United Statw
Environmental Protection
Agency
Industrial Environmental R
Laboratory
Cmonnati OH 45268
ch EPA 6OO 2 79 064
March 1979
Rweenti and Dซva4opmซn(
Selected
Methods for
Detecting and
Tracing Hazardous
Materials Spills
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into nine series. These nine broad cate-
gories were established to facilitate further development and application of en-
vironmental technology Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The nine series are:
1 Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
6. Scientific and Technical Assessment Reports (STAR)
7 Interagency Energy-Environment Research and Development
8. "Special" Reports
9. Miscellaneous Reports
This report has been assigned to the ENVIRONMENTAL PROTECTION TECH-
NOLOGY series. This series describes research performed to develop and dem-
onstrate instrumentation, equipment, and methodology to repair or prevent en-
vironmental degradation from point and non-point sources of pollution. This work
provides the new or improved technology required for the control and treatment
of pollution-sources to meet environmental quality standards.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161
-------
EPA-600/2-79-064
March 1979
SELECTED METHODS FOR DETECTING AND
TRACING HAZARDOUS MATERIALS SPILLS
by
Ditmar Bock and Paul Sullivan
Calspan Corporation
Buffalo, New York 14221
Contract No. 68-01-0110 and 68-03-0287
Project Officer
Joseph P. Lafornara
Oil and Hazardous Materials Spills Branch
Industrial Environmental Research Laboratory
Edison, New Jersey 08817
INDUSTRIAL ENVIRONMENTAL RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL. PROTECTION AGENCY
CINCINNATI, OHIO 45268
-------
This report has been reviewed by the Industrial Environmental Research
Laboratory - Cincinnati, U.S. Envionmental Protection Agency, and approved for
publication. Approval does not signify that the contents necessarily reflect
the views and policies of the U.S. Environmental Protection Agency, nor does
mention of trade names or commercial products constitute endorsement or
recommendation for use.
11
-------
FOREWORD
When energy and material resources are extracted, processed, converted,
and used, the related pollutional impacts on our environment and even on our
health often require that new and increasingly more efficient pollution con-
trol methods be used. The Industrial Environmental Research Laboratory -
Cincinnati (lERL-Ci) assists in developing and demonstrating new and improved
methodologies that will meet these needs both efficiently and economically.
This report describes studies conducted to evaluate the use of several
selected off-the-shelf techniques to detect and monitor spills of hazardous
materials in the water environment. Techniques studied included pH, con-
ductivity, colorimetry, ion specific electrodes, catalytic combustion sensors,
and dye tracing. Based on these techniques both a prototype device to give
early warning of spills of heavy metal compounds and a field detection kit to
trace the location of several hazardous materials were built and tested.
This report should be of value to Federal, state and local government
personnel as well as to individuals from the chemical process and trans-
portation industries who are involved in responding to discharges of
hazardous substances. Information on this subject beyond that supplied in
the report may be obtained from the Oil and Hazardous Materials Spills
Branch (IERL), Edison, New Jersey 08817.
David G. Stephan
Director
Industrial Environmental Research Laboratory
Cincinnati
111
-------
ABSTRACT
Detection of hazardous chemicals by a wide range of phenomena including
electrical conductivity, catalytic combustion, and colorimetry was investigated.
This study showed that simple, fieldable instruments are available or can
readily be made available for detecting spills of most common, hazardous
materials at or near the threshold for deleterious biological effects.
Several applicable commercial instruments were identified. A novel apparatus
employing chemical indicators was developed for the early warning of spills
of a wide range of pollutants in natural water bodies. A prototype spill
tracing kit was designed and fabricated for use by laymen and its effective-
ness demonstrated with volunteer firemen as operators.
This report was submitted in fulfillment of Contract Nos. 68-01-0110 and
68-03-0287 by Calspan Corp. under the sponsorship of the U.S. Environmental
Protection Agency. This report covers the period June 30, 1971 to February 25,
1974, and work was completed as of March 20, 1974.
