United States
Environmental Protection
Agency
National Exposure
Research Laboratory
Las Vegas, NV 89193-3478
Research and Development
EPA/600/SR-97/055 October 1997
4>EPA Project Summary
Field Validation of a
Penetrometer-Based Fiber-Optic
Petroleum, Oil, and
Lubricant (POL) Sensor
William C. McGinnis and Stephen H. Lieberman
Comprehensive comparisons of in situ
measurements from a cone penetrom-
eter-deployed laser induced fluores-
cence (LIF) petroleum, oil, and lubricant
(POL) sensor with traditional field
screening methods were performed. Op-
erational procedures were developed
to facilitate comparison between meth-
ods and across multiple sites. Using a
field screening detect/non-detect crite-
rion, agreement between sensor mea-
surements corresponding to the
sampled interval and the laboratory ana-
lytical measurements on those samples
was better than 85 percent. Compari-
son between measurements from the
two accepted analytical techniques, on
splits of the same sample, was only
slightly better. We conclude that the
LIF-POL sensor, deployed from a cone
penetrometer, provides significant ad-
vantages for subsurface field screen-
ing of POL-contaminated sites. The LIF
technique offers the advantages of
rapid, in situ, real-time measurements,
coupled with increased data density,
not possible with traditional screening
methods.
This Project Summary was developed
by EPA's Environmental Sciences Divi-
sion, National Exposure Research Labo-
ratory, Las Vegas, NV, to announce key
findings of the research project that is
fully documented in a separate report of
the same title (see Project Report order-
ing information at back).
Introduction
The U.S. Environmental Protection
Agency (EPA), Environmental Sciences
Division-Las Vegas (ESD-LV) evaluated
field screening techniques to expedite site
characterization and monitor corrective ac-
tions. In collaboration with the Naval Com-
mand, Control and Ocean Surveillance
Center, RDT&E Division (NRaD), the ef-
fort described here was undertaken to
evaluate the use of a cone penetrometer
system equipped with a fiber optic-based
laser-induced fluorescence (LIF) petro-
leum, oil, and lubricant (POL) sensor for
real-time field screening of subsurface
POL contamination.
The feasibility of using a truck-mounted
cone penetrometer system to push chemi-
cal sensors into the ground to delineate
subsurface contaminant plumes was first
demonstrated through the Department of
Defense (DOD) Tri-Service Site Charac-
terization and Analysis Penetrometer Sys-
tem (SCAPS) program. The LIF cone
penetrometer test (CPT) technology was
developed through a collaborative effort
of the Army, Navy, and Air Force under
the Tri-Service SCAPS program. To sat-
isfy the objective of this Inter-Agency
Agreement (IAG), a comprehensive inter-
comparison effort was established to di-
rectly compare sensor results with
conventional sampling and laboratory
analyses. This effort was proposed as a
jointly funded collaborative effort between
the U.S. EPA and the U.S. Navy and as
such leverages funding provided by Naval
Facilities Engineering Command. This IAG
enabled confirmatory sampling work to
be performed at two sites: Naval Air Sta-
tion, Alameda, CA, and Guadalupe Oil
Field, Guadalupe, CA. To date, confirma-
tory sampling has been performed at 16
sites in addition to the two above listed
-------
sites, therefore summary results will be
presented.
System Description
The SCAPS system uses a truck-
mounted CRT platform to advance its
chemical and geotechnical sensing probe
into subsurface soils. CRT has been
widely used in the geotechnical industry
for determining soil strength and soil type
from measurements of tip resistance and
sleeve friction via an instrumented probe.
The CRT platform provides a 20-ton static
reaction force associated with the weight
of the truck. The forward portion of the
truck-mounted laboratory is the push
room. It contains the rods, hydraulic rams,
and associated system controllers. Un-
derneath the SCAPS CRT push room is
the steam cleaning manifold for the rod
and probe decontamination system. The
rear portion of the truck-mounted labora-
tory is the isolatable data collection room
in which components of the LIF system
and onboard computers are located. The
combination of reaction mass and hy-
draulics can advance a 1-m long by 3.57-
cm diameter threaded-end rod into the
ground at a rate of 1 m/min in accor-
dance with American Society of Testing
and Materials (ASTM) Standard D3441,
the standard for CRT. The rods, various
sensing probes, or sampling tools can be
advanced to depths in excess of 50 m in
naturally occurring soils. As the rods are
withdrawn, grout can be injected through
Vi-in.-diameter tubing within the interior of
the probe's umbilical cable, hydraulically
sealing the push hole. The platform is
fitted with a self-contained decontamina-
tion system that allows the rods and probe
to be steam cleaned as they are with-
drawn from the push hole, through the
steam cleaning manifold, and back into
the CRT push room. Subsurface investi-
gation in this manner produces rinsate
but no soil cuttings as investigation de-
rived waste.
