatively consistent over time and indicate significant O2 consumption is likely occurring between about 0.6 to 0.9 m depth. Oxygen levels in near surface soil below the centre of the building slab were depleted and ranged between approximately 8 and 13 %. Carbon dioxide concentrations above 0.9 m depth generally ranged from 6 to 9 %, while below 0.9 m the concentrations generally ranged between 10 and 13 %. Methane concentrations were also high (i.e., at percent levels) below about 0.9 m depth (Figure 4.5). The O2 concentration profiles for the no building case exhibited greater variability. The O2 concentration profiles after a relatively long period of dry weather (September 1998 monitoring rounds) were consistent and showed a significant decline between about 0.7 m and 0.9 m depth. In contrast, O2 concentrations during wet periods (November 1998) indicated greater variability with lower concentrations at shallow depths. Rainfall during the week prior to each sampling event in November 1998 exceeded 30 mm. Lower 0 2 concentrations at shallow depths may have been a result of reduced O2 flux through the surface crust due to high moisture content, and O2 depletion at depth through biological activity. Oxygen concentrations at a non-contaminated reference site with similar soils were within 1.0 % of atmospheric levels while CO2 concentrations were equal to or below 0.1 % throughout the soil column. N A T U R A L S O I L C O V E R ( S G - B C , B R ) B e n z e n e C o n c e n t r a t i o n (mg/L) B E L O W B U I L D I N G ( S G - B C ) B e n z e n e C o n c e n t r a t i o n (mg/L) 0.0001 0.01 100 0.0001 0.01 1 100 JZ1 ra W o •S0.5 o c o O o a. o H E 1 o a. a> Q 1.5 -Sept 2/97 -Oct 21/97 -Nov 14/97 - Mar 18/98 -June 25/98 Figure 4.4. Benzene vapour concentrations below and adjacent to building (MC = moisture content). Chapter 4 58 BELOW BUILDING (SG-BC) Concentration (mg/L) 0.1 10 Oct 21/97 — Benzene —Toluene — m&p-Xylene 1000 BELOW BUILDING (SG-BC) 02, C02, CH4 (Landtec GA-90) (%) XI a (/> o +* CO l_ o c o o o a. o l-E o a 0) Q 0.5 1.5 10 15 20 > Ct 17/971 —•—CH4 - • - C 0 2 - * - 0 2 - • - M C Oct.23 u f i > Figure 4.5. B T X vapour and light gas concentrations below building (MC = moisture content. NATURAL SOIL COVER (SG-BC, SG-BR) 0 2 Concentration (%) 0 5 10 15 20 0 BELOW BUILDING (SG-BC) 02 Concentration (%) 0 5 10 15 20 Figure 4.6. Oxygen gas concentrations below and adjacent to building. The semi-continuous O2 monitoring below the slab indicated diurnal changes in concentrations (up to 2 %) due to temperature fluctuations and day-long to week-long O2 trends that appear to be correlated to barometric pressure. Pressure monitoring indicating that significant effects from diurnal and atmospheric pressure changes were limited to above 0.3 m Chapter 4 59 depth beneath the building slab. As discussed in subsequent sections of this paper, advective processes may be important in terms of 0 2 replenishment directly below the slab. Estimation of Field Biodegradation Rates Monitoring at the Chatterton site suggests that processes affecting fate and transport below the building can be divided in three zones based on depth below ground surface consisting of primarily diffusion in the deep zone, biodegradation and diffusion in the mid-depth zone, and advection, biodegradation and diffusion in the shallow zone. First-order degradation coefficients were estimated for the mid-depth zone by fitting model-predicted B T X concentrations, based on an analytical solution for one-dimensional steady-state diffusion and reaction, to measured B T X concentrations. The vapour profile between about 0.5 to 0.8 m depth was used since B T X attenuation was significant and approximately log-linear over this interval. The B T X vapour profiles below and adjacent to the building were also consistent over time suggesting approximate steady-state conditions (Figure 4.4). A first-order model was considered appropriate based on the log-linear B T X vapour profile and presence of sufficient O2 for biodegradation based