oEPA t
United States 'J
Environmental Protection
Agency
Investigation of a Sustainable
Approach to In-situ Remediation
of Arsenic Impacted Groundwater
KlJI
i * '
EPA/600/R-19/102 j September 2019 | www.epa.gov/research
" #rx* ft' t. -Vi* •' fa-
Office of Research and Development
Center for Environmental Solutions & Emergency Response | Groundwater Characterisation & Remediation Division
-------
AUTHORS
U.S. Army Corps of Engineers
Philadelphia District
Tricia North - Philadelphia, PA
Lily Sehayek - Philadelphia, PA
U.S. Environmental Protection Agency
Office of Research and Development
Richard W kin - Ada, OK
Diana Cutt - New York, NY
Region 2
Nica Klaber - New York, NY
Hunter Young* - New York, NY
*Currently with EPA Region 10
ACKNOWLEDGMENTS
The authors would like to thank the following individuals for their contributions to the project:
Dr. Anthony J. Bednar, Laura Bittner, J. Mark Chamberlain, Dr. Juan C. Corona, Dr. Brian Dempsey,
Molly Sexton, Tony Lee, and Kathy Tynsky.
-------
CONTENTS
Executive Summary
1.0 Introduction
2.0 Site Characterization
2.1 Background......
2.2 Soil Sampling
2.2.1 Soil Characterization - USAGE, ERDC-EL
2.2.2 Soil Characterization - EPA/ORD
2.3 Groundwater Characterization
2.4 Biotic Oxidation of As(lli)
2.5 Summary of Key Findings and Geochemical Modeling Approach
2.6 Quality Control/Quality Assurance...............................................
3.0 Bench Scale Tests....
3.1 Background
3.2 Laboratory Tests and Results
3.2.1 Time to Equilibrium Tests
3.2.2 Sorption/Reaction Tests
3.2.3 Column Test
3.3 Model Simulations
3.3.1 Model Calibration to Batch and Column Test Data
3.3.2 Model Simulations of Field Conditions
3.4 Sources of Uncertainty
4.0 Summary and Conclusion
-------
FIGURES
Figure 1.1 Geochemical zones arid dissolved arsenic concentrations 1-3
Figure 2.1 Conceptual site model 2-2
Figure 2.2 Soil sampling locations ASB-1 and ASB-2 2-3
Figure 2.3 Percentage of sample mass recovered in the <425-micron size fraction 2-6
Figure 2.4 X-ray diffraction scans for samples from the ASB-1 and ASB-2 cores 2-8
Figure 2.5 FUR spectra of samples ASB-1 27-28', ASB-1 28-29', ASB-2 26-27', and
ASB-2 30-31' 2-9
Figure 2.6 Concentrations of total As (AsT) and As extracted with 0.1 M HCI and
concentrations of extracted As present as As(lll) and As(V) 2-10
Figure 2.7 SEM images and EDX maps of particles in samples ASB-1 27-28 and
ASB-2 31-32 2-12
Figure 2.8 Pre and post air sparging dissolved iron and arsenic concentrations 2-14
Figure 3.1 Soil and groundwater collection locations. 3-2
Figure 3.2 Time to equilibrium batch test results: Dissolved arsenic vs. time 3-5
Figure 3.3 Time to equilibrium batch test results: Dissolved iron vs. time 3-6
Figure 3.4 Sorption/reaction batch test results: Sand mixed with groundwater
without pH adjustment.. 3-10
Figure 3.5 Sorption/reaction batch test results: Red/orange sand mixed with groundwater
without pH adjustment........ 3-11
Figure 3.6 Sorption/reaction batch test results: Sand mixed with groundwater with pH
adjustment.. 3-12
Figure 3.7 Sorption/reaction batch test results: Red/orange sand mixed with groundwater
with pH adjustment 3-13
Figure 3.8 Column tests 3-15
Figure 3.9 Column tests results: Influent and effluent arsenic and iron concentrations.... 3-17
Figure 3.10 Column tests results: Influent and effluent arsenic speciation 3-18
Figure 3.11 Column tests results: Parameters measured in the laboratory... 3-20
Figure 3.12 Sorption/reaction batch tests results: Comparison of modeled to measured
iron and arsenic concentrations 3-23
Figure 3.13 Column tests: Comparison of modeled and measured results 3-25
Figure 3.14 Simulated iron concentration: Pre-air sparge conditions 3-27
-------
TABLES
Table 2,1 XRF soil analysis 2-4
Table 2.2 EPA 6000/7000 series soil analysis 2-5
Table 2.3 Sequential extraction results 2-5
Table 2.4 Metals concentrations (mg/kg) in samples from locations ASB-1 and ASB-2 2-7
Table 2.5 Concentrations of As and Fe (mg/kg); solid-phase As speciation (mg/kg), and
recovery (%) in selected chemical extracts 2-10
Table 3.1 Soils used in batch and column tests 3-1
Table 3.2 Time to equilibrium test results 3-4
Table 3.3 Sorption/reaction batch test results using sand with a low iron content 3-8
Table 3.4 Sorption/reaction batch test results using sand with a high iron content 3-9
Table 3.5 Sorption/reaction batch test results using sand with a low iron content and pH
adjustment. 3-9
Table 3.6 Sorption/reaction batch test results using sand with a high iron content and pH
adjustment.. 3-9
Table 3.7 Column test analytical results 3-16
Table 3.8 Column test laboratory measured parameters 3-19
DISCLAIMER
This document has been reviewed in accordance with U.S. Environmental Protection Agency policy and approved for publication.
Any mention of trade names or commercial products does not constitute endorsement or recommendation for use.
iv
-------
EXECUTIVE SUMMARY
The Vineland Chemical Company operated from 1949 to 1994 producing arsenical herbicides and
fungicides. A Record of Decision (ROD) signed in 1989 addressed remedial actions at the site which
included a pump and treat (P&T) system to remediate arsenic in groundwater. A Remediation System
Evaluation (RSE) optimization study conducted in 2010, ten years after the onset of the P&T system,
found that the current P&T system was unlikely to restore the aquifer within a reasonable time period as
specified by the ROD. Because the annual cost for the P&T system is very high, the study listed several
recommendations designed to optimize or replace the P&T system, including in-situ remediation for
arsenic immobilization. The challenges of operating the P&T system at the Vineland Chemical Company
site created an opportunity to find a sustainable in-situ approach to remediate arsenic and identify the
key processes involved.
A large-scale pilot air sparge system was developed in response and began operation in 2015. After
several years of operating the pilot system, results showed that arsenic in iron-rich groundwater
was immobilized successfully. Key processes and parameters controlling arsenic immobilization were
determined through a novel approach that combined bench-scale tests, geochemical modeling, and
groundwater and soil characterization. Results from these tests can be used to optimize the design and
operation of the fullscale system and provide guidance for the design of air sparge systems at other sites
with arsenic-impacted iron-rich groundwater at variable redox conditions.
Field data from the large-scale air sparge system show that arsenic and iron were reduced from levels
around 1,000 and 15,000 ng/Lto levels as low as 10 and 1,000 ng/L, respectively, in specific areas. Key
processes that account for arsenic immobilization in groundwater include:
• Oxidation and precipitation of iron to amorphous hydrous ferric oxide (HFO) - This process
was modeled as a function of dissolved oxygen (DO) and pH using a non-equilibrium kinetic rate
equation. Iron precipitation does not occur instantaneously at the pH range of 5.5 to 6.5
encountered at this site. Precipitation of HFO occurred downgradient of the air sparge wells
prior to reaching compliance points, reducing the need for well maintenance due to clogging.
• Sorption of arsenic to amorphous HFO and iron in soil - This process was modeled using the
surface complexation model available in PHREEQC. The majority of arsenic was immobilized by
the freshly oxidized iron with a small fraction of arsenic immobilized by iron in soil. Soil and
groundwater characterization demonstrated that oxidation of arsenic was not required for
arsenic immobilization because the reduced form of arsenic was found to be the dominant
species in both the aqueous and solid (immobilized) phases in the area of the pilot study.
• Degassing of C02 - The pH of groundwater is controlled by opposing processes: oxidation
of iron and degassing of C02. Iron oxidation decreases pH and air sparging reduces C02
concentrations which lowers the concentration of carbonic acid and results in a pH increase.
Identifying the processes responsible for arsenic immobilization was an important factor in the
sustainable operation of the air sparge system. Air sparging for arsenic immobilization can be applied
to other sites where iron is present in groundwater in sufficient quantities, and a similar procedure of
groundwater and soil characterization combined with bench-scale testing and modeling can be applied
to identify the parameters most influential on pH and iron oxidation rate.
V
-------
SECTION 1
Introduction
The presence of arsenic in water and soil is a global concern since it is identified as a carcinogen and
presents a serious threat to human health. Despite the rise in scientific research in this area, remediation
of arsenic contamination in groundwater is mostly done by ex-situ methods, whereas more sustainable
in-situ methods often do not receive consideration.
The ongoing remedial activities at the Vineland Chemical Company Superfund Site in New Jersey
provided a unique opportunity to address arsenic contamination in groundwater using an in-situ
approach. The Vineland Chemical Company operated from 1949 to 1994 producing arsenical herbicides
and fungicides, resulting in the contamination of soil and groundwater. A ROD signed in 1989 addressed
remedial actions at the site, which included a pump and treat (P&T) system to remediate arsenic in
groundwater. An RSE optimization study conducted in 2010, ten years after the onset of the P&T system,
found that the current P&T system was unlikely to restore the aquifer within a reasonable time period
as specified by the ROD. Because the annual cost for the P&T system is high, the study listed several
recommendations designed to optimize or replace the P&T system, including in-situ remediation for
arsenic immobilization. Air sparging was ultimately selected to immobilize arsenic in-situ after a series of
bench scale treatability studies were conducted (Sehayek et al., 2014b).
1-1
-------
Successful in-situ remediation of arsenic contaminated sites requires a thorough understanding of the
factors influencing arsenic transport as well as the ability to predict the behavior of arsenic in soils and
aquifer systems under future conditions (Bundschuh and Bhattacharya, 2014). The overall goal of this
study is to understand and, when possible, quantify the processes controlling the fate and transport of
arsenic under ambient, as well as, under redox conditions created downgradient of an air sparge system.
Previous testing has shown that the ambient geochemical composition of groundwater along the arsenic
plume can be divided into three general categories (Sehayek et al., 2014a).
1. Anoxic (with DO < 1 mg/L), arsenic-impacted and iron-rich groundwater located along the
leading edge of the arsenic plume that discharges to ponds and a stream (the Blackwater
Branch),
2. A mixture of oxic groundwater and the anoxic arsenic-impacted iron-rich groundwater located
in the middle of the arsenic plume along the segment where it partially discharges to a surface
water body, and
3. Arsenic impacted iron deficient oxic/anoxic groundwater located near the source.
The areas of the site where these three geochemical zones exist are shown on Figure 1.1 along with the
arsenic plume in (a) the lower portion of the surficial aquifer (well depths ranging between 25 and 75 ft
below ground surface) and (b) the upper portion of the surficial aquifer (well depths ranging between
13 and 35 ft below ground surface). The mechanisms controlling the fate and transport of arsenic could
be different within each of these zones, but changes in the project limited the study area to the leading
edge of the arsenic plume where conditions prior to air sparging were anoxic and iron-rich (category 1
above). Because the air sparge pilot was located along the leading edge of the arsenic plume, this study
addressed the overall objective of qualifying and quantifying processes taking place under ambient
conditions and under air sparge conditions.
The following tasks were performed to accomplish the goals of the study:
Task 1-Quantification of Abiotic Processes-This task included characterization of groundwater
and soil, batch bench-scale tests, and development of a quantitative tool that accounts for arsenic
partitioning to the solid phase using the geochemical model PHREEQC. Soil and groundwater for
characterization and bench-scale testing were collected from locations on the site that reflected the
geochemical conditions of interest along the leading edge of the arsenic plume.
Task 2 - Determine whether biotic oxidation of arsenite (As(lll)) is taking place - This task included
determining whether the genes aioA (aerobic arsenite oxidation) and/or arxA (anaerobic arsenite
oxidation) involved in As(lll) oxidation can be detected in groundwater or soil samples obtained from
the Vineland Chemical Company Superfund site. Task 2 was implemented using soil and groundwater
samples from the leading edge of the arsenic plume.
Task 3 - Fate and transport of arsenic - This task included performing column tests to determine
whether the quantitative tool developed in Task 1 can be used to conservatively predict the fate and
transport of arsenic in the columns and subsequently applied to predict the fate and transport of arsenic
in the field.
These tasks were accomplished as part of ongoing remedial actions and supported by EPA/ORD funding
through the Superfund Technology Liaison (STL) Extramural funding program. This report summarizes
the work performed by the U.S. Army Corps of Engineers (USACE), the Environmental Protection Agency
(EPA) Office of Research and Development (ORD), and EPA Region 2.
-------
1. Leading Edge
of Plume
2. Middle
of Plume
3. Primary
Arsenic
Source Area
Legend
+ Air SDarae Wells
4 Monitoring Wells
Dissolved Arsenic
Concentration (|jg/L)
^Source: Esri. DigitalGIobe. GeoEyl
IDS. U.SDA. USGS, AercGRID. IGNJ
10
50
100
1000
General direction of
groundwater flow
ll. Leading Edge
of Plume
2. Middle
of Plume
3. Primary
Arsenic
Source Area
jrce: Esri, DigitalGIobe-, Ge o Ey.6fSgoflBaij.
USDA USGS. AeroGRID, IGN and t
Figure 1.1 Geochemical zones and dissolved arsenic concentrations: (a) lower portion of surficial aquifer;
(b) upper portion of surficial aquifer.
1-3
-------
2.1 Background
The geology at the Vineland Chemical Company Site consists of an upper sand unit that has been
conceptually divided into an upper zone and lower zone known at the site as the shallow aquifer and
mid-depth aquifer. In many locations, iron staining and iron-cemented sands were noted at the base of
the mid-depth aquifer. A banded zone consisting of interbedded clays and silts lies underneath the mid-
depth aquifer. Below the banded zone is an oxic zone known as the middle sand unit. A schematic of the
conceptual site model is shown in Figure 2.1.