LIF sensors rely on impinging ultravio-
let (UV) light to excite molecular elec-
trons to higher energy states. As the
electrons return to lower energy states,
the transition produces UV fluorescence
photons of longer wavelength than the
UV excitation. The LIF POL sensor probe
consists of a standard penetrometer probe
modified with a 1/4-in.-diameter flush-
mounted sapphire window which is 24 in.
behind the probe tip. Two 500-um silica
clad silica optical fibers, one for laser
excitation and one for fluorescence emis-
sion, are included in the 300-ft umbilical
cable and are internally mounted in the
probe terminating at the sapphire win-
dow. Excitation light at 337 nm, gener-
ated from a pulsed nitrogen laser (0.8 ns
pulse width, 1.4 mJ pulse energy), travels
down the optical fiber and excites fluores-
cence from polycyclic aromatic hydro-
carbons (PAHs) in the soil. The method
detects PAHs in the bulk soil matrix
throughout the vadose, capillary fringe,
and saturated zones. The emission fiber
collects the laser-induced fluorescence
and returns it to the surface.
At the surface, the fiber is coupled to a
spectrograph where the light is spectrally
dispersed. The dispersed light then im-
pinges on an intensified linear photo di-
ode array detector (1024 pixels) which is
gated on for 100 ns at the time of signal
return. An optical trigger from the pulsed
laser via a pulse delay generator is used
to gate the detector. The laser-induced
fluorescence signal is emitted over a broad
range of wavelengths longer than the ex-
citation wavelength. Approximately 16
msec is required to read the fluorescent
signal from a single laser shot. The maxi-
mum spectral resolution is approximately
0.5 nm. The detector is set to measure
the wavelength range from 350 to 720
nm. In practice, the system usually inte-
grates the emission from 20 laser shots
with detector pixels grouped by four. Since
the laser repetition rate is 10 Hz, the total
time to collect a fluorescence emission
spectrum is two seconds. This represents
approximately a 2.5- depth resolution. The
spectral resolution for pixels grouped by
four is approximately two nm. An optical
multichannel analyzer accumulates the
detector readings and reports the sum as
a single measurement to the data acqui-
sition computer.
Data acquisition is automated under
software control using a 486 host com-
puter. The computer sets and controls
the sensor system, stores fluorescent
emission spectra and strain gauge data,
and generates the real-time depth plots.
From the spectral curve at each depth,
the SCAPS software extracts the maxi-
mum intensity and associated peak wave-
length for real-time depth display. SCAPS
standard electrical cone penetrometer in-
strumentation consists of strain gauges
measuring tip resistance and sleeve fric-
tion in accordance with ASTM Standard
D3441. An empirical relationship between
tip resistance and sleeve friction provides
a soil type classification. This data is con-
tained in the real-time display strips as
cone pressure, sleeve friction, and soil
classification. As the probe is forced into
the ground, the real-time display presents
a 10-ft interval on a scrolling basis.
LIF is a nonspecific field screening tech-
nique which detects PAH compounds with
at least two aromatic rings but is most
effective for three and more aromatic rings.
To date, LIF measurements over optical
fibers have not been used extensively for
detection of the single ring aromatic com-
pounds, BTEX (benzene, toluene,
ethylbenzene, and xylenes). The general
trend is the fewer the number of rings the
shorter the excitation wavelength required.
Greater attenuation of this shorter wave-
length UV radiation in optical fibers is a
technological barrier for transmitting the
excitation pulse over long fiber lengths.