on the biodegradation capacity (BC). Steady-state vapour-phase diffusion and reaction for a homogeneous soil can be represented by the following equation: Df c7Cg/dZ? = G =( V Bw/H') CR (5) where Dgeff is the effective vapour-phase diffusion coefficient (L2 T1), Co is the vapour concentration (mol L3), G is the first-order rate of mass consumption (mol L T1), kj is the first-order degradation rate based on the pore-water concentration (T1), 6w is the water-filled porosity (L L ), and H' is Henry's Law Constant (dimensionless). Assuming the top and bottom boundary conditions can be represented by constant concentrations, an analytical solution can be derived that enables calculation of a Damkohler number, or the ratio of the degradation rate over the diffusion rate (DeVaull et al., 1997), as follows: C ( Z ) . - ) . - » • C g ( Z = 0) e - - e ~ Dam = d = Degradation Rate / Diffusion Rate = kj 0wL2/( H' Dgeff) (1) where Dam is the dimensionless Damkohler number, L is the domain length over which transport occurs (L) and /3 is the vapour concentration at distance L. For the Chatterton site, the effective diffusion coefficient was calculated using the measured tortuosity factor obtained from in situ push-pull tracer tests using helium (Hers et al, 1999) and the free-air diffusion coefficient. Between 0.5 and 0.8 m depth, the mean Dgeff was 0.044 cm2/sec while and 6 were equal to 0.18 and 0.36, respectively. Using the above input values, the estimated first-order degradation rate constants based on soil pore water concentrations (kj) were on the order of 1.2 hr" for benzene, 0.9 hr"1 for toluene and 0.5 hr"1 for m&p-xylene. / Chapter 4 60 Zero-order biodegradation rates (kw , mol L T ) were estimated from in situ respiration tests using the method in U.S. EPA (1995). Respiration tests were conducted for the lower part Of the mid-depth zone, and deep zone described above. A zero-order model was considered appropriate based on approximately linear O2 consumption curves and since total B T X concentrations in vapour exceeded about 1 mg/L at the respiration test locations. The zero-order total hydrocarbon (i.e., BTX) degradation rates in pore water ranged between 0.8 and 6 mg/L-hr and were highest at 0.9 m depth, which again was consistent with the depth interval over which the highest biodegradation rates are expected. A relatively good comparison is obtained between the first- and zero-order rates measured at the Chatterton site and published biodegradation rates based on laboratory and field studies, compiled by DeVaull et al. (1997) (Figure 4.7). In addition, the Chatterton site total B T X first-order rate intersects the zero-order rate at a pore-water concentration of about 2 mg/L, which is consistent with the biodegradation model described in Section 2.3.3. Together the first-order degradation rates (from vapour profiles), and zero-order degradation rates (from the in situ respiration tests), provide a practical field-based method to approximate the relationship between biodegradation rate and concentration based on Monod kinetics. Comparison to Other Case Studies The estimated Damkohler numbers and first-order aerobic biodegradation rates for the Chatterton site and several other case studies are presented in Table 4.2. The degradation rates are highly sensitive to the Damkohler number, effective diffusion coefficient and moisture content. Biodegradation rates are overestimated when there are thin unquantified high moisture content layers, since these layers represent a partial barrier to diffusive transport. At sites where there are unresolved moisture content effects, fitted biodegradation rates are, in effect, lumped parameters. Due to the various sources of uncertainty, the estimated biodegradation rates ( k w 1 ) should be considered order-of-magnitude estimates. The vadose zone first-order degradation rates are about two to four orders-of-magnitude higher than those obtained for dissolved B T E X plumes in groundwater (Wiedemeier et al, 1996). While the vadose zone biodegradation