Oxic water from the middle sand unit upwells to the mid-depth aquifer in a portion of the site where the
silts and clays of the banded zone are discontinuous or missing. The oxic water dissolves pyrite minerals
that are found at the bottom of the mid-depth aquifer and in the banded zone, where present, resulting
in groundwater that has a high dissolved iron concentration and a low pH. At times, the hydraulic
conditions at the site result in oxic conditions after water from the middle sand unit has mixed with
water from the upper sand unit. These oxic conditions cause iron to precipitate out of solution, forming
iron minerals that have been detected during previous field investigations (Sehayek et al., 2014a).
2-1
-------
2-2
Shallow Aquifer -
Mid-Depth Aquifer-
Sanded Zone (silty clay unit)
is located ~ 30 - 60 feet
below ground surf ace
Upper Sand
Goethite &Hematite
Pyrite
Banded Zone
m
»w
Oxic Water in Middle Sand
Recharge Upper Sand Sand
—
Middle Sand
K'5!w
wr
mmm
iV//A AVW^V< AV/
-------
2.2 Soil Sampling
A soil sampling program was implemented in an attempt to gain direct evidence of the minerals that
are formed under in-situ air sparge conditions. Soil samples were collected from two different locations
along the downgradient edge of the groundwater plume; one location was hydraulically upgradient of
the air sparge area (A5B-1 shown on Figure 2.2) and one location was impacted by air sparging (ASB-
2 shown on Figure 2.2). Soil boring ASB-1 was completed next to piezometer PZ-9 where the arsenic
concentration in groundwater is approximately 1,000 ng/L. Soil boring ASB-2 was completed next to
piezometer PZ-10 where air sparging has decreased the arsenic concentration in the groundwater from
approximately 1,000 [ig/Lto 10 jig/L. In addition to the decrease in groundwater arsenic concentration,
the impact of the air sparging was demonstrated by changes in iron minerology.
¦i-m
W:
j&r\
-^-Air Sparge Well
A Monitoring Well
O Soil Boring Location
Groundwater Flow Direction
(approximate)
Figure 2.2 Soil sample locations ASB-1 and ASB-2.
Beginning at a depth of approximately 25 ft below land surface, XRF was used to approximate the
arsenic and iron concentrations of the soil collected from each boring at one-foot intervals. XRF results
for arsenic and iron concentrations in soil for ABS-1 and ABS-2 are summarized in Table 2.1. Soil sample
intervals with the highest arsenic concentrations in each boring (four from ASB-1 and three from ASB-
2, as indicated on Table 2.1) were selected for additional characterization. Each of the seven one-foot
sample intervals selected for additional characterization was split in half. One half was shipped to the
USAGE Engineer Research and Development Center Environmental Laboratory (ERDC-EL), and the other
half was shipped to the Environmental Protection Agency Office of Research and Development (EPA/
ORD). Soil tests performed by ERDC-EL and EPA/ORD and the results are summarized in Sections 2.2.1
and 2.2.2, respectively.
-------
Table 2.1 XRF soil analysis
Location
Identification
Soil Description
Date
Collected
Depth
Interval
Sample
Type
Collected
XRF Result (Arsenic, ppm)2
XRF Result (Iron, %)
(ft bgs)
XRF
only
XRF
and
LAB1
Result
RUN 1
+/"
Result
RUN 2
+/"
Result
RUN 3
+/"
AVERAGE
Result
RUN 1
+/"
Result
RUN 2
+/"
Result
RUN 3
+/"
AVERAGE
ASB-1
sand with gravel
4/6/2016
25.0-26.0
X
15
2
12
2
10
2
12
0.19
0.03
0.16
0.03
0.16
0.03
0.17
ASB-1
sand with clay
4/6/2016
26.0-27.0
X
193
5
230
5
142
4
188
4.10
0.10
4.70
0.10
2.80
0.10
3.87
ASB-1
iron-stained sand (between clay layers)
4/6/2016
27.0-28.0
X
321
6
364
7
443
8
376
4.91
0.10
6.39
0.06
8.36
0.08
6.55
ASB-1
iron-stained sand with clay
4/6/2016
28.0-29.0
X
348
7
408
7
297
6
351
4.53
0.10
6.03
0.06
3.46
0.10
4.67
ASB-1
iron-stained sand
4/6/2016
29.0-30.0
X
186
4
186
4
172
4
181
1.93
0.10
1.85
0.10
1.68
0.10
1.82
ASB-1
iron-stained silty/wet sand
4/6/2016
30.0-31.0
X
126
4
108
3
105
3
113
1.55
0.10
1.27
0.10
1.15
0.10
1.32
ASB-1
iron-stained sand
4/6/2016
31.0-32.0
X
52
3
41
3
46
3
46
1.01
0.10
0.87
0.10
1.05
0.10
0.98
ASB-1
clay
4/6/2016
32.0-33.0
X
37
3
32
3
49
3
39
2.22
0.10
2.52
0.10
2.42
0.10
2.39
ASB-1
slightly stained sand with clay
4/6/2016
33.0-34.0
X
< 6.0
-
< 6.0
-
< 6.0
-
<6.0
1.84
0.10
1.00
0.10
0.72
0.08
1.18
ASB-1
slightly stained sand
4/6/2016
34.0-35.0
X
7
2
< 6.0
-
< 6.0
-
< 4
0.76
0.08
0.60
0.07
0.43
0.06
0.60
ASB-1
slightly stained sand
4/6/2016
35.0-36.0
X
< 6.0
-
< 5.4
-
< 5.6
-
< 5.7
0.38
0.20
0.48
0.06
0.29
0.04
0.38
ASB-1
slightly stained sand
4/6/2016
36.0-37.0
X
< 5.7
-
< 5.7
-
< 5.9
-
< 5.8
0.38
0.05
0.41
0.05
0.41
0.05
0.40
ASB-1
clay
4/6/2016
37.0-38.0
X
14
2
7
2
8
2
10
4.13
0.10
2.19
0.10
0.95
0.10
2.42
ASB-1
sand
4/6/2016
38.0-39.0
X
< 5.5
-
< 5.6
-
< 5.4
-
< 5.5
0.33
0.04
0.43
0.05
0.38
0.05
0.38
ASB-1
sand with clay
4/6/2016
39.0-40.0
X
< 6.0
-
6
2
< 6.0
-
< 4
1.42
0.10
1.42
0.10
0.8625
0.09
1.23
1
ASB-2
sand with gravel
4/5/2016
25.0-26.0
X
33
2
30
2
24
5
29
0.31
0.10
0.28
0.10
0.31
0.10
0.30
ASB-2
sand with gravel
4/5/2016
26.0-27.0
X
137
4
140
5
179
5
152
4.12
0.10
4.80
0.10
4.01
0.10
4.31
ASB-2
sand
4/5/2016
27.0-28.0
X
40
4
47
3
51
3
46
1.16
0.10
1.51
0.10
1.44
0.10
1.37
ASB-2
sand
4/5/2016
28.0-29.0
X
54
3
50
3
44
5
49
1.07
0.10
0.86
0.10
1.02
0.10
0.98
ASB-2
sand
4/5/2016
29.0-30.0
X
93
3
89
3
102
4
95
2.25
0.10
1.71
0.10
1.53
0.10
1.83
ASB-2
slightly stained sand
4/5/2016
30.0-31.0
X
118
4
179
4
73
3
123
2.21
0.10
2.91
0.10
1.47
0.10
2.20
ASB-2
iron-stained sand
4/5/2016
31.0-32.0
X
150
5
81
3
104
4
112
2.25
0.10
1.16
0.10
1.53
0.10
1.65
ASB-2
iron-stained sand
4/5/2016
32.0-32.5
X
87
3
76
3
101
3
88
1.19
0.10
1.08
0.10
1.58
0.10
1.28
ASB-2
silty sand
4/5/2016
32.5-33.0
X
27
3
25
3
27
2
26
1.03
0.10
1.01
0.10
1.02
0.10
1.02
ASB-2
sand
4/5/2016
33.0-34.0
X
10
2
12
2
8
2
10
0.48
0.06
0.71
0.08
0.65
0.07
0.62
ASB-2
sand
4/5/2016
34.0-35.0
X
< 7.0
< 5.8
< 7.0
<6.6
0.94
0.10
0.94
0.10
1.45
0.10
1.11
ASB-2
wet sand
4/5/2016
35.0-36.0
X
< 5.4
< 5.6
< 5.5
< 5.5
0.51
0.06
0.58
0.06
0.62
0.07
0.57
1Soil samples sent to the USACE's Engineering Research and Development Center (ERDC) laboratory in Vicksburg, Mississippi for analysis
2For instances where an interval had both detected and non-detect results between runs, half the detection limit was used for the average calculation
ft bgs = feet below ground surface; N/A= not applicable; ppm = parts per million; XRF = X-ray fluorescence
-------
2.2.1 Soil Characterization - USACE, ERDC-EL
Soil characterization at USACE, ERDC-EL included metal analysis using EPA 6000/7000 series method,
XRD, and sequential chemical extraction.
Arsenic and iron concentrations in soil determined by EPA 6000/7000 series method are provided in
Appendix A and summarized in Table 2.2. The ranges in arsenic and iron concentrations of soil samples
collected from ASB-1 were 170 to 567 mg/kg, and 21,600 to 50,200 mg/kg, respectively. The arsenic and
iron contents at sample location ASB-2 were slightly lower, with arsenic concentrations ranging from
83.9 to 94.7 mg/kg and iron concentrations ranging from 10,100 to 22,000 mg/kg. The results from the
laboratory analysis were within reasonable agreement to the concentrations determined in the field
using XRF (see Table 2.1).
Table 2.2 EPA 6000/7000 series soil analysis
Location
Identification
Soil Description
Date Collected
Depth Interval
As
(mg/kg)
Fe
(mg/kg)
(ft bgs)
ASB-1
sand with clay
4/6/2016
26.0-27.0
170
22600
ASB-1
iron-stained sand
4/6/2016
27.0-28.0
519
50200
ASB-1
iron-stained sand with clay
4/6/2016
28.0-29.0
567
45300
ASB-1
iron-stained sand
4/6/2016
29.0-30.0
229
21600
1 1
ASB-2
sand with gravel
4/5/2016
26.0-27.0
94.7
22000
ASB-2
slightly stained sand
4/5/2016
30.0-31.0
88.3
11800
ASB-2
iron-stained sand
4/5/2016
31.0-32.0
83.9
10100
As = Arsenic; Fe = Iron; ft bgs = feet below ground surface; mg/kg = milligram per kilogram
Sequential extraction was performed by USACE, ERDC-EL to determine the fraction of arsenic and iron
in the soluble, exchangeable, carbonate, Fe-Mn oxides, organic matter and sulfide, and residual forms.
Results from sequential extractions are provided in Appendix A and summarized in Table 2.3. The
analyses showed that a negligible amount of arsenic was soluble, and the majority of the arsenic and
iron were associated with the oxide and the stable residual components. Arsenic was not detected under
the exchangeable and organic matter and sulfide extractions.
Table 2.3 Sequential extraction results
Location
Identification
Depth Interval
Carbonate
Oxide
Residual
Soluble
(ft bgs)
As
(mg/kg)
Fe
(mg/kg)
As
(mg/kg)
Fe
(mg/kg)
As
(mg/kg)
Fe
(mg/kg)
As
(mg/kg)
Fe
(mg/kg)
ASB-1
26.0-27.0
7.02
7.95
44.8
1150
33.4
4530
2.92
54.4
ASB-1
27.0-28.0
19.4
20.9
104
923
62.9
6780
8.18
25.9
ASB-1
28.0-29.0
24.3
10.8
92.9
1270
108
8750
10
5.58
ASB-1
29.0-30.0
13.7
9.78
74.9
524
38.6
4400
5.31
24
1 1
ASB-2
26.0-27.0
0.665
10.9
6.42
817
18
4250
ND
34.6
ASB-2
30.0-31.0
0.636
7.46
6.15
412
14.9
2090
ND
35.9
ASB-2
31.0-32.0
3.66
11.2
23.2
427
14.1
1850
1.4
78.8
As = Arsenic; Fe = Iron; ND = Not detected; ft bgs = feet below ground surface; mg/kg = milligram per kilogram
2-5
-------
XRD results are provided in Appendix B. Results indicate that the most frequently encountered form of
iron mineral is goethite. The less stable form of iron mineral, ferrihydrite, also defined in some places
as the dried form of amorphous ferric hydroxide (i.e. hydrous ferric oxides or HFO), was encountered
only in one interval located in the air sparge area (specifically boring ASB-2 interval 26-27'). The limited
detection of ferrihydrite in the current and historical field investigations and the high frequency and
quantity, expressed by percent weight, of the more thermodynamically stable and more crystallized
goethite suggests that iron oxidation results in the formation of amorphous ferric oxide which transforms
over time to the more stable goethite or other iron minerals.
Hematite, was detected by USACE, ERDC- EL only in one soil sample. EPA/ORD detected hematite in all
soil samples. The discrepancy between EPA/ORD and the USACE, ERDC-EL can be attributed to the faster
scan speed used by USACE, ERDC-EL and the lower relative mass of hematite as compared to goethite.
2.2.2 Soil Characterization - EPA/ORD
Soil characterization at EPA/NRMRL/
GWERD included XRD; Fourier-
transform infrared spectroscopy (FTIR);
scanning electron microscopy (SEM);
energy dispersive X-ray spectrometry
(EDX); and metals analysis using
microwave digestion, targeted
chemical extractions, and arsenic
speciation using liquid chromatography
(LC) coupled on-line to ICP-mass
spectrometry (LC-ICP-MS).
Samples from locations ASB-1 and
ASB-2 were prepared in an anaerobic
glovebox (Coy Inc.) with a maintained
N2 plus 5% H2 atmosphere. Samples
were allowed to dry for two weeks
within the glovebox at ambient
conditions aided by alumina desiccant
plates. The dried samples were
disaggregated using an agate mortar
and pestle and sieved to recover
aquifer particles in the <40 mesh size
range (<425 micron). For each sample,
mass recovery in the <40 mesh sieve is
shown on Figure 2.3.
tfl 50
so
cN
(/I
V)
re
£ 30
3
tn 20
CM
<3-
V 10
Figure 2.3 Percentage of sample mass recovered in the
<425 micron size fraction.