IV
-------
CONTENTS
Foreword iii
Abstract iv
Figures vi
Tables viii
1. Introduction 1
2. Conclusions 3
3. Recommendations 4
4. Early Warning Techniques 5
Conductivity Measurements 5
Specific Ion Electrodes 12
Catalytic Combustion Sensors . 12
Multicolor Transmissometer 16
Cyclic Colorimeter 22
Reaction Chamber 23
Electronics 28
Laboratory Tests 31
Field Evaluation of the Cyclic Colorimeter 31
5. Spill Tracing Techniques 46
pH 46
Odor 47
Indicators 47
Spill Tracing Kit 49
In Situ Colorimetry . 49
Dye Tracer Experiments 52
6. Summary 55
References 56
-------
FIGURES
Number Page
1 Radio-Frequency Oil Sensor 6
2 Solution Impedance vs Frequency 7
3 Early Conductance Meter 8
4 Conductivity Meter Record - Simulated Spill 10
5 Simplified Conductance Meter 10
6 HASP Conductance Meter Mod "C" 11
7 Sulfide Ion Concentration vs Time at Several Sampling Points ... 13
8 Specific Ion Probe Electronics 14
9 "Sniffer" Evaluation Apparatus 15
10 Spectral Discrimination of Several Materials by Use of
Weighting Functions 18
11 Near-Ultraviolet Oil Sensor 20
12 Multispectral Transmissometer 21
13 Cyclic Colorimeter - Block Diagram 22
14 Reaction Chamber Configurations 24
15 Breadboard of Cyclic Colorimeter 26
16 Cyclic Colorimeter Plumbing 27
17 Cyclic Colorimeter Electronics - Block Diagram 28
18 Cyclic Colorimeter Electronics - Schematic Diagram 29
19 Cyclic Colorimeter Calibration for Iron, Nickel, and Zinc .... 32
20 Cyclic Colorimeter Calibration for Lead, Copper, and Manganese . . 33
VI
-------
FIGURES (continued)
Number Page
21 Cyclic Colorimeter Calibration for Cobalt, Mercury, and Cadmium . . 34
22 Cyclic Colorimeter Response 35
23 Cyclic Colorimeter Calibration - Turbidity Immunity 36
24 Hydraulic System 38
25 Sampling Module 39
26 Intake Level Control 41
27 Experimental Cyclic Colorimeter 42
28 Test Site 43
29 Spill and Storm Record 45
30 Colorimetric Discrimination Range 50
31 Spill Tracing Kit 51
32 Colorimetric Monitoring of Treatment Efficacy -
Laboratory Simulation 53
VII
-------
TABLES
Number Page
1 "Sniffer" Response to Group A Chemicals 17
2 pH Indicator 46
3 Smell Intensities 48
4 Attenuation of Dye Fluorescence by Selected Heavy Metal Salts ... 54
Vlll
-------
SECTION 1
INTRODUCTION
For the effective treatment of spills of hazardous materials, it is
imperative that the spill be detected and identified rapidly and that the
geographical extent and concentration of the spilled material be monitored
during the course of the treatment. Rapid detection of the spill facilitates
treatment, which is almost always more effective with the concentrated spills,
and permits reduction in the contact between biotae (including humans) and the
hazardous substance by localized treatment, impoundment, closing of municipal
water intakes, etc. Delineation of the amount and extent of untreated material
permits cost-effective treatment and limits the exposure of the environment to
excess treatment chemicals.
At the present time early notification of a spill depends on the report
of the person responsible or on environmental signs which are obvious to un-
trained observers (gross changes in color, odor, fish behavior, etc.). On
the other hand, in certain high-probability locations, such as harbors and
industrial rivers, arrays of automatic spill detection/alarm systems could be
effective. To maximize the ratio of spill damage prevented to instrumentation
cost, such systems should be broad based, i.e., react to a wide spectrum of
possible pollutants. Furthermore, to be suitable for such applications, instru-
mentation must require little maintenance and be resistant to deterioration in
the hostile environment characteristic of sewers and industrial rivers.
The requirements for equipment and methods of monitoring the extent of
a spill and the progress in its treatment are somewhat different. Such
equipment does not require the same degree of environmental immunity as the
spill alarm probes since it need be deployed only during the treatment of a
spill and need survive only a few hours of environmental exposure. Temporary
deployment also obviates the need for extended periods of maintenance-free
operation and permits rather rapid expenditure of consumables, e.g., indicator
chemicals. Since an operator can be deployed with the equipment, automatic
operation is not necessary and a rather complex signal processing computer
(the operator) is available for interpreting the equipment outputs. Finally,
a few highly portable instruments can cover a rather wide area of possible
pollution sites by being deployed to a previously reported spill site. On
the other hand, this field-monitoring equipment must give semiquantitative
results -- such as an indication of the endpoint of the spill-treatment
titration.and should provide a fairly specific indication of the nature of
the spilled material. Furthermore, although an operator is available, he
may not be skilled in chemistry or electronics. The instrument therefore
should require only the observational skills of the average person, such as
-------
viewing a color change from a colorimetric indicator, or reading a pre-cali-
brated meter.