rates are high compared to saturated zone rates, they are on the same order or lower than rates obtained for biofilters, suggesting that vadose zone rates given in Table 4.2 may be reasonable. For example, Andreoni et al. (1997) report a toluene removal rate of 6 mg/L-hr for a biofilter constructed of wood bark, while Conti et al. (1999) report a toluene removal rate of 135 mg/L-hr for a biofilter constructed of peat beads. Chapter 4 61 1.E+04 E £ l . E + 0 2 > o> CO 1.E+00 1.E-02 c i_ d) Q. CD •«—' n DC c •I 1.E-04 m T> (0 l -O) 0) Q 1.E-06 H From DeVaull et al. (1997) - — Monod-type kinetics • - -first-order kinetics DeVaull et al. (1997) Holman & Tsang (1995) Jin et al. (1994) Karlson & Frankenberger (1989) Ostendorf & Kampbell (1989) Ostendorf & Kampbell (1989) Salanitro et al. (1989) Salanitro et al. (1989) • Fitted first-order total BTX degradation rates from vapour profile kw = 2.6 hr'1 • Maximum measured total BTX degradation rate - in situ res-piration test K w = 6 mg/L-hr 1.E-06 1.E-04 1.E-02 1.E+00 1.E+02 Pore Water Concentration (mg/L) 1.E+04 Figure 4.7. Comparison of field degradation rates to published data in DeVaull et al. (1997) (used with permission). Table 4.2. Measured aerobic biodegradation rates from field studies Site Chemical Class Chemical Biodegrada-tion Layer Thickness (m) Damkohler Number (Dam) First-order Degradation Rate (water-(phase)Gir"1) Chatterton (this paper) BTX Benzene Toluene m&p-xylene 0.3 0.3 0.3 7.4+/-1 6.05+/-1 5.2+/-1 0.5-2.0 (1.2/ 0.3-1.5 (0.9)3 0.2-0.8 (0.5)3 Alameda Fischer et al. (1996) Gasoline iso-pentane 0.2 -61 ~2' Traverse City Ostendorf & Kampbell (1991) Aviation Fuel Total Hydrocarbon 3 4 -0.012 California Gasoline Benzene 2 9 0.42 Ririe and Sweeney (1995) Notes: ' Degradation rate overestimated due to likely presence of low moisture content layer (i.e., physical barrier). 2 Order-of-magnitude estimate due to significant uncertainty in air-filled and total porosity, and physical-chemical properties. 3 Range and best estimate (in parentheses) based on probable range of values for Damkohler Number and moisture content. Chapter 4 62 NUMERICAL SIMULATION OF VADOSE ZONE BTX TRANSPORT A two-dimensional numerical model (VADBIO) for multispecies transport in the unsaturated zone was developed to compare various types of biodegradation models, and was based on conditions observed at the Chatterton site. The V A D B I O model was used to evaluate the effects of a building, variations in moisture content, and advection on biodegradation processes and kinetics. Model Description The differential equation for vadose zone chemical transport incorporating two-dimensional diffusion, gas-phase advection, absorption and first- or zero-order reaction, assuming equilibrium chemical partitioning between the vapour, aqueous and sorbed phases, and linear isotherms, for a single chemical species may be written as follows: where 6g is the gas-phase porosity (L3L3), R is the retardation coefficient (dimensionless), D ^ , Dxy, Dyx and Dyy comprise the dispersion tensor (L2 T1), (Vg)x, (Vg)y are the gas-phase velocities (LT1), OL, OCTH are the longitudinal and transverse dispersivities (L), Dg and Dw are the vapour-and aqeous-phase diffusion coefficients (L2 T1), pb is the dry bulk density of the soil (ML3) and Kd is the distribution coefficient between the sorbed and aqueous phases (VM1). Chapter 4 63 The VADBIO model simulates vapour-phase hydrocarbon transport described by equations 8 to 11 and models biodegradation processes using four different methods: 1. First-order biodecay (model no.l). 2. Combined first and zero-order biodecay. For lower concentrations where the substrate concentration limits biodecay, a first-order process is assumed, while above a critical concentration, a zero-order process is assumed (model no. 2). 