2-6
-------
Solid-phase concentrations of metals in the bulk and <40 mesh size fraction were determined using
microwave-assisted digestion and inductively coupled plasma - optical emission spectrometry (ICP-OES)
following EPA Method 200.7. A certified reference material from ERA (Golden, CO; cat. #540) was analyzed
along with the sample batch and all QC performance acceptance limits were met, except for uranium
and silver. For these elements, mass concentrations determined were low by up to 10% compared to the
certified values. Solid-phase concentrations for selected elements are provided in Table 2.4. In all cases
where an analyte was detected, concentrations were greater in the fine-grained fraction (<425 micron)
compared to the bulk sample, which is expected. Across all of the samples, iron concentrations ranged
from about 0.55 to 7.37 wt%; arsenic concentrations ranged from 53 to 465 mg/kg. Overall, these metals
concentrations are in reasonable agreement with results of XRF and digestion analyses presented in Tables
2.1 and 2.2.
Table 2.4 Metals concentrations (mg/kg) in samples from locations ASB-1 and ASB-2
Concentrations in the bulk and fine (<40 mesh) fractions are designated as b and f, respectively
ASB-1
ASB-2
26-b
26-f
27-b
27-f
28-b
28-f
29-b
29-f
26-b
26-f
30-b
30-f
31-b
31-f
Al
2855
6725
2224
7600
4870
7417
1300
1386
324
6350
434
855
324
960
Ca
12 J
25 J
9 J
30 J
18 J
27 J
5 J
6 J
46
79
<40 U
8 J
<40 U
8 J
Mg
62
117
21J
58
126
154
12 J
14 J
57
84
<40 U
5 J
<40 U
8 J
Fe
24690
44890
27190
73660
26740
36970
13350
14310
19150
32460
5800
9350
5520
14390
Mn
<8 U
<8 U
4 J
20
<8 U
<8 U
<8 U
<8 U
<8 U
<8 U
<8 U
<8 U
<8 U
<8 U
As
201
357
311
465
309
389
159
174
65
95
53
87
58
146
Pb
<8 U
11
<8 U
15
<8 U
11
<8 U
<8 U
<8 U
10
<8 U
<8 U
<8 U
<8 U
Cd
3 J
6
3 J
8
3 J
4
1J
2 J
2 J
4
<4 U
1J
<4 U
2 J
Ni
<4 U
1J
<4 U
1J
<4 U
<4 U
<4 U
<4 U
<4 U
2 J
<4 U
<4 U
<4 U
1J
V
7
13
6
13
18
22
6
6
31
47
2
4
2
5
Concentrations determined using microwave-assisted sample digestion and ICP-OES analysis following EPA Method 200.7.
Data qualifiers: J, analyte was detected above the method detection limit (MDL) but below the quantitation limit (QL). U, analyte was not detected.
XRD analyses were carried out using a Rigaku Miniflex II diffractometer with Mn-filtered FeKa radiation
(X = 0.1937 nm). Diffraction data were collected from 5° to 90° 20 with 0.01° 20 step increments at a scan
rate of 6 s per step. NIST 640b standard reference material (silicon powder) was periodically scanned
as a quality control check of d-spacing accuracy. Data collected from the XRD scans were imported into
the Jade (Materials Data, Inc.) software package for analysis and matched to the Powder Diffraction File
Data Base (PDF, International Centre for Diffraction Data). XRD scans for all core samples (<40 mesh size
fraction) are shown on Figure 2.4. The dominant mineral components identified were quartz, kaolinite,
goethite, and hematite. Similar results were obtained by USACE, although hematite was less frequently
identified in the samples. This finding is attributed to the fact that EPA/ORD used the <40 mesh size
fraction in their analysis and adopted a slower scan rate in the XRD analysis to improve sensitivity.
Ferrihydrite (hydrous ferric oxide) was not detected in the XRD scans, but identification of this poorly
crystalline material in the presence of other strongly diffracting sample components is notoriously
difficult. A further attempt was made by ultra-sounding the <40 mesh size fraction in methanol and
collecting the dispersed particulates in suspension. Again, quartz, kaolinite, goethite, and hematite were
detected (results not shown), but kaolinite was more abundant and illite was identified in several samples
(ASB-1 28-29; ASB-2 26-27; ASB-2 30-31).
2-7
-------
h
V.
< c
*•——
3 C
5
,
H
A il
/
/
\SB-2 26-27'
^SB-2 30-31'
^SB-2 31-32'
i
i i
Lit
.1
i
l ¦ i ii
A. i
1
—r
20
r*
i
40
1 1
60
«- |
80
ASB-1 26-27'
ASB-1 27-28'
ASB-1 28-29'
ASB-1 29-30'
20 40 60 80
2-Theta, degrees
Note: The dominant minerals identified in the aquifer solids were
quartz (Q), kaolinite (K), goethite (6), and hematite (H).
Figure 2.4 X-ray diffraction scans for samples from the ASB-1 and ASB-2
cores (<40 mesh size fraction).
2-8
-------
FTIR spectra were collected with a Bruker Vertex 70 spectrometer. Samples were prepared as pellets
with KBr (weight ratio 195 mg KBr to 5 mg sample). Samples were scanned 32 times from wavenumber
400-4000 cm1. As a quality control check, calcium carbonate was scanned and results were compared
with spectral data from Adler and Kerr (1962). FTIR spectra for selected samples are shown in Figure 2.5
and results are consistent with XRD analyses in showing the presence of kaolinite and goethite in the
samples. Arsenic sorbed onto iron oxide surfaces typically shows an absorption band at about 1617 cm 1
(Hsia et al., 1993). No absorption features could be attributed to arsenic in the measured spectra which
indicates that the resolution of the method requires higher arsenic surface loadings.
ASB-2 30-31'
CD
O
C
03
-Q
ASB-2 26-27
ASB-1 28-29'
ASB-1 27-28"
4000
3200
2400
Wavenumber, cm"1
1600
800
Note: The spectra clearly show absorption bands indicative of kaolinite and goethite. Features at
wavenumbers <1200 cm-1 may be contributed by several minerals including quartz, kaolinite, and
goethite. Absorption bands characteristic of arsenic on iron oxides were not detected in the samples.
Figure 2.5 FTIR spectra of samples ASB-1 27-28', ASB-1 28-29', ASB-2 26-27', and ASB-2 30-31'.
2-9
-------
More sensitive chemical
extraction tests were conducted
under anaerobic conditions
using deoxygenated water and
0.1 M HCI. Water extractions
were expected to dissolve the
most labile forms of arsenic
in the samples and 0.1 M HCI
was expected to extract poorly
crystalline hydrous ferric oxides
and associated arsenic (e.g., Kostka
and Luther, 1994). Note that dilute
HCI is not expected to dissolve
goethite and/or hematite, but will
dissolve hydrous ferric oxide. Solid-
phase concentrations of arsenic
measured are shown on Figure 2.6
and Table 2.5. In core ASB-1, water-
extractable arsenic ranged from 23
to 119 mg/kg, representing about 9
to 25% of the total arsenic; in core
ASB-2, water-extractable arsenic
ranged from 2 to 12 mg/kg, or
about 2 to 10% of the total arsenic.
Solid-phase arsenic concentrations
determined using 0.1 M HCI ranged
from 44 to 881 mg/kg, and, in
most cases, the majority of the
arsenic was recovered using 0.1
M HCI. This finding suggests that
much of the solid-phase arsenic is
associated with poorly crystalline
iron-bearing materials.
Table 2.5 Concentrations of As and Fe (mg/kg), solid phase As speciation (mg/kg), and recovery (%) in
selected chemical extracts
Sample
As
Total
As
h2o
As
0.1 M HCI
As
Recovered
As(lll)
h2o
As(lll)
0.1 M HCI
As(V)
0.1 M HCI
Fe
Total
Fe
H2O + 0.1MHCI
Fe
Recovery
mg/kg
mg/kg
mg/kg
%
mg/kg
mg/kg
mg/kg
wt%
wt%
%
ASB-1 26-27
357
34
276
87
58
224
20
4.49
0,29
6.4
ASB-1 27-28
465
119
881
215
120
663
54
7.37
0.71
9.6
ASB-1 28-29
389
38
317
91
61
227
30
3.70
0.30
8.1
ASB-1 29-30
174
23
97
69
25
88
10
1.43
0,06
4.2
1 1
ASB-2 26-27
95
1.5
44
48
12
7.9
22
3.25
0.19
5.8
ASB-2 30-31
87
9,1
52
70
5.0
28
4.8
0.94
0.07
7.4
ASB-2 31-32.
182
12
62
41
7.6
46
7,3
1.84
0.09
4.9
2-10
1200
1000
800
GC
E
600
i/i
<
400
200
0
h
in
I..
*sa *sg 4S-&,
la. Ili_
¦ AsT ¦ As, 0.1M HCI + H20 ¦SumAs(lll)
Sum As(V)
8
7
6
3? 5
S 4
oT
3
2
1
0
I
I
4%, ^e., ^s., "4%,
-------
The sequential extraction testing performed by ERDC (Section 2.2.1) similarly found that a significant portion of
the arsenic was associated with iron oxides, but those results also suggested that a notable fraction of the arsenic
remained in the stable residual form. It is unknown if the discrepancy between the results proposed by EPA/
ORD and the results proposed by ERDC are a result of the extraction method, sample preparation/handling (e.g.
use of bulk vs. <40 mesh size fraction), or mineralogical differences in the soil sub-samples sent to each lab. The
differences in the fraction of arsenic found to be associated with the poorly crystalline iron-bearing minerals has
implications for desorption of arsenic but should not impact the interpretation of arsenic sorption mechanisms.
The distribution of arsenate and arsenite species in the water and 0.1 M HC1 extracts was determined using liquid
chromatography (LC) coupled on-line to ICP-mass spectrometry (LC-ICP-MS; Thermo Electron Spectra HPLCi
using collision cell technology to remove spectral interferences. Samples for arsenic speciation were filtered,
acidified with HCI (pH<2; Optima, Fisher Scientific), and retained chilled in amber-plastic bottles. Chromatographic
separation of arsenic species was accomplished using a PRP-X100 guard column (Hamilton), a PRP-X100 separator
column (Hamilton), and by pumping an isocratic eluent [1.0 mL min ', 10 mM (NH4)H PO.,/NH...NO,]. Eluent was
directly aspirated into a Thermo Electron X series II ICP-MS and arsenic was detected by monitoring the m/z = 75
signal. In core ASB-1, most of the extracted arsenic (88-92%) was present as As(lll). Similar results were obtained
for samples from core ASB-2 [26-86% As as As(lll); Table 2.5], although the fraction of As as As(V) increased at
shallower depth intervals.
SEM-EDX results are shown in Fig. 2.7 for samples ASB-1 27-28 and ASB-2 31-32. In both samples, arsenic was
found at low concentrations (<2 wt%) associated with iron-rich coatings on aquifer particles. These results show
that at the micro-scale, arsenic is associated with iron and is found mainly associated with particle coatings.
o2°
©s°
-------
VEGA3 IE SCAN I
VEGA3 TESCAN
AS812728
As
d
Fe
CI
Si
-
s
Figure 2.7 SEM images and EDX maps of particles in samples ASB-1 27-28 and ASB-2 31-32.
ASB-2 31-32
ASB-1 27-28
Note: These images show that As is closely associated with Fe-rich coatings present
on the aquifer particles.
2-12
-------
2.3 Groundwater Characterization
Groundwater was sampled prior and during air sparging and analyzed for total and dissolved metals and
for arsenic speciation. Significant observations from the groundwater sampling program that took place
between the fall of 2015 and the spring of 2016 are described below.
• Prior to air sparging, the dissolved arsenic concentrations were approximately 1 mg/L
adjacent to the air sparge wells. During air sparging, the most significant decrease in arsenic
was observed approximately 15 ft downgradient of the air sparge line where dissolved arsenic
dropped to 0.007 mg/L. Arsenic concentrations in a monitoring well located approximately
40 ft downgradient of the air sparge wells dropped to 0.234 mg/L, below the alternate
concentration limit (ACL) of 0.350 mg/L stipulated in the 1989 ROD. The arsenic concentration
in the well located about 100 feet downgradient dropped from 1 mg/L to 0.388 mg/L (slightly
above the ACL).
• Prior to air sparging, dissolved iron concentrations ranged between about 15 to 20 mg/L.
During air sparging, iron concentrations declined to a range of 0.3 to 3 mg/L downgradient
of the air sparge line.
• Prior to air sparging, the groundwater was anoxic with DO concentrations less than 1 mg/L
and the oxidation reduction potential (ORP) in the arsenic-impacted area ranged between
approximately -50 and -125 mV. These DO and ORP values are indicative of iron-reducing
conditions. During air sparging, the DO concentrations increased to between 2 and 11 mg/L
and ORP increased to approximately 100 mV to 300 mV.
• Historical pH levels in the area adjacent to the air sparge wells ranged between about 4.5
and 6, with the majority of the readings between 4.5 and 5.5. In general, during air sparging,
the pH slightly increased to values between 5.5 and 6. Oxidation and precipitation of iron
is expected to result in decreasing pH; however, the pH during air sparging slightly increased.
Previous investigations at the site and geochemical modeling indicate that groundwater
at the site is supersaturated with carbon dioxide (C02). The air sparging resulted in a release
of gases into surface water bodies at distances greater than 50 feet. This resulted in
degassing of C02 and a consequent increase in pH. The source of C02 is believed to be
biological activity in the clay layer at the base of the aquifer. This layer contains high total
organic carbon (TOC) of about 100 mg/kg.
• The ORP of anoxic groundwater is often controlled by the iron chemistry. The pH and ORP
measured in four existing piezometers were plotted on an Eh-pH diagram for arsenic and iron
published by Ford et al. (2007). Measured pH/ORP data from these piezometers aligned along
the goethite/Fe2+ slope, demonstrating that ORP appears to be controlled by the iron chemistry.
This was observed throughout the site areas with elevated concentrations of dissolved iron.
• Historical investigations indicate that only the inorganic arsenic species [i.e., arsenite (As(lll))
and arsenate As(V))] are encountered in the portion of the arsenic plume that was investigated
during the pilot aeration tests. As(lll) and As(V) speciation analyses that were performed prior to
air sparging indicated that As(lll) was dominant over As(V) in the groundwater. EPA/ORD
investigations confirm that most of the arsenic extracted from the soil samples collected
upgradient of and within the air sparge areas was in the As(lll) form, similar to the chemical
extraction and speciation testing as described in Section 2.2.2. These results indicate that prior
to and after air sparging, the oxidation of As(lll) to As(V) was not occurring at a significant rate.
2-13
-------
Dissolved Arsenic
Dissolved Iron
Pre Air Sparge
After 4 Months of Air Sparging
1.07/.