Table 1 presents detection limits for
common fuel products found as soil con-
taminants. Measurements were made in
the laboratory on spiked soils over a 50-
m fiber and are reported at the 95 percent
confidence level. Detection limits vary with
fuel type depending on constituent com-
pound abundance. Detection limits also
vary with soil type due to particle size and
mineralogy. Most importantly, these limits
fall well within the range of utility consid-
ering regulatory action limits.
Table 1. Detection Limits for the LIF POL
Sensor
Fuel Type
Soil
Type
Soil A
Sand
Soil B
China Lake
Soil C
Columbus
Unleaded
Gasoline
17
ppm
36
ppm
121
ppm
Diesel
Fuel
#2
329
ppm
25
ppm
83
ppm
Diesel
Fuel
Marine
(DFM)
14
ppm
4
ppm
5
ppm
Sampling Procedure
SCAPS site operations typically con-
sist of two phases: site investigation and
validation. In the investigation phase,
pushes were performed to delineate the
plume boundaries. During the validation
phase, areas of interest were selected
from the first phase and then revisited. At
the selected locations, a validation push
was performed followed immediately by
collection of confirmatory soil samples.
During validation, the SCAPS CPT
pushed the LIF POL sensor probe into
the ground and acquired fluorescence and
geotechnical data. After the probe was
pushed to the total depth anticipated or
was blocked from further penetration, the
probe was retracted. The CPT rig moved
away from the location and a hollow stem
auger (HSA) drill rig was positioned ap-
proximately 20 cm (8 in.) from the push
hole. The HSA rig drilled a hole such that
-------
the advancing auger flights destroyed the
push hole while allowing for the collec-
tion of split spoon soil samples within
approximately 7.5 cm (3 in.) (horizontally)
of the push cavity. This offset between
the push hole and the auger boring per-
mitted sampling far enough apart so that
the soil samples were not affected by
possible cross contamination due to
sloughing down the penetrometer hole,
yet near enough to minimize variability
due to small scale spatial heterogeneities
of the soil and the contaminant distribu-
tion. This sampling strategy ensured that
samples were representative of the re-
gion sampled by the LIF POL sensor.
Each borehole was logged by a geolo-
gist. Soil samples were collected with a
split spoon sampler lined with 15-cm (6-
in.) long stainless steel tubes. The sam-
pler was driven in advance of the lead
auger using a 63.5-kg (140-lb) slide ham-
mer falling over a 75-cm (30-in.) distance,
in accordance with ASTM 1586, the Stan-
dard Penetration Test.
A California modified split spoon sam-
pler was used for sample collection. The
split spoon sampler was a 75-cm (30-in.)
long, 7.5-cm (3-in.) diameter steel tool.
The sampler consisted of a 10-cm (4-in.)
long (reduced to 5 cm (2 in.) when fully
threaded) cutting head or shoe section,
followed by either a 45-cm (18-in.) or 60-
cm (24-in.) long sample barrel containing
three or four 15-cm (6-in.) long stainless
steel soil sampling tubes, and ending in a
waste soils catch barrel section.
Soil samples were collected at depth
intervals to confirm the LIF POL sensor
depth profile in both background and el-
evated fluorescence intervals. The sam-
pler was overdrilled approximately 15 cm
(6 in.) prior to retrieval to reduce the
amount of sloughed soils typically found
in the bottom of the borehole. Only tubes
containing sample soils that appeared
relatively undisturbed were used. Samples
for confirmatory analysis were collected
from the lower and middle (deeper) 15-
cm (6-in.) soil tubes in the 45-cm (18-in.)
sampler. The sample was Teflon-sealed,
capped, taped, labeled, logged, and
placed into a chilled ice chest. Each con-
firmatory sample was analyzed by EPA
Method 418.1 Total Recoverable Petro-
leum Hydrocarbon (TRPH), a water analy-
sis method modified for soil, and EPA
Method 8015-Modified Total Petroleum
Hydrocarbon (TPH). Samples for
geotechnical analysis (soil moisture, grain
size, and density) were sealed and
shipped in the stainless steel tubes re-
trieved from the split spoon sampler.
Those samples chosen for geotechnical
analysis were generally the uppermost
(shallowest) tube of the three from the
split spoon sampler, but only if the tube
appeared full as a result of complete
sample recovery by the split spoon sam-
pler. In each boring, these sampling pro-
cedures were usually repeated four to
eight times to gather samples for tradi-
tional laboratory analytical measurements.