3. Instantaneous-reaction model where the hydrocarbon and O2 concentrations are sequentially computed, and at each time step combined using superposition to simulate the reaction between O2 and hydrocarbon compounds. The O2 consumed is linked to hydrocarbon degraded through the stochiometric ratio for complete mineralization of the hydrocarbon compound using the following equations: H(t+l) = H(t)-O(t)/SR;O(t+l) = 0(zero) where H(t) > O(t) /SR (12) 0(t+l) = O(t) - H(t) * SR ; H(t+1) = 0 (zero) where O(t) > H(t) * SR (13) where H is the total hydrocarbon concentration, O is the O2 concentration and SR is the average stochiometric ratio of O2 to hydrocarbon consumed. The instantaneous reaction model is appropriate for single hydrocarbon compounds, or multiple compounds with similar SR values (e.g., BTEX) (model no. 3). 4. Combined first- and zero-order biodecay as described above except that the O2 availability in soil at each time step is checked by simultaneously solving the O2 transport equation. The mass of O2 consumed is obtained using the stochiometric relationship for complete mineralization and the mass of hydrocarbon degraded by either a first- or zero-order process (model no. 4). The numerical model solution assumes a rectangular domain and hydrocarbon transport from a subsurface source at the base of the domain to ground surface (Figure 4.8). Boundary conditions for hydrocarbon transport are as follows: 1. Bottom layer: constant or time-varying concentrations (Dirichlet condition). Possible compositional changes in N A P L and source vapour concentrations over time are not accounted for in this version of the model. 2. Top layer: constant concentration equal to atmospheric hydrocarbon concentration, which assumes chemicals are subject to instantaneous mixing and dilution in the atmosphere. Chapter 4 64 Centre of Greenhouse ox g m Atmospheric Concentration (constant) c g = 20.9% <5x -=o Building Slab -4.3m Y t A x=0.05 Ay=0.05 X <5c: <5x -=0 -=o • 7.6m C g = Source Cone. (6=15, T=20, X-0.9 mg/L) Figure 4.8. Model domain and boundary conditions used for 2-D simulations. The boundary conditions for O2 transport are as follows: 1. Bottom layer: constant flux set equal to zero (Neumann condition). 2. Top layer: constant atmospheric concentration with time (20.9%). Initial conditions used for simulations presented in this paper are hydrocarbon concentrations equal to average atmospheric concentrations (except at the bottom boundary), and O2 concentrations equal to 20.9 % (except at the bottom boundary). The model allows for input of variable soil properties and an at-grade building with cracked floor slab. The effective diffusion coefficient for soil is calculated using the Millington and Quirk (1961) relationship: Dgeff = dpg = Tpair = (6/0/3/02) * D a . r ( 1 4 ) Dweff = 0J)w/H' = Tj)water = (0JO/3 / ( 0 2 H ' ) ) * D w a t e r ( 1 5 ) where Tg and tw are the gas-phase and aqueous-phase tortuosity factors. Diffusion through the building slab occurs primarily through cracks and other openings, and through intact concrete. The approach for simulations presented in this paper is to assume that diffusion only occurs through cracks, but to use a somewhat conservative crack ratio to account for possible diffusion through intact concrete. Radon research suggests that diffusion through both cracks and intact concrete can be significant (Renken & Rosenberg, 1995; Nielson et al., 1997). Little research on hydrocarbon diffusion through building foundations has been conducted. Since discretization of crack-sized openings requires too fine a model grid, diffusion through concrete is modeled assuming that a rough estimate of the bulk tortuosity of the concrete slab can be obtained as follows: Chapter 4 65 Dgcan* =(Ac*dust/Ab) *Dair (16) where Dgconeff is the effective diffusion coefficient of the concrete, Ac is the area of the cracks in concrete, Tdust is the tortuosity factor for dirt-filled cracks and A/, is the building slab area. The model output includes concentration versus time data, and calculated hydrocarbon flux across the top boundary of the model domain. The numerical solution utilizes a forward, or explicit, temporal discretization scheme. The scheme is subject to the following stability constraints: At<(AX)2/(2*(Dg'ff+Dw'ff)); At < (AY)2 /(2 *(Dgeff + Dweff)); At<l/kw (17) where AX and AY are the grid block size in the x and y directions. The V A D B I O model is similar to the R-UNSAT computer model developed by the U.S. Geological Survey (Lahvis and Baehr, 1997). The main difference relates to advection in that V A D B I O only incorporates gas-phase transport while R-UNSAT only incorporates aqueous-phase transport. The VADBIO model includes additional options for biodegradation modeling and incorporation of building slab properties whereas R-UNSAT has greater flexibility in terms of boundary conditions, and includes an analytical model. Comparison to Analytical Solutions Model-predicted benzene transport was compared to analytical solutions for: (i) steady-state one-dimensional diffusion with first-order decay (equation 6), and (ii) one-dimensional uniform soil gas flow, two dimensional dispersion (and diffusion) and first-order decay for a finite, constant source, as adapted from the extended pulse model (approximate solution) by Domenico (1987). The diffusion solution (equation 6) assumes constant concentrations at the top and bottom boundary, while the advection (Domenico) solution assumes a finite, constant source combined with an infinite flow domain. A homogeneous, isotropic porous media is assumed by both models. For the diffusion model case, the numerical solution closely approximates the analytical solution for a grid size equal to or less than 0.05 m (Figure 4.9). For the advection model case, the numerical solution closely approximates the analytical solution for the grid sizes evaluated. Subsequent model simulations in this paper use a grid size equal to or less than 0.05 m. Chapter 4 66 Benzene Concentration (mg/L) 0.00000001 0.00001 0.01 10 Analytical Diffusion & Decay Numerical Diffusion & Decay 0.1 m grid Numerical Diffusion & Decay 0.05 m grid Numerical Diffusion & Decay 0.025 m grid Analytical Advection, Dispersion & Decay Numerical Advection, Dispersion & Decay 0.1 m grid Pe=0.85 Numerical Advection, Dispersion & Decay 0.025 m grid Pe=0.21 Figure 4.9. Comparison of numerical and analytical solutions for steady-state conditions, (input parameters: C'(source) = 15 mg/L, dw = 0.14, 6= 0.36, H' = 0.23, kj = 1.2 hr 1 , U = 5.6xl0"6 m/s, at = 0.1 m, U = specific discharge, at = longitudinal dispersivity, Pe = Peclet number, all other symbols defined in text) Comparison of Biodegradation Models No. 1,2,3 and 4 (1-D Diffusion and Biodecay) The numerical solutions for the four biodegradation models were compared for steady-state one-dimensional diffusion of B T X with biodecay with results presented in Figure 4.10. Two cases were simulated: (i) a uniform soil with relatively low soil moisture content (6w equal to 0.14, 6 equal to 0.36), and (ii) a two-layer soil profile with a relatively high moisture content for the surface layer (dw equal to 0.28, 6 equal to 0.36). Model input parameters are provided in Table 4.3. The first-order degradation rates were estimated by fitting the analytical model previously described to measured hydrocarbon vapour profiles at the Chatterton site while the zero-order rates were obtained from in situ respiration tests. Individual zero-order rates for B T X constituents were assumed to be in the same proportions as the first-order rates. For the uniform soil layer case, all biodegradation models predict a sharp decrease in benzene concentration just above the B T X source and an approximate linear decrease in O2 concentrations over the depth range. No significant difference was obtained between model no. 2 (first- and zero-order) and model no. 4, which incorporates an O2 availability check, since biodegradation is not limited by O2 availability over most of the soil profile. Results for model no. 3 (instantaneous reaction) showed a sharper decrease in benzene concentration than models no. 2 and 4 since again O2 is not rate limiting. For the two soil layer case (i.e., high surface soil moisture content), there was a significant difference in