0.026 7l0-254a
234 ¦
0444
After 4 Months of Air Sparging
O
EffiSl -
'O-V ^2'86WHg
* n
o
F"f|
Groundwater
Flow Direction
Air Sparge
Well
Note: This figure
shows the dissolved
arsenic and iron
concentrations
from two sampling
events that serve as
"typical" pre and
post air sparging
conditions. Minor
variations in
concentrations
have been
observed due to
seasonal variations
or variations in air
sparge operations.
Figure 2.8 Pre and post air sparging dissolved iron and arsenic concentrations.
2-14
-------
2.4 Biotic Oxidation of As(lll)
At the onset of this work, it was not clear whether oxidation of As(lll) to As(V) during air sparging was
an important process that needed to be considered when quantifying partitioning of arsenic from
groundwater to soil; therefore, part of the investigation included determining the overall physiological
potential for microbiological transformation of As(lll) in soils and groundwater at the site. Specifically,
the physiological potential toward the oxidative biotransformation of As(lll) to As(V) was evaluated by a
PCR-based survey targeting three relevant genes. Genes of interest that were targeted for PCR included
aioA and arxA, which are involved in oxidation of As(lll) to As(V), and dsrl which is a likely candidate for
providing an electron acceptor for As(lll) oxidation via dissimilatory sulfate reduction. The final report
of activities conducted under this task (Task 2) are provided in Appendix C. The conclusion of this study
was that organisms responsible for oxidation of As(lll) to As(V) by specific physiological mechanisms are
either not present or are present in extremely low concentrations. It is possible that the oxidation of
As(lll) may be occurring anaerobically via coupling of As(lll) oxidation with S04 reduction.
2.5 Summary of Key Findings and Geochemicai Modeling Approach
The soil and groundwater data were used to evaluate arsenic partitioning from groundwater to soil and,
hence, the fate and transport of arsenic in groundwater under ambient anoxic or under air sparging
aerobic conditions. Key findings include:
1) Decreases in the arsenic concentrations in groundwater coincide with decreases in iron
concentrations. Iron is oxidized, precipitates, and the precipitated iron oxide removes arsenic
from the groundwater. HFO forms first and over time transforms to more thermodynamically
stable iron minerals (goethite and hematite). Both the chemical extraction analysis performed
by EPA/ORD and the sequential extraction analysis performed by USACE ERDC-EL identified a
significant portion of the solid-phase arsenic as associated with poorly crystalline iron minerals
(i.e. ferrihydrite or HFO). The sequential extraction analysis performed by ERDC-EL also found
that a significant portion of the solid-phase arsenic was associated with the stable residual minerals
while the majority of the chemical extraction analyses performed by EPA/ORD showed that only
a small portion of the arsenic was associated with more crystalline forms of iron. Therefore, there
is some uncertainty in the site-specific conditions and timeframes under which HFO transforms into
the more crystalline forms of iron.
It has been noted in the literature that ferrihydrite transforms to goethite and to hematite under
certain conditions (Cornell and Schwertmann, 1996) and that factors affecting the rate of
transformation include:
• Temperature-Transformation of HFO to hematite has been observed at temperatures as low
as 4 °C. The rate increases with increasing temperature.
• Presence of seed crystals - Formation of goethite from HFO involves dissolution of HFO
followed by nucleation and growth of goethite in solution. This occurs naturally, since the
solubility of HFO is greater than the solubility of goethite. Goethite formation is catalyzed
by the presence of seed crystals of either goethite or hematite. There is evidence that the
ordered regions in HFO can also serve as sites for goethite crystallization.
• pH - The rate of transformation of HFO increases as the pH of the system rises from 2 to 12.
2-15
-------
2) As(lll) is the dominant form of arsenic in groundwater and in soil. Oxidation of As(lll) to As(V) in
groundwater under ambient conditions does not appear to influence partitioning of arsenic
from groundwater to soil, possibly because oxidation of As(lll) to As(V) occurs much slower than
HFO formation and arsenic partitioning to soil. It has been reported that under natural water
conditions, oxidation of As(lll) or reduction of As(V) occurs at a sufficiently slow rate such that water
samples can be collected, transported and analyzed before excessive change in species distribution
takes place (Cherry et al., 1979).
3) Oxidation of Fe(ll) is not instantaneous at the pH range encountered in the field (4.5 to 6). This
allows precipitation of iron oxides over a large volume of the aquifer and decreases the
accumulation of iron oxides in the immediate vicinity of the air sparge wells.
4) When groundwater is exposed to the atmosphere, the pH of groundwater is influenced by
both iron oxidation and precipitation (which decreases the pH) and the degassing of C02 from
the supersaturated groundwater (which increases the pH).
PHREEQC, version 3 (Parkhurst and Appelo, 1999; Parkhurst and Appelo, 2013) was selected to model
the fate and transport of arsenic under ambient and air sparge conditions. The database for PHREEQC
was evaluated against the key findings to ensure that key processes were incorporated. Reactions that
are slow under the site conditions such as pH-controlled reactions and microbial processes were not
included in the modeling effort. PHREEQC, with the Wateq4f database that addressed all key processes
outlined in this section, was used to calibrate the model to bench-scale data. Summary of key findings
and the processes used to address them are as follows:
1) Iron (Fe (II)) oxidation to HFO is not instantaneous (i.e., kinetics control the rate and extent of
reaction). Rate constants for Fe(ll) oxidation to HFO were added to PHREEQC to account for this
phenomenon. Specifically, Dietz and Dempsey (2001) rate equations and constants that were
developed and applied to oxidation of iron in acid mine drainage were added to PHREEQC.
2) Arsenic partitioning from groundwater to HFO: According to the literature, surface complexation
models can be used to quantify sorption on variably charged surfaces such as Fe oxides (Dzombak
and Morel, 1990). There are three options available in PHREEQC for modeling surface-complexation
reactions: the generalized two-layer model of Dzombak and Morel (1990); a model with an explicitly
calculated diffuse layer from Borkovec and Westall (1983); and, the non-electrostatic model of
Davis and Kent (1990). The Dzombak and Morel diffuse-layer model accounts for the dominant
anions and cations and does not account for the sorption of trace metals. This model was selected
because it has been widely used by EPA and others, successfully applied to field cases, and has an
extensive database. Surface complexation constants taken from Dzombak and Morel (1990) are
available in the PHREEQC databases phreeqc.dat and wateq4f.dat. Wateq4f.dat was selected for
the analysis since phreeqc.dat does not include arsenic.
3) As(lll) is the dominant form of arsenic in groundwater, and As(lll) in groundwater does not appear
to oxidize to As(V) prior to partitioning to soil. PHREEQC will allow the oxidation or reduction
of arsenic to proceed to equilibrium, with most arsenic oxidizing to As(V) in the presence of oxygen
and most arsenic reduced to As(lll) under anaerobic conditions regardless of the length of time it
takes this process to occur. Since DO is present in the system at concentrations < 1 mg/L under
ambient groundwater and > 1 mg/L under aerobic conditions, PHREEQC converts most of the
2-16
-------
As(lll) to As(V) prior to partitioning to HFO. For that reason, PHREEQC was modified to remove
the oxidation of As(lll) since the partitioning of As(lll) to soil takes place much more quickly than
the oxidation of As(lll) to As(V).
4) Degassing of carbon dioxide (C02) is a key reaction that influences the pH of the system. The pH of
the system is pertinent since both the rate of oxidation of Fe(ll) and partitioning of arsenic to HFO
are functions of pH. Degassing of C02 is incorporated in the model by allowing a given percent of C02
in groundwater to partition to the gas phase.
Detailed information regarding the incorporation of the above processes in the model and the use of the
model to calibrate results from the bench-scale tests and the field are provided in Section 3.
2.6 Quality Control/Quality Assurance
As required by EPA's quality assurance policy, data collection efforts and modeling studies described
in this report were conducted under approved Quality Assurance Projects Plans (QAPPs) and followed
standard quality control (QC) procedures. Sampling conducted by the USACE was completed following
procedures in the Department of Defense (DoD) Environmental Field Sampling Handbook, Revision
1.0 (April 2013). Work conducted by the USACE ERDC-EL was performed following standard method
QC procedures, including, but not limited to: blanks, blank spikes, duplicates, and matrix spikes.
Instruments were calibrated using NIST-traceable, commercially available standards with second source
NIST-traceable calibration verification standards. Recovery ranges for all QC samples followed method
guidance (e.g. 10% for ICV and CCV recoveries, 20% for BS and MS recoveries). Internal standards
were added in-line for all ICP-AES (e.g. Scandium and Yttrium) and ICP-MS (e.g. Germanium, Rhodium,
Terbium, and Bismuth) analyses to correct for instrumental drift, and palladium-magnesium or nickel
nitrate matrix modifiers were used for all GF-AAS analyses. Laboratory characterization studies
conducted by EPA's Office of Research and Development were conducted under the QAPP titled
"Monitored Natural Attenuation of Metals" (G-GWERD-0014907-QP-1-5). Data qualifiers were applied
as appropriate and are noted in the tabulated data.
2-17
-------
SECTION 3
Bench Scale Tests
3.1 Background
Laboratory tests were conducted at the Vineland Chemical Company
Superfund Site in order to investigate sorption of arsenic onto iron oxides
in native soil and when freshly precipitated from groundwater. Data
collected from the laboratory tests were used to determine if published
thermodynamic constants were adequate in describing the arsenic
partitioning between groundwater and iron oxides in the site soil as
well as iron oxides that precipitate from the groundwater. Additionally,
laboratory test data were used to calibrate the model to describe the
rate of iron oxidation using a rate equation that incorporates both
heterogeneous and homogeneous oxidation.
Iron and arsenic concentration data were collected prior to and during
the air sparging pilot test. The data were used to verify that the fate and
transport mechanisms identified during model calibration were consistent
with the field observations.
Three types of bench-scale tests were performed in the lab following EPA
(1992) guidance for batch-type procedures for estimating soil adsorption
of chemicals when relevant. These tests include:
1. Time to Equilibrium Test - Groundwater was mixed with soil and
the concentration of arsenic and iron were monitored for a period
up to 24 hours to determine the time it takes the system to come
into equilibrium.
2. Sorption/Reaction Test - Groundwater was mixed with soil at
different ratios to determine the impact of soil quantity on
arsenic sorption.
3. Column Test - Groundwater was passed through a column filled
with soil to simulate groundwater flow in the aquifer.
Each test was performed on two types of soil. The soils that were used
in the lab tests had low levels of arsenic (< 10 mg/kg), but each soil type
had a different iron content so that the role of the native iron in soil in
immobilizing arsenic could be assessed. Table 3.1 lists the arsenic and iron
content of each soil.
fable 3.1 Soils used in batch and column tests
Soil Type
Sample Interval
(ft bgs)
Arsenic
(mg/kg)
Iron
(mg/kg)
Sand
Red/Orange Sand
20-25
35-36.5
3.06-5.34
3.97-9.52
634 -1,950
9,200-10,300
-------
Both types of soil were collected from the site by sonic drilling at location MP-1 as shown on Figure 3.1.
The soil was collected prior to any air sparging activities on the site, so the subsurface in this region was
anoxic at the time of soil collection. The soil was immediately transported (<24 h) to an on-site lab in
acetate sleeves. Soil at the edges of the sleeves was removed and the soil was placed into an anaerobic
chamber. These actions were performed shortly after collection to minimize oxidation of the iron in the
soil. All batch tests were assembled and sealed inside the anaerobic chamber. Due to space limitations,
the batch test bottles were then rotated on a tumbler outside of the anaerobic chamber. Batch test
bottles were transferred back inside the anaerobic chamber before they were unsealed for sampling.
All bottles were covered with foil for the duration of the tests to minimize exposure to light. Column tests
were conducted entirely inside the anaerobic chamber. Groundwater that was used in the lab tests was
collected from MW75S, shown on Figure 3.1 using standard low flow sampling methods. Groundwater in
this location is anoxic, iron rich, impacted by arsenic, and generally had a pH between 5.20 and 6.12 and
an ORP between -6 and 74 upon sampling.
Groundwater Collection Location
MW75S
Soil Collection Location
MP-l|
>£lc earth
C 2010 Goals'
Figure 3.1 Soil and groundwater collection locations.
3-2
-------
3.2 Laboratory Tests and Results
The following sections provide a description of each laboratory test and a brief discussion of the
analytical results from the test.
3.2.1 Time to Equilibrium Tests
Time to equilibrium tests were conducted by filling four bottles with approximately 550 mL of
groundwater and 100 g soil. Another four bottles were filled with about 550 mL groundwater and
no soil to serve as the "groundwater control" so that changes in concentrations due to soil could be
differentiated from those occurring in groundwater under ambient laboratory conditions. All eight
bottles were placed in a rotary agitator and sampled after different contact times.
Four separate sets of Time to Equilibrium batch tests were conducted. The first set included sand with
a lower iron content. The second set included red/orange sand with a higher iron content. The third
and fourth sets of batch tests replicated the first two sets except that the pH of the groundwater was
adjusted to 4 using H2S04 before being introduced into the batch tests. The results of all four sets of
batch tests are listed in Table 3.2 and shown on Figure 3.2.
In batch tests that included soil, the majority of the observed decrease in dissolved arsenic occurred
within the first one to two hours of the test. Arsenic concentrations continued to decrease after two
hours, but the rate slowed. Dissolved iron was also monitored throughout the 24-hour test period
(Figure 3.3). A period of 24 hours was deemed sufficient for the subsequent sorption/reaction tests.
Some iron precipitated out of the groundwater control tests and resulted in a decrease in dissolved
arsenic. The decrease in arsenic in the groundwater control tests was always less than in the tests
involving soil. In some tests, the dissolved iron concentration increased when soil was added, but
these results did not impact the decision to continue with a 24-hr testing time. An increase in dissolved
iron concentration was also observed in some of the sorption/reaction batch tests and column tests
involving soil. The batch and column tests did not directly demonstrate the mechanism responsible
for the increase in dissolved iron, so it was assumed that either a release of adsorbed ferrous iron or a
dissolution of an iron mineral in the soil had taken place.