From three to eight validation borings were
performed at a site.
Analytical Methods
EPA method 418.1 (TRPH by infrared
absorption) and EPA method 8015- Modi-
fied (TPH by GC/FID) represent two of the
most frequently used methods employed
for delineating nonvolatile POL contami-
nation. It is important to note that these
analytical methods do not measure ex-
actly the same constituents that are tar-
geted by the LIF POL sensor but were
selected because they represent the tech-
nology that is currently being used on a
day-to-day basis to make decisions about
the distribution of subsurface POL con-
tamination. This data is then compared
with the in situ fluorescence data gath-
ered with the sensor.
Data Analysis
LIF POL sensor data is evaluated on a
detect/non-detect basis to determine per-
centage agreement between sensor data
above or below a fluorescence threshold
and both TRPH and TPH results above or
below a sensor detection threshold.
SCAPS independently provides detect/
non-detect data relative to a specific de-
tection limit derived for a specific fuel
product on a site-specific soil matrix. The
detection limit is determined for the site
by generating a concentration calibration
response curve for a set of calibration
standards (spiked site-specific soil
samples) prepared by standard addition.
Results and Discussion
For an in situ field screening measure-
ment technique, such as LIF, determining
the accuracy of the technique presents a
particular challenge. This is because it is
not a simple matter to confidently assign
a "true" value to a subsurface contami-
nant distribution. With conventional labora-
tory-based measurements, the accuracy
of the method is a function of both the
sampling errors and errors associated with
the measurement method. To evaluate
the accuracy of a laboratory method, the
conventional approach is to compare the
results obtained from analysis of a spiked
sample of known concentration. It should
be recognized, however, that this ap-
proach does not address the issue of
whether the result is an accurate repre-
sentation of the true value of the contami-
nant in the ground. In other words, errors
related to sampling are not addressed.
Because there is no independent mea-
sure of the subsurface value of contami-
nant concentration, it will be necessary to
evaluate the accuracy of the in situ mea-
surement by comparing in situ results with
results from conventional methods that
may not provide a true value of the sub-
surface contaminant distribution because
of errors associated with the sampling
process.
It should be noted that the three meth-
ods for quantifying hydrocarbon contami-
nation discussed in this document (namely
the analytic EPA Methods 418.1 and 8015-
Modified, and the LIF method) all mea-
sure and quantify the amount of
contaminant using a different physical
property of the contaminant. The EPA
Method 418.1 measures the infrared ab-
sorption of the extract from the soil
sample. The EPA Method 8015-Modified
passes the extract from the soil sample
through a gas chromatograph and uses a
flame ionization detector to measure the
contaminant according to the retention
time of the constituents. The LIF method
measures the fluorescence (under laser
excitation) from the PAHs present in the
contaminant.
The two EPA Method measurement
techniques require comparison to a simi-
lar measurement of a target fuel in order
to quantify the contaminant. Note that it is
not possible to ensure that the target fuel
is identical in composition to the contami-
nant extract. The EPA Method 418.1 uses
a single standard hydrocarbon mixture for
quantification, while EPA Method 8015-
Modified quantifies using a target fuel that
produces a similar chromatogram. The
LIF method does not use an extract from
the soil sample, but measures the con-
taminant in situ as it is presented to the
window of the probe. For this reason, the
LIF POL sensor is more sensitive to ma-
trix effects. Because of this matrix sensi-
tivity, the LIF POL sensor does not employ
a target fuel for quantification but only to
set a detection threshold for the site.
Another difference between in situ and
conventional laboratory-based measure-
ments is that laboratory measurements
usually employ extraction or matrix sim-
plification procedures, whereas in situ
measurements offer limited opportunities
for controlling matrix effects. For the LIF
POL sensors, studies have shown that
variability in sensor response results from
changes in the sample matrix and from
variations in fluorescence response re-
-------
lated to fuel product type, age, and origin
of the hydrocarbon contaminant. Since it
is not possible to account for all sources
of variability that affect sensor response
at this time, the sensor is intended to
operate as a field screening method. It
will provide only qualitative data on the
distribution of petroleum hydrocarbon con-
tamination.