models no. 2 and 4 since biodegradation at depth is limited by O2 availability, but there was virtually no difference between models no. 3 and 4 for benzene vapour concentrations above 1 mg/L. The results highlight the importance of Chapter 4 67 O2 diffusion rates (and hence moisture content) and selection of a biodegradation model that incorporates appropriate reaction kinetics combined with O2 availability (e.g., model no. 4). Table 4.3. Input Parameters for Comparison of Biodegradation Models Parameter Comparison of Bio Models Chatterton Comparisons Sand Crust (0.2m) Sand Crust (0.2m) Soil Properties Water-filled Porosity (-) 0.14 0.28 variable variable Total Porosity (-) 0.36 0.36 0.36 0.403 Organic Carbon Content (-) 0.006 0.006 0.006 0.006 Bulk Density(g/cm3) 1.67 1.67 1.67 1.58 Benzene toluene m&p-xylene Oxygen Chemical Henry's Law Constant (-) 0.23 0.28 0.23 31.6 Properties Diffusion Coef in Air (m2/sec) 8.44E-06 7.60E-06 7.00E-06 2.06E-05 Diffusion Coef in Water (m2/sec) 1.00E-09 9.4E-10 8.5E-10 1.00E-09 Partitioning coef (log KoC) (cnrVg) 1.96 2.12 2.56 1 Concentrations Cg(x,y,t=0) (mg/L) 0 0 0 279 Cg(x,y=0,t) (mg/L) 15 20 0.9 Variable Cg(x,y=1.4,t)(mg/L) 0.000005 0.00001 0.000003 279 Biodegradation First-order rates (kw') (hr1) 1.2 0.9 0.5 N/A Rates Zero-order rates (kw°) (mg/L-hr) 1.36 1.06 0.56 N/A Average stochiometric ratio 3.1 3.1 3.1 N/A Q . O Q Benzene & 0 2 Concentration (mg/L & %) Benzene & 0 2 Concentration (mg/L & %) 0.0001 0.01 1 100 0.0001 0.01 1 10c UNIFORM SOIL CONDITIONS 4-9W = 0.28, 6 = 0.36 3&4 HIGH MOISTURE S U R F A C E L A Y E R - • — B e n z e n e 1 s t o r d e r ( n o . 1 ) — B e n z e n e I R ( n o . 3 ) • * - - 0 2 I R ( n o . 3 ) • B e n z e n e 1 s t / z e r o o r d e r ( n o . 2 ) • B e n z e n e 1 s t / z e r o o r d e r w \ 0 2 c h e c k ( n o . 4 ) • • • - 0 2 1 s t / z e r o o r d e r w \ 0 2 c h e c k ( n o . 4 ) Figure 4.10. Comparison of biodegradation models for uniform and non-uniform soil conditions using 1-D diffusion and reaction model (steady-state conditions, IR = instantaneous reaction). Chapter 4 68 Comparison to Chatterton Vapour Profiles (2-D Diffusion and Biodecay) Model-predicted and measured soil vapour concentrations for the Chatterton site were compared using the biodegradation model incorporating first- and zero-order biodegradation with O2 availability check (model no. 4). Two additional refinements were made to the model for the purposes of simulations presented in this section. First, biodegradation was turned off once O2 concentrations reached 1.5 % since biodegradation rates are significantly reduced at low concentrations (DeVaull, 1997). Second, biodegradation was turned off once hydrocarbon concentrations reached atmospheric levels. The justification for this refinement is that there is insufficient substrate to support biodegradation at low hydrocarbon concentrations. Estimation of the precise concentration at which biodegradation becomes substrate limiting is beyond the scope of this assessment; however, an arbitrary cut-off point equal to atmospheric concentrations was considered sufficiently accurate for the purposes of these simulations. Input parameters were identical to those given in Table 4.3 except that soil moisture was varied, and an at-grade building slab was simulated over a portion of the model domain (Figure 4.8). Due to the dilution that occurs through building ventilation, B T X concentrations inside the building (i.e., top left boundary condition) were assumed to be equal to atmospheric concentrations. The model domain was chosen to simulate conditions at the Chatterton site. Since vapour transport is likely symetrical in relation to the building slab, only half the building is incorporated in the domain; the left edge of the model domain corresponds to concentrations below the centre of the building while the right edge corresponds to concentrations adjacent to the building. The ratio of the crack area to building slab area (TJ = AJAu) ranged from 0.0003 to 0.001, and tdust was varied between 0.25 