MEMBRANE INTERFACE PROBE PROTOCOL FOR CONTAMINANTS IN LOWPERMEABILITY ZONES
David Adamson, Steven Chapman, Nicholas Mahler, Charles Newell, Beth Parker, Thomas
Sale, Seth Pitkin, Michael Rossi, and Mike Singletary
SUPPORTING INFORMATION
Groundwater Data Collection Methods
To complement the MIP and soil data, groundwater data were collected (see Figure S.2 for
results). Groundwater samples were collected at all six of the MIP locations (Figure 1) using
one or both of the following methods:

WaterlooAPS
TM
:
Stone Environmental Inc.’s proprietary subsurface data acquisition
system was used to collect both discrete-depth groundwater samples and an integrated
set of companion data in a single, continuous direct push. Groundwater samples were
collected by pumping water through screened ports located in the profiler tip; with
stainless steel lines leading from the downhole tooling to 40-mL VOA vials for CVOC
analysis using EPA Method 8260 at Stone’s off-site analytical laboratory. Profiling was
completed at all six locations where MIP screening was completed, and a total of 56
samples were collected from these locations. However, due to system limitations in
collecting water from finer-grained soils, no samples were collected from the lowerpermeability clay intervals that are the focus of the current study. During advancement,
this system was also used to generate real-time relative k data in the form of an index of
hydraulic conductivity (Ik) at all locations.

Groundwater Sampling using Geoprobe SP16 and Temporary Piezometers:
At
three locations where MIP screening was completed, groundwater was collected from at
least 3 depths within the lower permeability zone at each location, along with at least 1
sample from both the higher permeability sands above and the below this interval. In
higher-permeability zones, Screenpoint (SP) groundwater sampling, consisting of the
installation of protected screens within standard Geoprobe rods with an expendable
drive point, was used. Once reaching the desired depth, the screen is held in place with
extension rods while the drive rods are retracted. For this project, the SP16 system
custom screens (1.0-in PVC, 2.5-ft long, 0.010-in slot size) was used. At intervals with
lower-permeability soils, temporary piezometers with similar screen characteristics were
instead installed to provide more adequate time for well development. Using these two
methods, groundwater was collected from a total of 6 to 7 different depths at each
location. Groundwater was collected from each of the points using a peristaltic pump
after sufficient water was present. All CVOC analyses were completed at the University
of Guelph.
QA/QC Procedures
QA/QC checks were made throughout the MIP sampling program, including. pre- and postresponse tests. These involved measuring detector response (trip time and total response)
when the probe is placed in a solution of known concentration (in this case PCE at 1 mg/L for
ECD response and benzene at 5 mg/L for PID response). Although the MIP was operated using
non-standard conditions (e.g., varying flow rates, drive rates, and temperatures), nearly all of
the pre- and post-response tests were within standard acceptance criteria (greater than 150 mV
for ECD and greater than 2 mV for the PID). Parameters such as flow rates, temperature and
pressure were monitored continuously and were also mostly within acceptance criteria.
Soil sample QA/QC included collection of field duplicates (minimum 5%), field blanks (methanol
blanks to identify pre-existing contamination in methanol used for soil sample preservation and
extraction, trip blanks and equipment blanks), lab storage blanks (used to replace trip blanks
which were analyzed on arrival at the lab, and kept with the samples during the extraction
period), lab method blanks, lab duplicates (minimum 5%), continuing calibration checks, matrix
spikes and matrix spike duplicates.
Data Normalization Procedures
The statistical comparisons of MIP data to CVOC concentrations in soil required establishing a
quantitative method for assessing the “accuracy” of the MIP contaminant distributions under the
various operating conditions. Of the various statistical methods that could be appropriate for
this type of evaluation, two relatively simple methods were chosen. The first was to calculate
the relative percent difference (RPD) between soil concentrations and the MIP
“concentrations” at corresponding depths. The following stepwise procedure was employed at
each location were MIP data was collected:
1. At each depth, the concentrations of individual constituents were summed to obtain a
total CVOC concentration from the soil data. The total CVOC concentration served as
the “known” value (i.e., assumed to be the accurate benchmark).
2. The total CVOC concentration data was log transformed to be more consistent with its
expected distribution.
3. At each depth, the MIP ECD and PID values were converted to the geomean of values
obtained across the nearest 1-ft interval (e.g., the equivalent MIP signal at 20 ft bgs was
calculated by taking the geomean of all MIP signals measured from 19.5 ft to 20.5 ft).