3-3
-------
Table 3.2 Time to equilibrium test results
Description
Date
Contact
Time
(hr)
MW75S
Groundwater
(mL)
Soil (g)
Aqueous
Concentration
(mg/L)
Measured in the lab
As
(Diss)
Fe
(Diss)
PH
ORP
(mV)
Temp
(°F)
a.) Time To
Equilibrium,
sand, no pH
adjustment
9/8/2015
0
N/A-lnitial Condition
2.19
10.5
5.78
94
9/8/2015
2
550
100
1.5
7.8
5.8
-51
68.2
9/8/2015
4
550
100
1.44
8.12
5.78
-34
69.1
9/8/2015
6
550
100
1.33
7.55
5.69
-16
71
9/9/2015
24
550
100
0.965
4.89
5.9
1
70.6
9/8/2015
2
550
0
2.12
10.4
5.85
-35
68
9/8/2015
4
550
0
2.07
10.3
5.6
-17
70.1
9/8/2015
6
550
0
1.83
10.2
5.69
-26
73.6
9/9/2015
24
550
0
1.48
8.58
5.93
-13
73.1
b.) Time To
Equilibrium,
Red/orange
sand, no pH
adjustment
11/12/2015
0
N/A-lnitial Condition
1.39
7.81
5.70
98
64.5
11/12/2015
1
550
0
1.37
7.7
5.65
93
67.8
11/12/2015
1
550
100
0.545
9.28
5.78
70
67.8
11/12/2015
3
550
0
1.34
7.7
5.68
89
72.1
11/12/2015
3
550
100
0.432
10.3
5.74
23
71.4
11/12/2015
5
550
0
1.37
7.82
5.70
61
75.8
11/12/2015
5
550
100
0.277
8.63
5.63
55
74.9
11/13/2015
24
550
0
1.02
5.63
5.53
50
76.8
11/13/2015
24
550
100
0.143
7.9
5.59
-5
75.6
c.) Time To
Equilibrium,
sand, pH=4
12/8/2015
0
N/A-lnitial Condition
1.87
9.08
3.96
260
64.5
12/8/2015
1
550
0
1.89
9.16
3.99
176
68.1
12/8/2015
1
550
100
1.58
10.5
5.19
-21
69.2
12/8/2015
3
550
0
1.75
9.1
3.94
179
79.7
12/8/2015
3
550
100
1.56
11.1
5.32
33
71.5
12/8/2015
5
550
0
1.9
9.21
3.95
153
77
12/8/2015
5
550
100
1.51
10.2
5.38
40
76.4
12/9/2015
24
550
0
1.76
8.2
3.86
297
74.5
12/9/2015
24
550
100
1.41
11
5.48
93
73.9
d.) Time To
Equilibrium,
Red/orange
sand, pH=4
11/17/2015
0
N/A-lnitial Condition
1.76
9.4
4.00
268
62.5
11/17/2015
1
550
0
1.74
9.34
4.01
249
67.3
11/17/2015
1
550
100
0.76
12.5
5.15
11
67
11/17/2015
3
550
0
1.77
9.35
4.02
260
70.8
11/17/2015
3
550
100
0.586
14.9
5.43
-55
71.3
11/17/2015
5
550
0
1.77
9.44
4.03
254
74.3
11/17/2015
5
550
100
0.484
14.2
5.37
-30
74.3
11/18/2015
24
550
0
1.69
8.86
3.99
344
79.7
11/18/2015
24
550
100
0.169
14.8
5.50
29
79.7
-------
a.)
As (D) Concentration vs. Time
S 1.6
< 1.2
in L1.6
=! 0.4
8 10 12 14 16
Contact Time (hrs)
-GW Only
-GW & Soil
b-)
As (D) Concentration vs. Time
2.2
2
1.8
1.6
1.4
1.2
1
0,8
0.6
0.4
0.2
0
8 10 12 14 16
Contact Time (hrs)
18 20 22
24
- GW Only
- GW & Soil
c.)
As (D) Concentration vs. Time (pH=3.9-5.5)
d.) As (D) Concentration vs. Time (pH=4-5.5)
oo l.S
8 10 12 14 16 18 20 22 24
Contact Time (hrs)
8 10 12 14 16
ContactTime (hrs)
18 20 22 24
GW Only
GW & Soil
GW Only
- GW & Soil
Notes: a.) Sand
b.} Red/orange sand
c.) Sand and pH adjustment
d.) Red/orange sand and pH adjustment
Figure 3.2 Time to equilibrium batch test results: Dissolved arsenic vs. time.
3-5
-------
a.)
Fe (D) Concentration vs. Time
8 10 12 14
Contact Time (hrs)
16 18 20 22 24
GW Only
GW & Soil
b.)
Fe (D) Concentration vs. Time
12
a 10
8 10 12 14 16
Contact Time (hrs)
GW Only
- GW & Soil
c.)
Fe (D) Concentration vs. Time (pH=3.9-5.5)
d.) Fe (D) Concentration vs. Time (pH=4-5.5)
0 2 4 6 8 10 12 14 16 18 20 22 24
Contact Time (hrs)
8 10 12 14 16
Contact Time (hrs)
18 20 22 24
GW Only
GW &Soil
GW Only
GW & Soil
Notes: a.} Sand
b.) Red/orange sand
c.) Sand and pH adjustment
d.) Red/orange sand and pH adjustment
Figure 3.3 Time to equilibrium batch test results: Dissolved iron vs. time.
3-6
-------
3.2.2 Sorption/Reaction Tests
Four separate sets of sorption/reaction batch tests were conducted:
1) Sand (low iron content) and groundwater were mixed at different ratios. No pH adjustment
was made.
2) Red/orange sand (high iron content) and groundwater were mixed at different ratios.
No pH adjustment was made.
3) Sand and groundwater were mixed at similar ratios. pH was varied by adding H2S04 or NaOH.
4) Red/orange sand and groundwater were mixed at similar ratios. pH was varied by adding
H2S04 or NaOH.
In each set of batch tests, bottles containing groundwater or soil plus groundwater were assembled and
sealed inside the anaerobic chamber. The bottles were rotated on a tumbler for 24 hours before being
transferred back into the anaerobic chamber, unsealed, and sampled. The results of all four tests were
used for model calibration (discussed in Section 3.3 below).
TEST 1: Sand (low iron content) and groundwater mixed at different ratios. No pH adjustment.
Seven bottles were filled with different ratios of sand and groundwater to evaluate the effect of soil/
water ratios on the sorption of arsenic. Soil amounts ranged from 30 g to 470 g. One bottle was also
filled only with groundwater to serve as a control. The details and results of each test are listed in Table
3.3. Plots of the results are shown on Figure 3.4. As the ratio of soil to water increased, the concentration
of arsenic in groundwater decreased. Batch tests in which more arsenic was removed also corresponded
to tests where more iron precipitated out of solution. The controls also experienced a decrease in arsenic
and iron, suggesting that freshly precipitated iron contributed to arsenic sorption. This hypothesis was
explored in more detail with geochemical modeling and is discussed in Section 3.3 below.
TEST 2: Red/orange sand (high iron content) and groundwater were mixed at different ratios.
No pH adjustment.
This set of batch tests was conducted in a similar way to Test 1 above, except the sand used in Test 2
was higher in iron than the sand used in Test 1. Six bottles were filled with differing ratios of soil and
groundwater, and one bottle was filled with only groundwater to serve as the control. Soil amounts
ranged from 30 g to 400 g. The results of this set of batch tests are shown on Figure 3.5 and listed in
Table 3.4. In general, as the ratio of soil to groundwater was increased, so did the rate and efficiency
of arsenic removal. Almost all of the arsenic in solution was removed with the highest ratio of soil to
groundwater that was tested.
Dissolved iron concentrations were also plotted on Figure 3.5. The iron concentration decreased in
batch tests involving low amounts of soil but increased above the initial measured value in several of
the batch tests with higher amounts of soil and this behavior was considered to be a result of either iron
desorption or dissolution.
3-7
-------
TEST 3: Sand and groundwater mixed at similar ratios. pH was varied.
In this set of batch tests, six bottles were filled with groundwater and 70 g of soil. One bottle was filled
with only groundwater to serve as the control. H2S04 or NaOH was added to five of the six bottles
containing soil to adjust the pH to a range between 4.0 and 6.8 which spans the pH values typically
expected to be encountered in the field. The pH of the remaining bottle with soil and groundwater was
not adjusted. Details of how much soil, groundwater, and acid or base was added to each batch test are
shown in Table 3.5, which also includes the batch test results. Final dissolved arsenic, dissolved iron,
As(lll) and As(V) concentrations are plotted against pH in Figure 3.6. In general, more arsenic and iron
remained in solution at lower pH values.
A decrease in arsenic was observed at all pH values. However, at pH values below 5, iron concentrations
were higher than initially measured, suggesting either release of adsorbed ferrous iron or dissolution of
iron minerals in the soil.
TEST 4: Red/orange sand and groundwater mixed at similar ratios. pH was varied.
This set of batch tests was conducted in much the same way as Test 3 above, except the sand used in
these batch tests had a higher iron content than the sand used in Test 3. Six bottles were filled with
groundwater and 70 g of soil. One bottle was filled with only groundwater to serve as the control. H2S04
or NaOH was added to five of the six bottles containing soil to adjust the pH to a range between 5.0
and 6.3 which covers the range of pH values expected to occur in the field under normal conditions.
The pH of the remaining bottle with soil and groundwater was not adjusted. Details of how much soil,
groundwater, and acid or base were added to each batch test are shown in Table 3.6 along with the
results of each test.
Final dissolved arsenic, dissolved iron, As(lll) and As(V) concentrations are plotted against pH in Figure
3.7. As the pH decreased, more arsenic and iron generally remained in solution, but the pattern is much
less distinct than was observed in Test 3. All of the batch tests in which the pH was decreased showed an
increase in dissolved iron concentration above the initially measured value.
Table 3.3 Sorption/reaction batch test results using sand with a low iron content
MW75S
Groundwater
(mL)
Aqueous Concentration (mg/L)
Measured in the lab
Description
Date
Soil
(S)
As
(Diss)
Fe
(Diss)
As
(Total)
Fe
(Total)
PH
ORP
(mV)
Temp
(°F)
Alkalinity
(mg/L as
CaC03)
10/1/2015
N/A-lnitial Condition
2.04
10.5
2.24
11.3
5.53
36
62.8
20
10/1/2015
575
0
1.73
9.26
2.18
11
5.56
14
72.7
1)
10/1/2015
550
33.25
1.36
5.5
N/A
N/A
5.51
58
72
15
Sorption/
10/1/2015
550
66.5
1.18
4.32
N/A
N/A
5.43
57
72
Reaction,
10/1/2015
500
133
1.04
3.64
N/A
N/A
5.58
39
72
15
sand, no pH
10/1/2015
450
266
0.962
4.17
N/A
N/A
5.7
-39
71.8
adjustment
10/1/2015
400
332.5
0.803
3.55
N/A
N/A
5.83
-37
71.7
10/1/2015
350
339
0.765
3.32
N/A
N/A
5.88
-200
71.9
20
10/1/2015
300
465.5
0.638
3.28
N/A
N/A
5.77
-239
70.7
3-8
-------
Table 3.4 Sorption/reaction batch test results using sand with a high iron content
Description
Date
MW75S
Groundwater
(mL)
Soil
(g)
Aqueous Concentration (mg/L)
Measured in the lab
As
(Diss)
Fe
(Diss)
As
(Total)
Fe
(Total)
As
(HI)
As
(V)
PH
ORP
(mV)
Temp
(°F)
2.)
Sorption/
Reaction,
Red/orange
sand, no pH
adjustment
12/3/2015
N/A-lnitial Condition
1.832
9.247
2.147
10.52
1.462
0.28
5.66
24
64.3
12/3/2015
575
0
1.422
6.896
N/A
N/A
1.212
0.135
5.58
-13
68.4
12/3/2015
550
33.25
0.428
6.283
N/A
N/A
0.392
0.0428
5.5
23
69.7
12/3/2015
550
67.5
0.267
5.855
N/A
N/A
0.234
0.0262
5.46
21
70.2
12/3/2015
500
133
0.118
9.379
N/A
N/A
0.103
0.0113
5.64
-97
67.1
12/3/2015
450
266
0.054
11.5
N/A
N/A
0.042
0.0046
5.47
-67
70.5
12/3/2015
350
399
0.03
13.81
N/A
N/A
0.0262
0.0033
5.51
-98
71.2
Table 3.5 Sorption/reaction batch test results using sand with a low iron content and pH adjustment
Description
Date
MW75S
Groundwater
(mL)
Soil
(g)
h2so4
(mL)
NaOH
(mL)
Aqueous Concentration (mg/L)
Measured in the lab
As
(Diss)
Fe
(Diss)
As
(Total)
Fe
(Total)
As
(IN)
As
(V)
PH
ORP
(mV)
Temp
(°F)
3.)
Sorption/
Reaction,
Low Fe
sand,
variable pH
3/31/2016
N/A-lnitial Condition
2.05
10
2.51
12.2
1.55
0.181
5.7
54
57.9
4/1/2016
552
0
0
0
1.41
6
N/A
N/A
1.17
0.11
5.68
77
71.3
4/1/2016
517
67.5
0
0
1.34
6.34
N/A
N/A
1.08
0.107
5.81
23
71.4
4/1/2016
505
67.5
10
0
1.43
8.77
N/A
N/A
1.16
0.11
5.38
99
72
4/1/2016
503
67.5
20
0
1.56
13.7
N/A
N/A
1.26
0.146
4.73
153
71.5
4/1/2016
492
67.5
30
0
1.61
15.2
N/A
N/A
1.3
0.147
4.41
173
72
4/1/2016
487
67.5
40
0
1.64
15.2
N/A
N/A
1.26
0.176
4.19
200
72.5
4/1/2016
475
67.5
50
0
1.65
17.9
N/A
N/A
1.3
0.145
4.01
234
70.6
4/1/2016
515
67.5
0
5
1.33
2.65
N/A
N/A
0.891
0.269
6.83
-117
71.1
Table 3.6 Sorption/reaction batch test results using sand with a high iron content and pH adjustment
Description
Date
MW75S
Groundwater
(mL)
Soil
(g)
h2so4
(mL)
NaOH
(mL)
Aqueous Concentration (mg/L)
Measured in the lab
As
(Diss)
Fe
(Diss)
As
(Total)
Fe
(Total)
As
(IN)
As
(V)
PH
ORP
(mV)
Temp
(°F)
4.)