and 1. The lower values for 7] and Tdust represent an estimate of the crack ratio based on measurements of visible cracks while the upper values for rj and Tdust are considered conservative estimates that take into account micro-scale cracks and pores through which diffusion can occur. Two scenarios were simulated corresponding to approximate measured moisture content conditions during dry periods in summer and early fall (Case 1) and during wet periods in late fall and winter (Case 2). The moisture content profiles used for the simulations are provided in Figure 4.11. As shown, the most significant difference in moisture content is for the surface crust, which tends to retain moisture for longer periods of time during wet periods. The crust is not present below the greenhouse since it was replaced by the foundation slab and sand fill. Laterally constant moisture contents were assumed for the simulations. Based on the measured moisture content variation (Figure 4.3), it is recognized that the moisture content profile adjacent to the building is approximate. The predicted benzene and oxygen concentrations below the centre of the building for simulations corresponding to 0.5, 1, 2 and 3 months are presented in Figure 4.12 (Case 2 only). Both predicted and measured concentrations below (centre) and adjacent to the building at 3 months are presented in Figure 4.13 (Case 1 and 2). The upper ranges of the crack ratio and tortuosity factor were chosen for these simulations. The time to reach approximate steady-state conditions varies depending on moisture content and location. Approximate steady-state conditions were reached in about one month adjacent to the building for Case 1 (dry summer) conditions (results not shown). As shown in Figure 4.12, benzene concentrations below the building were still slowly increasing after three months for Case 2 (wet winter) conditions. Chapter 4 69 Lateral O2 diffusion from adjacent to the building has a slight effect on O2 concentrations below the centre of the building. The 2-D O2 concentrations were about 0.1 % higher than 1-D O2 concentrations, for Case 1 conditions. Moisture Content (%) 5 10 15 0.5 a. Q 1.5 20 •Case 1 (Dry Conditions) •Case 2 (Wet Conditions) Figure 4.11. Moisture content profile used in numerical simulations (see Table 4.3 for porosity values). 0.000001 0 Benzene Concentration (mg/L) 0 . 0 0 0 1 0 . 0 1 1 1 0 0 E 0 . 5 a. at a 1 . 5 — • — T = 0 . 5 m o n t h —s—T=1 m o n t h - A - T = 2 m o n t h — * — T = 3 m o n t h BELOW BUILDING 1 . 5 Oxygen Concentration (%) 5 1 0 1 5 - T = 0 . 5 m o n t h • T = 1 m o n t h - T = 2 m o n t h • T = 3 m o n t h BELOW BUILDING Figure 4.12. Predicted vapour concentrations below centre of building for 2-D diffusion with reaction (Model No. 4) - wet (winter) conditions (rj = 0.001, Tdust =1). For Case 1, there is a significant difference between the benzene and O2 concentrations below and adjacent to the building (Figure 4.13). This is a result of the much higher diffusive O2 flux through the surface crust, compared to the concrete slab. For Case 2, there is less of a difference between the benzene and O2 concentrations below and adjacent to the building since the diffusivity through the crust is lower. Adjacent to the building, there is a difference in O2 Chapter 4 70 profiles for wet and dry conditions but virtually no difference between the benzene concentration profiles. This lack of sensitivity of benzene to moisture content is caused by O2 levels that remain sufficiently elevated for biodegradation to occur. Benzene Concentration (mg/L) 0.000001 0.0001 0.01 1 1001 1.5 - w e t ( w i n t e r ) c o n d i t i o n - d r y ( s u m m e r ) c o n d i t i o n BELOW BUILDING Benzene Concentration (mg/L) 0.000001 0.0001 0.01 1 1001 0 1.5 - w e t ( w i n t e r ) c o n d i t i o n - d r y ( s u m m e r ) c o n d i t i o n NATURAL SOIL COVER Oxygen Concentration (%) 5 10 15 20 0 H 0.5 1.5 • 1 1 t l! -1 \ '• 0 w e t ( w i n t e r ) c o n d i t i o n — a — d r y ( s u m m e r ) c o n d i t i o n m e a s u r e