The approach for evaluating accuracy
presented here depends on the direct
comparison of in situ sensor data with
the analysis of discrete samples collected
as close as possible to the soil sample
measured by the in situ sensor. Although
it is believed that this approach provides
the best opportunity for evaluating the ac-
curacy of the in situ measurement, it
should be noted that it will not be pos-
sible to account for all variability associ-
ated with the uncertainty in depth from
which the discrete samples are collected.
It is possible that the depth of the dis-
crete sample may be in error by up to 15
cm (6 in.) in the vadose zone. Due to
sloughing and flowing sand conditions in
the water saturated zone, depth measure-
ment uncertainty during discrete sampling
may be greater. In stratified soils, sharp
vertical boundaries of the contamination
plume may exist. This sampling error could
therefore lead to poor comparisons be-
tween in situ data and laboratory data.
For example, due to an error of 15 cm (6
in.) in the sample depth, contaminant con-
centration can change from strongly im-
pacted (greater than 10,000 ppm) to not
impacted (less than 100 ppm). For this
reason, the depth of the sample must be
known for the comparability of the
samples to be firmly demonstrated. In
addition, because there will be several
inches of horizontal offset between the
push location and the location of the split-
spoon sampler, there may also be some
small-scale horizontal variability that will
not be accounted for. Both the vertical
uncertainty and the small-scale horizontal
variability will not be a factor when com-
paring the two laboratory methods be-
cause splits of a homogenized sample
will be measured.
Summary Results for Sixteen
Sites
To date, validation efforts at 16 sites
have been completed. These sites pre-
sented varied hydrogeological conditions
including: (1) arid with deep groundwater
and (2) coastal with tidally-influenced shal-
low groundwater. These sites also pre-
sented various contaminant source
products including old refinery waste, heat-
ing oil, diesel fuel marine, and JP-5 jet
fuel.
Based on the results calculated for the
sites to date, the LIF POL sensor detec-
tion threshold varies somewhat from site
to site, but it is approximately 100 to 300
mg/kg as TRPH by EPA Method 418.1.
Scatter plots of the LIF POL sensor data
versus TPH and TRPH show a trend of
increased fluorescence with increased
TPH or TRPH. Table 2 contains the cu-
mulative contingency analysis results on
a percentage basis showing better than
85 percent agreement between sensor
and analytical measurements. As ex-
pected, comparison between analytical
methods is slightly better since these
measurements were made on splits of
the same sample.
Table 2. Sixteen Site Cumulative Contingency
Analysis Results Summary (n=552)
% % False % False
Comparison Correct Positive Negative
Table 3. MAS, Alameda Contingency Analysis
Results Summary (n=45)
% % False % False
Comparison Correct Positive Negative
LIF vs. TRPH 87
LIF vs. TPH 86
TPH vs. TRPH 95
5 7
7 7
2 3
Naval Air Station, Alameda
Results
SCAPS field operations were under-
taken at Naval Air Station (MAS), Alameda,
CA, Site 13 - Former Oil Refinery, from
the 17th of March through the 6th of April
1994. Validation operations were per-
formed from the 4th through the 6th of
April 1994. A total of 37 pushes were
performed during phase one SCAPS in-
vestigation. After review of phase one
data, eight validation pushes were per-
formed each directly followed by a hollow
stem auger boring and sample collection.
Forty-five samples were collected from
the eight validation borings.
The calculated fluorescence threshold
was 10,620 relative fluorescent counts,
with a corresponding detection threshold
of 137 ppm. Contingency analysis was
performed on the data. The percentage of
false negatives was 7 percent versus
TRPH and 4 percent versus TPH. The
percentage comparability or percentage
correct versus TRPH was 91 percent and
versus TPH was 87 percent. These re-
sults are very favorable considering the
comparison between laboratory analytical
methods with 11 percent false negative
and 87 percent comparability. Table 3
summarizes these contingency results.
LIF vs. TRPH 91
LIF vs. TPH 87
TPH vs. TRPH 87
2
9
2
7
4
11
Guadalupe Oil Field Results
The SCAPS was employed at the
Guadalupe Oil Field in San Luis Obispo
County, CA, for subsurface investigation
of diluent contaminated soils on the 23rd
of August through the 2nd of September
1994. Validation pushes with overborings
and sampling were performed on the 7th
through the 8th of September 1994. A
total of thirty-two SCAPS push holes were
advanced during phase one investigation.