Sorption/
Reaction,
Red/orange
sand,
variable pH
3/23/2016
N/A-lnitial Condition
1.54
7.81
2.54
12.1
1.28
0.138
6.01
35
66
3/24/2016
550
0
0
0
0.896
3.14
N/A
N/A
0.758
0.08
5.57
109
65.2
3/24/2016
508
67.5
0
0
0.176
15.9
N/A
N/A
0.149
0.018
5.92
-93
64.7
3/24/2016
507
67.5
5
0
0.256
18.5
N/A
N/A
0.217
0.024
5.84
-94
65.6
3/24/2016
505
67.5
10
0
0.24
21.4
N/A
N/A
0.208
0.026
5.89
-53
65.3
3/24/2016
498
67.5
20
0
0.284
22.6
N/A
N/A
0.249
0.028
5.45
82
66
3/24/2016
481
67.5
30
0
0.26
28.3
N/A
N/A
0.224
0.03
5.00
104
62.4
3/24/2016
515
67.5
0
4
0.179
1.4
N/A
N/A
0.159
0.019
6.27
-123
62.9
3-9
-------
Arsenic Removal vs. SoihWater Ratio
1.6
< 1.4
ao
E 1.2
| 0.8
£ 0.6
g 0.4
£ 0.2
0
0.5
1.5
g Soil/g Water
Fe (D) Decrease vs. SoihWater Ratio
8
—1
7
QjO
E
6
"(5
5
>
0
E
4
-------
Arsenic Removal vs. SoikWater Ratio
-Initial As (III) As(lll)
As(V) • Diss As
Initial As (V)
Initial Diss. As
0.4 0.6 0.8
g Soil / g Water
00
_£
c
o
k_
TJ
§j
>
o
Dissolved Iron Concentration vs. SoikWater Ratio
• Diss Fe Initial Diss Fe
14
12
10
8
6
4
2
0
0.0 0.2 0.4 0.6 0.8
g Soil / g Water
1.0
1.2
1.4
Notes:
Batch Test Description:
• Six bottles were filled with red/orange sand and
groundwater, one bottle was filled with only
groundwater to serve as a control. Soil amounts
were varied from approximately 30g to 400g.
• Bottles were sampled after 24 hours of mixing on a
rotary tumbler.
Initial Concentrations
Soil:
Red/Orange Sand from MP-1 (35-36.5' bgs)
As = 5.88 mg/kg
Fe = 10,300 mg/kg
Groundwater: MW75S
As = 1.83 mg/L (Dissolved)
Fe = 9.25 mg/L (Dissolved)
pH = 5.66
ORP = 24 mV
Figure 3.5 Sorption/reaction batch test results: Red/orange sand mixed with groundwater without pH adjustment.
3-11
-------
Final Aqueous Concentration vs. pH
• As(mg/L) • As (III) (mg/L) • As(V) (mg/L)
1.8
^L6
i1A
1.2
CO
< «
oa 1
LTt
< 0.8
V>
¦£
S 0.6
0.4
20
"58 15
E
-O 10
a;
>
O
• 0
•
0.3
•
0.25
3 •
•
•
¦
°'2 £
E
0.15 -
>
•
•
•
•
• •
•
0.1 g
0.05
0
4.5 5 55 6 6.5
pH
Dissolved Iron Initial Iron
4.5 5 5.5 6 6.5
pH
Batch Test Description:
• Six bottles were filled with sand and groundwater,
one bottle was filled with only groundwater to serve
as a control.
• Approximately 70 g of soil were added to each batch
test bottle.
• H2S04 or NaOH was added to each batch test in
order to adjust the pH.
• Bottles were sampled after 24 hours of mixing on a
rotary tumbler.
• Final pH values ranged from 4.0 to 6.8
Initial Concentrations
Soil: Sand from MP-1 (22.5 - 25' bgs)
As = 8.62 mg/kg
Fe = 1,950 mg/kg
Groundwater: MW75S
As = 2.05 mg/L (Dissolved)*
Fe = 10.0 mg/L (Dissolved)*
pH = 5.7
ORP = 54 mV
* Concentration before pH was adjusted by adding
acid or base.
Figure 3.6 Sorption/reaction batch test results: Sand mixed with groundwater with pH adjustment.
3-12
-------
Final Aqueous Concentration vs. pH
0.3
E 0.25 ®
= •
- 0.2
l/>
<
oa 0.15
to
<
-0 0.1
Diss As • As (III} • As (V)
«:
0
5.2
5.4 5.6
5.8
PH
• Dissolved Iron
Initial Iron
~3j
E
0.035
0.03
0.025
0.02
0.015 >
U">
0.01 <
0.005
0
6.2 6.4
5.2 5.4 5.6 5.8 6 6.2 6.4
PH
Batch Test Description:
• Six bottles were filled with sand and groundwater,
one bottle was filled with only groundwater to serve
as a control.
• Approximately 70 g of soil were added to each batch
test bottle.
• H2S04 or NaOH was added to each batch test in
order to adjust the pH.
• Bottles were sampled after 24 hours of mixing on a
rotary tumbler.
• Final pH values ranged from 5.0 to 6.3
Initial Concentrations
Soil: Sand from MP-1 (35-36.5' bgs)
As = 9.52 mg/kg
Fe = 10,300 mg/kg
Groundwater: MW75S
As = 1.54 mg/L (Dissolved)*
Fe = 7.81 mg/L (Dissolved)*
pH = 6.0
ORP = 35 mV
• Concentration before pH was adjusted by adding acid
or base.
Figure 3.7 Sorption/reaction batch test results: Red/orange sand mixed with groundwater with pH adjustment.
3-13
-------
3.2.3 Column Test
Arsenic-impacted groundwater from MW75S was run through a set of three columns: one column
packed with sand, one column packed with red/orange sand, and one column packed with "Filpro"
sand composed of at least 99% silicon dioxide which served as a control. These columns are shown in
Figure 3.8. The first sand has a low iron concentration (approximately 1,000-2,000 mg/kg iron) while the
concentration of iron in the red/orange sand is much higher (about 9,000-10,000 mg/kg iron). These
are the same soil types that were used in the batch tests. Groundwater was pumped at approximately
4 mL/min from the bottom of each column to the top of each column where one pore volume (about
118 mL) was allowed to accumulate above the packing material before removal. At this flow rate, one
pore volume took approximately 30 minutes to flush through the column and another 30 minutes to
accumulate above the packing material before it could be sampled. Nine pore volumes were completely
flushed through the column with the tenth filling the column, but not completely flushed through. Pore
volumes 1, 3, 5, 7, and 9 from each column were sampled for dissolved arsenic, iron, As(lll), and As(V).
Analytical sample results are presented in Table 3.7.
A single container was used to provide groundwater for all three column tests, which were run
concurrently. Several samples were collected from this container throughout the tests so that
the influent concentrations of arsenic and iron could be tracked. Figure 3.9a shows the influent
concentrations of arsenic and iron through time. Note that dissolved iron is plotted on the secondary
y-axis, and both arsenic and iron follow the same decreasing trend. Figure 3.9 also shows the column
effluent arsenic (Figure 3.9b) and iron (Figure 3.9c) concentrations through time as well as the influent
concentration corresponding to each sampled pore volume. Columns 1 and 3 approached arsenic
breakthrough while arsenic concentrations in the effluent of column 2 remained rather low throughout
the entire test.
Dissolution or desorption of iron in column 2 (sand with the high iron content) was observed in the
sample of the first pore volume to be flushed through the column. Figure 3.9c shows that the first
sample resulted in a higher concentration of dissolved iron in the effluent than was measured in the
influent suggesting that some of the iron in the sand had gone into solution under the conditions created
in the lab. The iron concentration observed in the effluent of column 2 then decreased as additional pore
volumes were flushed through the column.
An increasing trend in the iron concentration of the effluent was observed in columns 1 and 3 (Figure
3.9c). Iron concentrations in the first pore volume flushed through these columns were very low and
subsequently increased as more pore volumes were flushed through the columns. This suggests that
dissolution of the iron minerals was not occurring or was only minimally occurring.
Sample results of As(lll) and As(V) concentrations are plotted on Figure 3.10. As(lll) is plotted on the
primary y-axis, and As(V) is plotted on the secondary y-axis. The concentrations of As(lll) and As(V) in the
column influent followed the same decreasing trend with time. Effluent concentrations in column 1 and
column 3 increased to concentrations approximating the influent. Effluent concentrations remained low
(<0.01 mg/L total arsenic) in the high-iron column (column 2). Concentrations of As(lll) remained higher
than As(V) throughout the column tests.
3-14
-------
Figure 3.8 Column tests.
Column 1: Sand with low iron
(Fe = 1,000-2,000 mg/kg)
Column 2: Sand with high iron
(Fe = 9,000-10,000 mg/kg)
Column 3: Filpro silica sand
3-15
-------
Table 3.7 Column test analytical results
Column
Pore
Volume
Time Collected
As
(mg/L)
Fe
(mg/L)
As (III)
(mg/L)
As (V)
(mg/L)
CI
(mg/L)
N3"
(mg/L)
so42
(mg/L)
Mg2+
(mg/L)
K+
(mg/L)
Na+
(mg/L)
Ca2+
(mg/L)
Influent
l
8/24/2016 8:00
2.16
10.4
2.098
0.29
2.31
0.0264
8.73
0.306
1.35
2.31
1.48
Influent
3
8/24/2016 10:00
2
9.66
2.065
0.266
Influent
5
8/24/2016 11:30
1.91
9.42
2.024
0.265
Influent
7
8/24/2016 12:34
1.93
9.26
1.996
0.265
Influent
9
8/24/2016 13:45
1.9
9.17
1.976
0.252
Influent
10
8/24/2016 15:20
1.92
9.18
1.973
0.227
1
1
8/24/2016 10:15
0.393
0.437
0.431
0.0488
1
3
8/24/2016 11:35
0.987
0.225
1.059
0.1
1
5
8/24/2016 12:40
1.67
1.97
1.743
0.159
1
7
8/24/2016 14:00
1.8
6.28
1.83
0.201
1
9
8/24/2016 15:29
1.75
9.08
1.798
0.176
0.62
1.42
2.67
2.61
2
1
8/24/2016 10:40
0.204
13.5
0.0019
0.0006
2
3
8/24/2016 11:43
0.387
5.99
0.0042
0.001
2
5
8/24/2016 12:44
0.257
6.05
0.0027
0.0007
2
7
8/24/2016 14:05
0.0026
4.64
0.0025
0.0009 I
2
8
8/24/2016 14:54
2.34
0.0514
8.85
2
9
8/24/2016 15:36
0.337
3.77
0.0031
0.0007
0.428
1.37
2.71
1.38
3
1
8/24/2016 10:45
1.35
0.255
1.4
0.112
3
3
8/24/2016 12:08
1.76
2.3
1.897
0.168
3
5
8/24/2016 13:15
1.86
4.91
1.955
0.205
3
7
8/24/2016 14:40
1.79
5.83
1.91
0.201
3
8
8/24/2016 15:15
2.33
0.348
16.3
3
9
8/24/2016 15:41
1.89
6.6
1.881
0.18
0.51
1.46
2.81
1.78
Column 1 = Sand
Column 2 = Red/orange Sand
Column 3 = "Filpro" 99% Silica Sand
3-16
-------
(a.)
S
£
2.2
2.1S
2.1
§ 205
< 2
73
QJ
¦I 1.95
a
5 1-9
1.85
A*
Influent Concentrations
¦ Influent - As
¦ Influent-Fe
10.6
10.4
I
io.2 "a
10
(b.)
Effluent Arsenic Concentraiton
¦Coll —•— Col 2 —•—Col 3 Influent
Sample Time
(C.)
16
rr 14
"a 12
10
8
6
4
2
0
JE
St
j/j
b
Effluent Iron Concentraiton
- Column 1 - ~ - Column 2 —•— Column 3
Sample Time
Influent
Notes:
• Pore volumes 1, 3, 5, 7, and 9 were sampled and analyzed
for As and Fe. (Effluent samples)
• The influent groundwater was also sampled at the same
time.
• Observations:
• Decline in As in the influent follows the same trend
as the decline in Fe
• The highest As removal was observed in the sand
column with the highest Fe content (column 2)
Column 1 = Low Fe Sand (1,000-2,000 mg/kg Fe)
Column 2 = High Fe Red/Orange Sand (9,000-10,000 mg/kg Fe)
Column 3 = "Control"
Figure 3.9 Column tests results: Influent and effluent arsenic and iron concentrations.
3-17
-------
Influent
-As(lll)mg/L —As(V)mg/L
2.12
2.1
2.08
"35 2.06
£
^ 2.04
IX 2.02
<
2
1.98
1.96
0.3
2
0.29
1.8
1.6
0.28
I
1.4
0.27
~5s
1.2
E
E
0.26
-
1
>
—
0.25
Irt
V)
0.8
<
<
0.6
0.24
0.4
0.23
0.2
0.22
0
7:12 8:24 9:36 10:48 12:00 13:12 14:24 15:36 16:48
Column 1
-As(lll) mg/L —As(V)mg/L
9:36
10:48 12:00 13:12 14:24 15:36
0.25
0.2
0.15 |js
0.1
0.05
>.
<
0
16:48
Column 2
-As(lll) mg/L —•— As(V}mg/L
Column 3
-As(lll)mg/L —•— As(V)mg/L
0.0045
0.004
0.0035
4^ 0.003
25
E 0.0025
% 0.002
< 0.0015
0.001
0.0005
0
¦ l.
0.001
0.0008
0.0006
0.0004 <
~Sa
E
0.0002
2.5
2
1.5
1
0.5
9:36
10:48 12:00 13:12 14:24 15:36
L 0
16:48
9:36
10:48
12:00 13:12
14:24
15:36
0.25
0.2
0.15 ~|S
0.1
0.05
>
<
0
16:48
Notes:
Column 1 = Low Fe Sand (1,000-2,000 mg/kg Fe); Column 2 = High Fe Red/Orange Sand (9,000-10,000 mg/kg Fe); Column 3 = Control
Y-axis magnitude varies by plot
Figure 3.10 Column tests results: Influent and effluent arsenic speciation.
3-18
-------
The pH, ORP, and temperature of the influent and effluent were measured frequently throughout the
test. These values are shown in Table 3.8 and plotted with time in Figure 3.11. It is likely that degassing
of C02 during sample collection influenced the measured pH in most of the tests, especially in the
samples collected from the effluent. There are no clear trends in pH or ORP associated with the column
effluent samples. The pH of the influent appears to gradually increase throughout the duration of the
test while the ORP decreases, which is consistent with the slow degassing of C02.