This ensured a more representative dataset and minimized signal abnormalities that
might have occurred at a single depth.
4. Each MIP dataset (ECD and PID) was log-transformed and then adjusted to account for
the baseline signal (i.e., the minimum response for the detector over the interval being
characterized).
5. The MIP datasets were then normalized using the maximum signal response at any
depth at that location. This established a set range for each dataset between zero (no
response) to one (maximum response) across the entire depth interval.
6. A similar normalization procedure was followed on the depth-discrete total CVOC
concentrations obtained from the soil data.
7. The RPD between the normalized MIP data and the normalized soil CVOC
concentration was then calculated at each depth. Values below zero indicated the MIP
data generally underestimated the soil concentration, while values above zero indicated
that the MIP data overestimated the soil concentration. The median RPD value across
the entire depth interval was used as an overall indicator of accuracy.
directional median RPD.
This is the
A second metric—the non-directional median RPD—was
obtained by finding the median of the absolute value of the individual RPD calculations.
The non-directional median RPD disregards signs, such that it is a measure of the
overall variability associated with the dataset.
Table S.1 Matrix for MIP Operating Conditions and Rationale
High Concentration Area
Run ID
Operating Conditions Rationale
Heated Trunkline (HTL) Baseline
MIP Flow = 40 ml/min.
Drive rate = 1 foot/min.
Probe temperature set to
120 degrees C
PEEK
MIP Flow = 40 ml/min.
Drive rate = 1 foot/min.
Probe temperature set to
120 degrees C
HTL Up-Log - Probe pulled
MIP Flow = 40 ml/min.
from bottom of hole to the top. Drive rate = 1 foot/min.
Probe temperature set to
120 degrees C.
HTL system deployed to determine if the heated and non-sorbing stainless steel carrier gas line would
allow the MIP to be more effective at mapping the contaminant distribution and relative concentrations
than that of the PEEK system.
Fast Drive Rate
MIP Flow = 40 ml/min.
Drive rate = 2 foot/min.
Probe temperature set to
120 degrees C
By increasing drive rate through potentially heavily contaminated zones, there would be less time
available for the system to become overloaded with contamination.
High Flow Rate
MIP Flow = 80 ml/min.
Drive rate = 1 foot/min.
Probe temperature set to
120 degrees C
By increasing the flow rate, via an increase in delivery pressure, it was hypothesized that 1) there
would be less of a pressure gradient for driving contaminants into the tool and 2) the contaminants that
did enter the tool would be more diluted than normal due to the higher volumetric flow rate.
Low Temperature
MIP Flow = 80 ml/min.
Drive rate = 1 foot/min.
Probe temperature set
to 100 degrees C
By decreasing probe temperature, the amount of VOC volatilization would decrease and the zone of
influence would also likely decrease. These two factors would likely result in less contaminant mass
entering the gas phase and being available for migration across the membrane.
By conducting the MIP in reverse, from the bottom up, the true bottom of the contamination could be
determined. MIP detector signals should be at or near basline conditions prior to starting uplogging
procedure.
Low Concentration Area
Run ID
Operating Conditions Rationale
Heated Trunkline (HTL) Baseline
MIP Flow = 40 ml/min.
Drive rate = 1 foot/min.
Probe temperature set to
120 degrees C
PEEK
MIP Flow = 40 ml/min.
Drive rate = 1 foot/min.
Probe temperature set to
120 degrees C
HTL system deployed to determine if the heated and non-sorbing stainless steel carrier gas line would
allow the MIP to be more effective at mapping the contaminant distribution and relative concentrations
than that of the PEEK system.
HTL Up-Log - Probe pulled
MIP Flow = 40 ml/min.
from bottom of hole to the top. Drive rate = 1 foot/min.
Probe temperature set to
120 degrees C.
By conducting the MIP in reverse, from the bottom up, the true bottom of the contamination could be
determined. MIP detector signals should be at or near basline conditions prior to starting uplogging
procedure.
Slow Drive Rate
MIP Flow = 40 ml/min.
Drive rate = 0.5
foot/min. Probe
temperature set to 120
degrees C.
By decreasing the drive rate through the lesser contaminated zones, there would be more time
available for the volatilization and hence more gas phase contamination available for migration into the
tool.
Low Flow Rate
MIP Flow = 20 ml/min.
Drive rate = 0.5 foot/min.
Probe temperature set to
120 degrees C.