After review of phase one data, an addi-
tional four validation holes were pushed,
subsequently overbored, and a total of 23
soil samples collected. Soil samples were
sent to an analytical laboratory for analy-
sis by EPA Methods 418.1 (TRPH by IR)
and 8015-Modified (TPH by GC/FID).
The calculated fluorescence threshold
was 350 relative fluorescent counts, with
a corresponding detection threshold of 77
ppm. Contingency analysis was performed
on the data. There were no false nega-
tives versus TRPH and 6 percent versus
TPH. The percentage comparability or per-
centage correct versus both TRPH and
versus TPH was 88 percent. These re-
sults are very favorable considering 88
percent comparability between the two
laboratory analytical methods. Table 4
summarizes these contingency results.
Table 4. Guadalupe Oil Field Contingency
Analysis Results Summary
% % False % False
Comparison Correct Positive Negative
LIF vs. TRPH 88
LIF vs. TPH 88
TPH vs. TRPH 88
12 0
6 6
12 0
Conclusions and
Recommendations
The SCAPS technology was developed
to provide rapid, in situ, real-time field
screening of the physical and chemical
characteristics of subsurface soil at haz-
ardous waste sites. The current configu-
-------
ration is designed to quickly, and cost
effectively distinguish hydrocarbon-con-
taminated areas from unimpacted areas.
Although the LIF POL sensor induces only
the PAH portion of the petroleum hydro-
carbons to fluoresce, petroleum hydro-
carbons are the general target analytes.
This capability allows further investigation
and remediation decisions to be made
more efficiently, on site, and reduces the
number of samples that need to be sub-
mitted to the laboratory for costly confir-
matory analyses. A site can then be further
characterized with reduced numbers of
borings or wells placed on a plume spe-
cific sampling pattern rather than a grid.
Remediation efforts can be directed on
an expedited basis as a result of the
immediate availability of the LIF POL sen-
sor and soil matrix data. Further, the
SCAPS CPT platform: (1) allows for the
characterization of contaminated sites with
minimal exposure of site personnel and
the community to toxic contaminants, and
(2) minimizes the volume of investigation
derived waste (IDW) generated during typi-
cal site characterization activities.
As a result of field experience and ef-
forts undertaken to validate the SCAPS
LIF POL sensor, the following items are
recommended.
Further development should be pur-
sued to refine the LIF measure-
ment technique for expanded use
in additional applications such as
monitoring in situ remediation.
Continue research efforts to de-
velop better quantitative aspects in
defining the dominant chemical
source of the fluorescence. Improve
contaminant discrimination by
spectral signature using neural net-
work pattern recognition techniques
and developing a database of fluo-
rescent signatures.
Develop methods to compensate
for matrix effects using additional
sensors and algorithms account-
ing for grain size distribution and
volumetric moisture content.
The U.S. Environmental Protection
Agency, through its Office of Research
and Development, partially funded and
collaborated in the research described
here under Interagency Agreement
#DW17936217 with the U.S. Navy. It has
been subjected to the Agency's peer and
administrative review and has been ap-
proved for publication as an EPA docu-
ment. Mention of trade names or
commercial products does not constitute
endorsement or recommendation for use.
-------
William C. McGinnis and Stephen H. Lieberman are with Control and Ocean
Surveillance Center, San Diego, CA 92151.
Charlita G. Rosa/ is the EPA Project Officer (see below).
The complete report, entitled "Field Validation of a Penetrometer-Based Fiber-
Optic Petroleum, Oil, and Lubricant (POL) Sensor," (Order No. PB98-100472;
Cost: $21.50, subject to change) will be available only from
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at
Environmental Sciences Division
National Exposure Research Laboratory
U.S. Environmental Protection Agency
Las Vegas, NV 89193-3478
United States
Environmental Protection Agency
Center for Environmental Research Information
Cincinnati, OH 45268
BULK RATE
POSTAGE & FEES PAID
EPA
PERMIT No. G-35
Official Business
Penalty for Private Use
$300
EPA/600/SR-97/055
------- |