Table 3.8 Column test laboratory measured parameters
Column
Pore
Volume
Time Collected
PH
ORP
(mV)
Temp
(°F)
Influent
1
8:00:00 AM
5.8
40
67.8
Influent
3
10:00:00 AM
5.97
14
67.9
Influent
5
12:34:00 PM
6.08
-32
68.8
Influent
7
1:45:00 PM
6.17
-28
67.8
Influent
9
3:20:00 PM
6.22
-26
68
1
1
10:15:00 AM
6.35
71
69.1
1
2
10:52:00 AM
6.46
-44
69.2
1
3
11:35:00 AM
6.34
-5
69.2
1
4
12:05:00 PM
6.19
69
1
8
3:00:00 PM
6.09
-71
68.5
1
9
3:29:00 PM
6.31
-68
68.4
2
1
10:40:00 AM
5.61
72
68.2
2
3
11:43:00 AM
5.84
73
69.1
2
4
12:14:00 PM
6.2
63
68.8
2
8
2:45:00 PM
5.63
60
68.8
2
9
3:36:00 PM
6.03
56
68.5
3
1
10:45:00 AM
6.4
-10
68.3
3
3
12:08:00 PM
6.6
69.3
3
4
12:51:00 PM
6.24
-37
69
3
8
3:15:00 PM
6.19
-35
68.9
3
9
3:41:00 PM
6.22
-39
68.6
3-19
-------
pH
6.8
6.6
6.4
6.2
6
5.8
5.6
5.4
& , <$> <9 J§> J& 0? .<$> rP n<§> & J§> J? <§> , .<£ , & , cP , .0?
*¦' °> of- NCS' $¦' %S- ^ $¦' $¦' %s- ^ $¦' N«-'
¦ Column 1
Column 2
¦ Column 3
Influent
Notes:
The pH and ORP of the influent and effluent were monitored
throughout the column test. The plots to the left show the
variation in these parameters throughout the test. Degassing of
C02 was likely to have occurred and could have influenced pH
measurements.
Column 1 = Low Fe Sand (1,000-2,000 mg/kg Fe)
Column 2 = High Fe Red/Orange Sand (9,000-10,000 mg/kg Fe)
Column 3 = "Control"
Figure 3.11 Column tests results: Parameters measured in the laboratory.
3-20
-------
3.3 Model Simulations
The geochemical model PHREEQC, version 3 (Parkhurst and Appelo, 1999; Parkhurst and Appelo, 2013)
was used to simulate arsenic sorption and iron precipitation under laboratory and field conditions.
Model calibration involved identifying key parameters impacting the fate and transport of arsenic and
iron in the sorption/reaction batch tests and column tests, and adjusting those parameters until the
modeled arsenic and iron concentrations came close to the concentrations measured in the lab. The
calibration process resulted in a range of values for fate and transport parameters that could be used to
predict arsenic and iron concentrations in the field.
3.3.1 Model Calibration to Batch and Column Test Data
The default parameter values found in the wateq4f database that accompanies PHREEQC were used in
the model simulations. Any modifications made to the wateq4f database are discussed in 2.5 and are
shown in Appendix D. A summary of the conditions and assumptions applied during model calibration is
given below:
1) The oxidation of As(lll) to As(V) or reduction of As(V) to As(lll) occurs slowly under laboratory
and field conditions and is not significant over the course of the batch tests or column tests.
Therefore, these processes were not allowed to occur in the model.
2) All processes were assumed to proceed to equilibrium with the exception of iron oxidation.
Both homogeneous and heterogeneous oxidation mechanisms contribute to the rate of iron
oxidation. The kinetics of iron oxidation were described in the model using the following
equation:
rate = -fchom[Fe(//)][02][H+]"z - fehet[Fe(///)][Fe(//)][02][//+]-1
The homogeneous (khom) and heterogeneous (khet) rate constants (8.4E-14 mol/L/sec and
9.5E-04 L/mol/sec, respectively) were taken from Dietz and Dempsey (2001) and corrected
for temperature as needed using the Arrhenius equation. The initial solid Fe(lll) included in the
rate calculations was assumed to be equal to the difference between total and dissolved iron
initially measured in the groundwater and a small percentage of the iron in the soil (0.2%).
Initial oxygen was assumed to be less than 1 mg/L in the majority of the simulations.
3) Arsenic sorption was modeled using surface complexation and the parameters for HFO from
Dzombak and Morel (1990). These include a specific surface area of 600 m2/g and a weak site
density of 0.2 moles per mole of iron. Sorption of arsenic to HFO was assumed to be
instantaneous.
4) Iron that precipitated out of solution over the course of the batch or column tests was
considered to be HFO and available as a sorption surface for arsenic. In laboratory batch tests
involving soil, a small percentage of iron in soil, ranging from 0.5% to 5%, was also assumed to
be HFO and available as a sorption surface in order to obtain a good calibration. The percentage
of iron in soil assumed to be HFO was even smaller in the column tests and was at most 0.1%.
The higher sorption capacity in the batch tests could be due to suspension of the HFO compared
to precipitation on existing surfaces during the column tests.
3-21
-------
5) Degassing of C02 results in an increase in pH. In most instances, the amount of C02 degassing
allowed in the model ranged from approximately 10% to 65% of the initial concentration. There
were several instances where up to 92% degassing was allowed in order to match pH and iron
concentrations that were observed during the column tests.
6) As noted above, the final measured iron concentration was often higher than the initial
concentration in batch tests that included soil. This phenomenon could potentially be due
to either dissolution of iron minerals or desorption of ferrous iron. For modeling purposes,
it was assumed that dissolution of iron was taking place, and equilibrium with magnetite and
siderite was included in the model in order to simulate iron dissolution. It is possible that these
minerals exist naturally or were formed under the anaerobic conditions inside the chamber
where the soil was stored. Up to 5% of the iron in soil was modeled as siderite.
As noted above, the model was calibrated to match the dissolved iron and arsenic results from the batch
tests. Figure 3.12 shows the overall fit of the model to the measured results for all sets of sorption/
reaction batch tests (described in Section 3.2.2). The diagonal solid black line represents a perfect match
between modeled and measured concentrations. Points that fall close to the black line correspond to
batch tests where the modeled results were in good agreement with the measured results. Most points
fall close to this line with the exception of one noticeable outlier on the plot of modeled vs. measured
arsenic. This point represents the result from the batch test conducted with low iron sand and pH
adjustment with NaOH to a final pH of 6.83 where the model significantly over-predicted the amount of
sorption that occurs. This discrepancy between modeled and measured results is due to removing the
process of As(lll) oxidation to As(V) in the model. The measured As(V) at the end of the batch test was
higher than it was at the beginning of the batch test suggesting that, at this pH value, some conversion
of As(lll) to As(V) did take place during the test. If more arsenic was modeled in the As(V) phase, less
sorption would occur because the sorption of As(lll) is more favorable than the sorption of As(V) at this
pH, and the model predictions would be closer to the experimental observations. These observations are
consistent with the findings reported in Jang and Dempsey (2008).
Figure 3.12 also compares modeled to measured percent changes in arsenic and iron concentrations
from the batch tests. It can be seen from these plots that either large or small changes in dissolved
arsenic and iron concentrations could be replicated with the model.
3-22
-------
Modeled vs. Measured Dissolved Arsenic
Modeled vs. Measured Dissolved Iron
DO 1.4
« 0.8
0 0.5 1 1.5
Measured As (mg/L)
Dissolved Arsenic
Measured vs. Modeled Percent Change
100%
80%
60% •
0)
OD
c
ID
SI
U
\P
ON
H 40%
T3
o
20%
0%
0% 20% 40% 60% 80%
Measured % Change
100%
0 10 20
Measured Fe (mg/L)
Dissolved Iron
Measured vs. Modeled Percent Change
300%
30
£ 200%
* 150%
£ 100%
0% 100% 200%
Measured % Change
300%
Notes:
PHREEQC was used to
replicate batch test results.
This figure shows plots of
modeled vs. measured
results for all sets of batch
tests (batch tests using low
Fe sand and high Fe sand,
with and without pH
adjustment as well as
groundwater control batch
tests. Simulations where
the modeled results are
equal to the measured
results would fall along the
plotted black line.
Legend:
• Arsenic
• Iron
Line of Perfect Fit
Figure 3.12 Sorption/reaction batch tests results: Comparison of modeled to measured iron and arsenic concentrations.
3-23
-------
The model could also replicate the majority of the dissolved iron and arsenic concentrations that were
measured during the column tests (Figure 3.13) by using the modeling conditions and assumptions that
are listed above. Each sampled pore volume that passed through the column was modeled separately.
Approximately one hour elapsed from the time the groundwater entered each column to the time when
the pore volume was removed from the column for sampling. Only the column test data from columns
filled with site soil were modeled; the control column was not modeled.
Measured iron and arsenic concentrations in the effluent of the columns could generally be explained
by using model parameters within a reasonable range of those used to model the batch tests, but some
results were more difficult to replicate with the original model parameters. The column filled with low-
iron sand experienced very rapid precipitation of iron in the groundwater followed by a gradual increase
until the effluent concentrations of dissolved iron were almost the same as the influent concentrations.
The initial rapid drop in iron could only be replicated by assuming an initial oxygen concentration of
2 mg/L, which is higher than the 0.2 mg/L - 0.8 mg/L assumed during model calibration to batch test
results, and significant degassing of C02 (> 90%, also much greater than was used to model the batch
tests) in order to push the pH higher and speed up the rate of iron oxidation. Additionally, sorption
of arsenic was over-estimated when applying the same assumptions for calculating HFO that were
used during model calibration. In order to match arsenic results from most of the column tests, HFO
was assumed to be less than the amount of iron that precipitated out of solution. These observations
suggest that some aspects of the fate and transport processes were poorly modeled when solid iron
concentration was low and for the early pore volumes. The time scale in the field is closer to the
later parts of the column tests and the batch tests (or longer), both of which were modeled using a
reasonable set of parameters.
The results of the column test using sand with the higher iron concentrations were better fit using
parameters within the ranges applied when modeling the batch tests. The degassing of C02 ranged
between 10% and 60% of the initial concentration in order to match the effluent pH. The initial DO
concentration was set between 0.83 mg/L and 0.88 mg/L. Arsenic concentrations could be replicated
by assuming all freshly precipitated iron and between 0.02% and 0.1% of the iron in the soil was HFO, a
value that was less than was assumed for the batch tests. This assumption is reasonable considering the
batch tests were thoroughly mixed, exposing more of the HFO surface area. HFO precipitated on existing
surfaces in the column tests and the specific surface area (area per mass) is expected to be lower.
Model results and assumptions for the batch and column tests have been tabulated in Appendix E.
3-24
-------
-©- Column 1 Measured As • Column 1 Modeled As
O Column 2 Measured As • Column 2 Modeled As
2 7
Notes:
uo
J. 1.5
©
Pore
Volume
"5 #: 1
Column 1 = Low Fe Sand (1,000-2,000 mg/kg Fe)
9 Column 2 = High Fe Red/Orange Sand (9,000-10,000 mg/kg Fe)
Colored circles on the graphs represent model calculated values.
3 Empty circles represent measured values.
The pore volume number corresponding to the plotted data is
.5 0 5 3 posted next to each data point.
® © A ©
~ © o 7 9
Q o 1 5 6
9:36 10:48 12:00 13:12 14:24 15:36 16:48
Sample Time
O Column 1 Measured Fe • Column 1 Modeled Fe
O Column 2 Measured Fe • Column 2 Modeled Fe
16 Pore
— 1A Volume #: l
-> 14 ®
12
10 9
6>
8
ao
E
-a 5 7
£ 6 ® ® ®
"o 4 ®
8 5 7 ®r
5 2 1 3 ®
9
0 • •
9:36 10:48 12:00 13:12 14:24 15:36 16:48
Sample Time
Figure 3.13 Column tests: Comparison of modeled and measured results.
3-25
-------
3.3.2 Model Simulations of Field Conditions
Arsenic and iron concentrations were modeled under pre-air sparge and air sparge conditions using
the methodology that was developed from the batch and column test results. The model of pre-air
sparge conditions involved replicating the decrease in arsenic and iron observed between MP-9Land
PZ-15 (see Figure 3.14 for well locations). These wells are spaced approximately 165 ft apart. At an
average groundwater flow velocity of 2.5 ft/day, it takes approximately 66 days for groundwater to
travel from MP-9Lto PZ-15. Prior to air sparging, the concentrations of dissolved arsenic and dissolved
iron at MP-9L were 0.989 mg/L and 11.7 mg/L, respectively. By the time groundwater had reached PZ-
15, the concentration of arsenic had dropped to about 0.01 mg/L and iron had dropped to about 0.06
mg/L. The decline in dissolved iron was replicated by the model based on an initial pH of 5.8 and a DO
concentration of 0.1 mg/L, both of which are consistent with measured pre-sparge field conditions.
Degassing of C02 was not included in this model simulation because it is unlikely to occur in the field
under non-air sparging conditions. After 66 days, the model was used to calculate an iron concentration
of 0.044 mg/L, which is a good approximation considering there is some uncertainty in the exact pH,
DO, and travel time from MP-9L to PZ-15. Figure 3.14 shows a plot of the modeled dissolved iron
concentration through time. Dissolved arsenic was calculated by the model to be 0.02 mg/L (only 0.01
mg/L higher than the observed concentration) assuming that all the iron which precipitated out of
solution along with less than 1% of the iron in soil is the HFO sorption surface. With these assumptions,
the model predicted that 75% of the arsenic removed from groundwater was due to sorption to the
freshly precipitated HFO. The remaining 25% of the arsenic removed from solution was due to sorption
with iron oxides that previously were in the soil.
PHREEQC was also used to replicate the decrease in arsenic and iron that occurs between MP-1L and
locations that are downgradient from the air sparge system. The goal of this simulation was to capture
the general behavior of dissolved arsenic and iron as groundwater moves through the air sparged area.
MP-1L is upgradient of the air sparge system and is not impacted by the air sparging, but there was some
variation in iron and arsenic concentrations measured during the pilot study at this location. Dissolved
iron concentrations ranged from 14.8 mg/L to 18.9 mg/L, and dissolved arsenic concentrations ranged
from 1.31 mg/L to 1.87 mg/L. To reflect this variation in influent concentrations to the air sparged area,
average values of dissolved arsenic and dissolved iron (1.4 mg/L and 15.5 mg/L, respectively) were used
as the initial concentrations in the model.