By decreasing the flow rate, via an decrease in delivery pressure, it was hypothesized that 1) there
would be more of a pressure gradient for driving contaminants into the tool and 2) the contaminants
that did enter the tool would be less diluted than normal due to the lower volumetric flow rate.
High Temperature
MIP Flow = 20 ml/min.
Drive rate = 1 foot/min.
Probe temperature set
to 140 degrees C.
By increasing probe temperature, the amount of VOC volatilization would increase and the zone of
influence would also likely increase. These two factors would likely result in more contaminant mass
entering the gas phase and being available for migration across the membrane.
.
Table S.2. Evaluation of Accuracy of MIP Data Relative to Soil CVOC Concentrations:
Linear Regression of Normalized Datasets
R2 for Regression
ECD
Dataset
PCE+TCE
High Concentration Location (OU3-3)
Heated Trunk Line (Baseline)
PEEK Trunk Line
Uplogged Data
Low Temperature
Fast Drive Rate
High Flow Rate
Combined Data from All Runs
PID
PCE+TCE+DCE
PCE+TCE
PCE+TCE+DCE
0.11
0.25
0.25
0.07
0.09
0.09
0.09
0.02
0.004
0.48
0.06
0.04
0.03
0.00
0.47
0.43
0.006
0.52
0.44
0.37
0.23
0.59
0.21
0.14
0.22
0.32
0.35
0.23
0.20
0.23
0.15
0.34
0.25
0.18
0.13
0.02
0.04
0.32
0.03
0.02
0.01
0.01
0.22
0.17
0.06
0.29
0.28
0.33
0.12
0.07
0.22
0.20
0.07
0.12
0.16
0.09
0.28
0.03
0.04
0.21
0.62
0.29
0.41
0.02
Low Concentration Location (OU3-6)
Heated Trunk Line (Baseline)
PEEK Trunk Line
Uplogged Data
Low Temperature
Fast Drive Rate
High Flow Rate
Combined Data from All Runs
Other Locations
OU3-4-PK (Baseline)
OU3-5-PK (Baseline)
Table S.3. Evaluation of Accuracy of MIP Data Relative to Soil CVOC Concentrations:
Relative Percent Difference Between Normalized Datasets
Relative Percent Difference (%)
ECD
PID
PCE+TCE
PCE+TCE
PCE+TCE+DCE
Median
(Non-Directional)
Median
(Directional)
Median
(Non-Directional)
Median
(Directional)
Median
(Non-Directional)
Median
(Directional)
Median
(Non-Directional)
Median
(Directional)
61
55
52
56
57
56
56
47
55
52
40
50
36
47
62
47
52
56
60
56
56
41
42
52
30
41
26
39
39
18
69
47
56
129
57
-2
-11
-67
-44
-41
-118
-39
39
18
67
47
51
118
53
-14
-15
-67
-46
-48
-118
-47
109
116
121
93
113
145
121
57
87
-23
65
79
82
65
42
40
102
40
54
38
48
13
8
-90
-37
24
5
-9
160
103
200
99
100
112
143
-54
23
-134
19
40
37
14
156
90
193
95
106
110
128
-149
-90
-185
-95
-57
-110
-113
44
93
17
52
36
57
13
10
50
76
-40
22
48
76
-48
-76
Dataset
High Concentration Location (OU3-3)
Heated Trunk Line (Baseline)
PEEK Trunk Line
Uplogged Data
Low Temperature
Fast Drive Rate
High Flow Rate
Combined Data from All Runs
PCE+TCE+DCE
Low Concentration Location (OU3-6)
Heated Trunk Line (Baseline)
PEEK Trunk Line
Uplogged Data
Low Temperature
Fast Drive Rate
High Flow Rate
Combined Data from All Runs
Other Locations
OU3-4-PK (Baseline)
OU3-5-PK (Baseline)
Table S.4. Evaluation of Accuracy of MIP Data Relative to Soil CVOC Concentrations:
Impact of Data Corrections on Linear Regression
2
R for Regression
ECD
Dataset
Data Correction Applied
High Concentration Location (OU3-3)
All MIP Runs
None
Corrected with Uplogged Data
PID
All Soil Types
Low Permeability Soils
Only
All Soil Types
Low Permeability
Soils Only
0.09
0.21
0.004
0.002
0.23
0.26
0.25
0.25
0.13
0.20
0.001
0.09
0.09
0.12
0.03
0.16
Low Concentration Location (OU3-6)
All MIP Runs
None
Corrected with Uplogged Data
(A)
Location OU3-4
Depth (m bgs)
Soil Total CVOC Concentration
(mg/kg)
MIP ECD
(mV)
MIP PID
(mV)
Electrical Conductivity
(mS/m)
Approximate boundary of
low k zone based on soil lithology
< 35, sand/gravel
35-70, silty sand
70-105, sandy silt
105-140, clayey silt
> 140, silty clay
Soil Total CVOC Concentration (mg/kg)
(B)
Location OU3-5
Depth (m bgs)
Soil Total CVOC Concentration
(mg/kg)
MIP ECD
(mV)
MIP PID
(mV)
Electrical Conductivity
(mS/m)
Approximate boundary of
low k zone based on soil lithology
< 35, sand/gravel
35-70, silty sand
70-105, sandy silt
105-140, clayey silt
> 140, silty clay
Soil Total CVOC Concentration (mg/kg)
Figure S.1. Baseline MIP Characterization Data vs. Soil CVOC Concentration Data at Location OU3-4 (top panel) and Location OU3-5
(bottom panel). (A) CVOC Concentration Data (including by-products); (B) ECD Data; and (C) PID Data; (D) Electrical Conductivity Data.