Although concentrations downgradient of the air sparge system varied as the operational conditions
of the system were changed as part of the pilot test, concentrations and field parameters measured at
MP-4L serve as a good approximation of the average conditions. MP-4L is located 50 ft downgradient
of the air sparge system, which equates to a 20-day travel time if the groundwater velocity is assumed
to be 2.5 ft/day. Typical concentrations of iron and arsenic at MP-4L are approximately 2.9 mg/L and
0.15 mg/L, respectively, while the air sparge system was fully operational and providing a good supply
of oxygen to the groundwater. Assuming an initial DO concentration of 2 mg/L and an equilibrium pH of
5.5, the model predicted a dissolved iron concentration of 3 mg/L after 20 days, which is a reasonable
approximation of the dissolved iron at MP-4L.
The iron that precipitated out of solution between the line of air sparge wells and MP-4L was assumed to
be HFO. Additionally, a very small percentage of iron in the soil (less than 1%) was modeled as HFO, and
the oxidation of As(lll) was assumed to be insignificant. The oxidation of As(lll) was removed from the
PHREEQC database in the same way it was for simulations involving anoxic conditions. Under air sparge
3-26
-------
-Modeled Fe
-Modeled pH
Groundwater flow between MP-9L and PZ-15
Distance = 165 ft
Average Groundwater Flow Velocity = 2.5 ft/day
Travel Time = 66 days
s mi
.
r.ivw
~
i.r.
MP 7L
PZ-10 J-VXL
• ~
MP-3L
• MP-2L
~ '4
HP-1L
r.(P9L
PZ-11
Jjfv . - V-:.-
"0
Monitoring Location
Air Sparge Well
Figure 3.14 Simulated iron concentrations: Pre-air sparge conditions.
-------
conditions, the model predicts a dissolved arsenic concentration of 0.14 mg/L, which is very close to
what is observed at MP-4L. Using these model assumptions, 80% of the arsenic that is removed from the
groundwater sorbs to the freshly precipitated iron while only 20% sorbs to iron oxides in the soil.
Considering the variability in iron and arsenic concentrations observed at MP-1L, which served as
initial conditions for the model, the model provides a very reasonable approximation of the impact of
air sparging on the downgradient concentrations of dissolved iron and arsenic. However, it should be
noted that some of the soil downgradient of the air sparge system contains elevated concentrations of
arsenic (above 100 mg/kg), which could impact the concentrations of dissolved arsenic measured in the
monitoring wells. Since modeling the release of arsenic from soil with elevated initial concentrations
is beyond the scope of this investigation, the model results are not expected to precisely predict the
observed arsenic concentration downgradient of the air sparge system in locations where soil contains
high levels of arsenic.
3.4 Sources of Uncertainty
Although the model replicated iron and arsenic concentration data collected in the lab and in the field,
several sources of uncertainty exist. Future modeling work could focus on quantifying the sensitivity of
the model results to changes in input parameters for which there is some uncertainty. This section lists
the sources of uncertainty encountered during laboratory tests and geochemical modeling.
1) Effort was made to mimic the in-situ site conditions as much as possible when conducting
batch and column tests in the lab. Exposure of the groundwater to oxygen was minimized
by filling containers from the bottom up and allowing the containers to overflow when
collecting groundwater used in batch and column tests. Groundwater collection bottles were
capped with no head space and only opened once placed inside an anaerobic chamber.
Throughout the duration of the batch and column tests, bottles containing groundwater were
only opened inside the anaerobic chamber. However, it is possible that some contact with air
occurred either during initial groundwater collection or by seepage through seals on bottles
while the bottles were not inside the chamber. It is known that the oxygen in groundwater
collected from MW75S is generally low, but there was no way to precisely measure changes in
oxygen concentration throughout the batch and column tests. To some extent, DO was used as
a model calibration parameter in order to match the observed rate of iron oxidation, but this
value was not adjusted beyond what could be considered reasonable.
2) The potential impact of photo oxidation on batch test results is unknown. Exposure of the
groundwater to light was minimized by covering bottles with foil, but some light exposure
was unavoidable during groundwater collection and batch and column test sampling.
3) Some degassing of C02 is unavoidable under ambient laboratory conditions. Groundwater
in the vicinity of MW75S is supersaturated with C02 due to biological activity. Once groundwater
is removed from the aquifer, changes in pressure and temperature as well as groundwater
handling (mixing, pouring, pumping, etc.) lead to degassing, which can also impact the pH of
the groundwater. Since there is no reasonable way to measure the extent of degassing of C02
during the laboratory tests, this parameter was adjusted to achieve model calibration.
Degassing of C02 is not expected to occur under non-air-sparge ambient field conditions.
3-28
-------
4) Not all processes impacting pH were included in the model simulations, and, because of this,
the model is not expected to precisely match the measured pH. In some model simulations,
the pH of the solution was fixed to the final measured pH. This was mostly the case in model
runs simulating the batch tests involving pH adjustment with H2S04 or NaOH. Although it is
known that the pH of the groundwater has some impact on sorption of arsenic and the rate
of iron oxidation, the sensitivity of the model calibration to the final pH value was not
specifically explored.
5) Soil collected from the Vineland Chemical Company Site was used in the batch and column tests.
Although sampling of the soil for iron and arsenic confirmed that the concentrations of these
parameters did not vary significantly among the batch and column tests, the soil samples
were not completely homogeneous. Calculations of model inputs such as sorption surfaces
and Fe(lll) contributing to the rate of iron oxidation were based on the selection of one value
for iron concentration. The sensitivity of the model calibration to potential variation in the soil
iron concentration has not been explored.
6) Measurement errors of soil mass and groundwater volume could have impacted the overall
calibration results. PHREEQC performs calculations on a molar basis for a particular control
volume. Both soil mass and groundwater volume are incorporated into the calculation along
with concentrations of each chemical constituent to determine how many moles of iron,
arsenic, etc. are involved in each model simulation. The extent to which changes in soil mass
and groundwater volume impact the overall model results have not yet been explored.
7) The soil used in the batch and column tests was stored inside of the anaerobic chamber until
it was needed. A slight change in soil color was noted after it had been inside the anaerobic
chamber for some time. Some of the soil, which was originally tan to orange, developed a
greyish tint, suggesting the formation of iron minerals different than those naturally found
in the subsurface.
3-29
-------
SECTION 4
Summary and Conclusion
After a P&T system for arsenic treatment could
not meet the long-term remediation goals
established in the Vineland Chemical Company
Superfund Site ROD, a more sustainable approach
was pursued which involved immobilizing arsenic
in-situ with an air sparge system, The success of
in-situ remediation requires a comprehensive
understanding of the processes controlling the
transport of arsenic. The purpose of this study
was to determine the processes responsible for
arsenic immobilization at the site through the
collection and evaluation of site groundwater and
soil data, bench-scale testing, and geochemical
modeling. The evaluation successfully identified
the processes controlling immobilization, as
well as other key factors that contributed to the
sustainable operation of the air sparge system.
Key findings include:
1) As (III) is removed from solution by
sorption to HFO and iron in soil.
As(lll) can sorb to iron oxides without first being
oxidized to As(V). The presence of As(lll) in site
soil samples provides direct evidence that this
process is occurring. The process of As(lll)
sorption to HFO was successfully modeled using
the surface complexation model available in
PHREEQC. Indirect evidence through modeling
shows that only a small portion of arsenic
partitions to iron already in the soil, and the
majority of arsenic partitions to the freshly
oxidized Fe(ll) (HFO). Arsenic sorption to HFO
is believed to be the main mechanism for
arsenic attenuation in groundwater at the
Vineland Chemical Company Site. Direct
evidence to confirm this hypothesis could be
gathered through bench-scale testing of iron-
poor arsenic impacted groundwater with the
site soil.
4-1
-------
2)
The rate of HFO formation is highly dependent on pH.
Fe(ll) oxidation and precipitation is not rapid at pH values below 6.5. The pH of groundwater at the
Vineland Chemical Company Site downgradient of the air sparge system is controlled by oxidation
of iron, which decreases pH and degassing of C02 in the presence of air bubbles, resulting in
increased pH. The native soil also has some buffering capacity. The process of iron oxidation was
successfully modeled as a function of DO and pH using a non-equilibrium kinetic rate equation.
3) HFO can transform into the more thermodynamically and more crystalized
minerals goethite and hematite over time.
Both goethite and hematite were found in soil samples collected from the Vineland Chemical
Company Site upgradient and downgradient of the air sparge system. Goethite is encountered
at a higher frequency and mass fraction as compared to hematite. This is attributed to the
formation and growth mechanisms of goethite. However, it is unknown how quickly the
transformation of HFO to goethite and hematite can occur under site-specific conditions, which
impacts the long-term stability of the arsenic immobilized by the air sparge system. Arsenic can
desorb more easily from HFO than from more crystalline iron oxides. Further investigation is needed
to understand the rate of arsenic desorption if the air sparge system is turned off.
4) It is desirable to control the rate of iron oxidation to optimize in situ
system performance.
Post air sparging, the pH of the groundwater at the Vineland Chemical Company Site ranged
between 5.5 and 6.5, which ultimately controlled the rate of iron oxidation. At this pH range,
iron precipitated out of solution downgradient of the air sparge system, eliminating potential
problems due to iron fouling of the sparge wells. An elevated pH at the air sparge line would
result in virtually instantaneous precipitation of iron which could lead to clogging of the system.
Maintaining a pH between 5.5 and 6.5 allows iron to precipitate downgradient of the air sparge line
but before discharging to surface water. Since iron and arsenic are removed concurrently from
groundwater, dissolved arsenic concentrations were also reduced before the arsenic plume reached
the surface water compliance point.
The principles controlling iron oxidation and arsenic immobilization are not unique to the Vineland
Chemical Company Site. Air sparging for arsenic immobilization can be applied to other sites where
iron is present in groundwater in sufficient quantities, and a similar procedure of groundwater and
soil characterization combined with bench-scale testing and modeling can be applied to identify the
parameters most influential on pH and iron oxidation rates.
The factors that control the rate of iron oxidation can also be utilized to optimize any remedial approach
where iron removal is desired, such as abandoned mine drainage (AMD) sites. Understanding these
factors can allow for pH control so that iron is removed at a rate that will reduce maintenance needs due
to iron fouling and also meet water quality standards at compliance points.
-------
SECTION 5
References
Adler, H. H and Kerr, P. F., 1962. Infrared study of aragonite and calcite. American Mineralogist, 47:700-717.
Bundschuh, J. and Bhattacharya, P., 2014. In-Situ Remediation of Arsenic-Contaminated Sites, CRC Press.
Cherry, J. A., Shaikh, A. U., Tallman, D. E., and Nicholson, R. V., 1979. Arsenic species as an indicator of redox conditions in
groundwater. Journal of Hydrology, 43:373-392.
Cornell, R. M. and Schwertmann, U., 1996. The Iron Oxides: Structure, Properties, Reactions, Occurrence and Uses, VCH,
New York.
Dietz, J. M., and Dempsey, B. A., 2001. Treatment of Mine Drainage Using Recirculated Iron Oxides In: A Complete Mix
Reactor. Proceedings of the 2001 National Association of Abandoned Mine Lands Annual Conference, August 19-22, 2001,
Athens, Ohio.
Dzombak, D. A., and Morel, F. M. M., 1990. Surface complexation modeling—Hydrous ferric oxide, John Wiley, New York.
Ford, R. G., Kent, D. B., Wilkin, R. T., 2007. Monitored Natural Attenuation of Inorganic Contaminants in Groundwater,
Volume 2. Assessment for Non-Radionuclides Including Arsenic, Cadmium, Chromium, Copper, Lead, Nickel, Nitrate,
Perchlorate, and Selenium. EPA/600/R-07/140.
Geo Trans, Inc., U.S. Army Corps of Engineers - Philadelphia District, U.S. Army Corps of Engineers Environmental and
Munitions Center of Expertise, Remedial System Evaluation, Vineland Chemical Company Superfund Site, Vineland, New
Jersey, Final Report March 11, 2011.
Hsia, T.-H., Lo, S.-L., Lin, C.-F., and Lee, D.-Y., 1994. Characterization of arsenate adsorption on hydrous iron oxide using
chemical and physical methods. Colloids and Surfaces A, 85:1-7.
Jang, J.-H. and Dempsey, B. A., 2008. Co-adsorption of As (III) and As(V) onto Hydrous Ferric Oxide: Effects on Abiotic
Oxidation of As(lll), Extraction Efficiency, and Model Accuracy. Environmental Science and Technology, 42(8):2893-2898.
Kostka, J. E. and Luther, G. W., 1994. Partitioning and speciation of solid phase iron in saltmarsh sediments. Geochimica et
Cosmochimica Acta, 58(7):1701-1710.
Parkhurst, D.L. and Appelo, C. A. J., 2013. Description of input and examples for PHREEQC version 3-A computer program
for speciation, batch-reaction, one-dimensional transport, and inverse geochemical calculations: U.S. Geological Survey
Techniques and Methods, book 6, chap. A43, ittp://pubs.usgs.gov/tm/06/a43/.
Parkhurst, D.L. and Appelo, C. A. J., 1999. User's guide to PHREEQC (Version 2)-A computer program for speciation, batch-
reaction, one-dimensional transport, and inverse geochemical calculations: U.S. Geological Survey Water-Resources
Investigations Report 99-4259, http://pubs.er.usgs.gov/publication/wri994259.
Roy, W. R., Krapac, I. G., Chou, S. F. J., and Griffin, R. A., 1991. Batch-Type Procedures for Estimating Soil Adsorption of
Chemicals, EPA/530-SW-87-006-F.
Sehayek, L. Klaber, N., Chamberlain, M., Bittner, L., Modica, E., 2014a. Optimization of Remedial Action for Arsenic
in Groundwater at Vineland Chemical Co. Superfund Site, New Jersey", Battelle's Tenth International Conference on
Remediation of Chlorinated and Recalcitrant Compounds, May 19-22,2014, Monterey, CA.
Sehayek, L., Klaber, N., Corona, J., North, T., Chamberlain, M., Loeper, J., Bittner, L., and, Charlier. E., 2014b. Optimization
of Remedial Action of Arsenic in Groundwater Vineland Chemical Co. Superfund Site, New Jersey, Part II: Column Tests
Results, Platform Presentation, 2014- Contaminated Site Management: Sustainable Remediation & Management of Soil,
Sediment and Water, November 17-20, 2014, San Diego, CA.
USEPA, April 1992, Batch-Type Procedures for Estimating Soil Adsorption of Chemicals.
USEPA, September 1989, EPA Superfund Record of Decision, EPA/ROD/R02-89/084
5-1
-------
------- |