(A)
Location OU3-3
Index of Hydraulic
Conductivity (Ik)
Soil Lithology
Depth (m bgs)
Soil CVOC Concentration (mg/kg)
Groundwater CVOC Concentration (mg/L)
(B)
Location OU3-4
Index of Hydraulic
Conductivity (Ik)
Soil Lithology
Depth (m bgs)
Soil CVOC Concentration (mg/kg)
Groundwater CVOC Concentration (mg/L)
Figure S.2. Groundwater and Soil Data Collected at (A) Location OU3-3; and (B) Location OU34. First panel shows total CVOC concentration data collected using Geoprobe SP16 and
Temporary Piezometers vs. Waterloo APS (WP) vs. Soil cores. Second panel shows soil and
groundwater (WP) concentration data for individual CVOCs. Third panel shows index of
hydraulic conductivity data collected in real-time by Waterloo APS. Fourth panel shows soil
lithography based on United Soil Classification System (SP = sand (poorly graded); CL = clay
(inorganic with low plasticity; SP/CL = sand/clay mix, SC = clayey sands; NC = not collected).
Other groundwater analyses (field and geochemical parameters) collected but not shown.
Geoprobe SP16/temporary piezometers not used to collect groundwater at location OU3-4.
(C)
Location OU3-5
Index of Hydraulic
Conductivity (Ik)
Soil Lithology
Depth (m bgs)
Soil CVOC Concentration (mg/kg)
Groundwater CVOC Concentration (mg/L)
(D)
Location OU3-6
Index of Hydraulic
Conductivity (Ik)
Soil Lithology
Depth (m bgs)
Soil CVOC Concentration (mg/kg)
Groundwater CVOC Concentration (mg/L)
Figure S.2. Groundwater and Soil Data Collected at (A) Location OU3-3; and (B) Location OU34. First panel shows total CVOC concentration data collected using Geoprobe SP16 and
Temporary Piezometers vs. Waterloo APS (WP) vs. Soil cores. Second panel shows soil and
groundwater (WP) concentration data for individual CVOCs. Third panel shows index of
hydraulic conductivity data collected in real-time by Waterloo APS. Fourth panel shows soil
lithography based on United Soil Classification System (SP = sand (poorly graded); CL = clay
(inorganic with low plasticity; SP/CL = sand/clay mix, SC = clayey sands; NC = not collected).
Other groundwater analyses (field and geochemical parameters) collected but not shown.
Figure S.3. Relative Percent Difference between Baseline MIP Characterization Data and Soil
CVOC Concentration Data at Location OU3-3. PID data are compared to sum of PCE, TCE,
and cis-1,2-DCE soil concentrations. ECD data are compared to sum of PCE and TCE soil
concentration
Response Test Summary Results
Large Change in Response between compounds,
Minimal response to c-DCE
Response/as Rise in Baseline
Compound
PID
ECD
c-DCE
19
22
TCE
23
690
PCE
16
1300
Notes:
1. Response Test solutions comprised of 1 mg/L of
study compound.
Small Change in Response between compounds,
Response to all compounds very similar
FIGURE S.4. Sensitivity of MIP Detectors for Individual CVOCs Present at NAS Jacksonville OU3. Tests were completed similarly to a
standard response test for MIP operation (i.e., the probe is submerged in a container with a known quantity of the study compound (1
mg/L)) to account for diffusion across the membrane. All tests were performed twice.