During 1971-72, Calspan Corporation, under contract to the Environmental
Protection Agency (EPA), undertook a study to select methods and instrumenta-
tion that are commercially available or could be quickly developed to meet
the aforementioned spill alarm and/or field monitoring criteria, with major
emphasis on applications to the detection and monitoring of spills in water-
courses. A wide range of chemical and physical phenomena were explored in
this effort, and are detailed in the body of this report. Among the devices
evaluated were: conductivity meters for ionic solute spills; pH probes for
acid or base spills; specific ion probes; catalytic combustion sensors for
volatile organics; and a multicolor transmissometer for non-volatile
organics. In addition an automatic "cyclic colorimeter", which uses modu-
lation of indicator injection to compensate for turbidity and fouling, was
developed under Contract 68-03-0110 and field tested under Contract 68-03-
0287. Other experiments indicated the effectiveness of dyes as spill tracers
and of the sense of smell as a spill detector. Perhaps the most significant
outcome of the effort was the development of a spill detection kit containing
a conductivity meter, pH indicator, odor samples, and colorimetric reagents
geared to the prescribed spill treatments. This kit has been proven effect-
ive in tests involving volunteer firemen as operators.
-------
SECTION 2
CONCLUSIONS
Electrical conductivity was found to be highly effective in detecting
the presence and measuring the extent of untreated ionic solute spills but
was essentially worthless in indicating the effectiveness of the treatment.
Commercial pH probes and certain other specific ion probes proved to be
highly effective in detecting and measuring spills of acids, bases, and
metallic salts and in monitoring the effectiveness of their treatment.
Volatile organics triggered indications from commercial catalytic com-
bustion sensors while less volatile organics were detected with a multi-
color transmissometer.
Colorimetric indicators proved to be extremely helpful and were used
both in a spill detection kit designed for use by volunteer firemen and in
an automatic "cyclic colorimeter", which uses modulation of indicator
injection to reduce its sensitivity to turbidity and fouling.
Colorimetric indicators were not reliable when added to a water body
(e.g., as a stripe sprayed from a low flying airplane) to delineate the
extent of an acid or base spill.
Experiments indicated the effectiveness of dyes as spill tracers and
of the sense of smell as a spill detector.
Perhaps the most significant outcome of the effort was the development
of a spill detection kit containing a conductivity meter, pH indicator, odor
samples, and Colorimetric reagents geared to the prescribed spill treatments.
This kit has been proven effective in tests involving volunteer firemen as
operators.
The cyclic colorimeter is a useful instrument for field monitoring of
spills. An evaluation of its characteristics showed that it maintained
adequate sensitivity of a few parts per million of heavy metal ion for a
period of about two weeks without maintenance and despite noticeable fouling
due to scale buildup and stream turbidity. However, it proved difficult to
make a simulated spill that would not be suppressed by natural processes
in the creek without spilling unconscionable quantities (tens of moles) of
simulant.
-------
SECTION 3
RECOMMENDATIONS
A kit of instruments and reagents should be developed to detect and
define the location of previously identified spills. The kit should be made
available to appropriate federal, state, local agencies that may be required
to respond to spills of hazardous materials. A short educational movie on
the proper use of the kit should also be produced and made available to these
agencies.
The fact that pH electrodes have been provided with considerable resis-
tance to fouling and other interferences suggests that the same could be done
for other specific ion electrodes, e.g., sulfide, bivalent metal, and cyanide
probes. Some research and development in this area is in progress. This
effort should be encouraged.
Instruments to be incorporated in a device for treating spills of hazard-
ous substances were developed to a point where they could be tested in a
controlled environment. Such tests should be conducted to establish stability
and accuracy of this treatment approach.
A multidetector hazardous spill warning system should be developed to
detect spills of broad classes of compounds.
-------
SECTION 4
EARLY WARNING TECHNIQUES
CONDUCTIVITY MEASUREMENTS
Conductivity is traditionally used as a measure of purity for distilled
and deionized water. Since the conductivity of fresh water in most streams
and lakes is fairly low (50 to 500 S/cm*)(l), a sudden increase in conduc-
tivity indicates the presence of an ionic solute, probably a pollutant.
Conductivity is usually measured with alternating current to avoid interfer-
ence from a plethora of ionic reactions which occur when direct current is
applied to electrodes in water. In the laboratory one uses platinum elec-
trodes whose geometry is calibrated against solutions of known concentration
in a sample container, an AC source operating at a few hundred hertz, and
electronics of varying complexity, depending on the amount of automation one
can afford. For field use, a rugged conductivity meter that uses very little
power and is capable of being read at some distance from the electrodes that
are to be dipped into a stream is desired.
One such instrument, (shown in Figure 1) was developed for use in sewers,
where it discriminates between sewage and oil accumulations in traps. (2)
It is extremely rugged, as attested bv its continuous use for several years
in an inverted syphon in Buffalo, New York. Conductivity is measured by the
load presented to a small radio frequency oscillator immersed in the fluid.
(3) Regrettably, it is most sensitive to variations of conductivity in a
range typical for oil pollutants and very pure water. An attempt during this
study to extend its range to the less pure water in natural waters failed.
Figure 2 presents data on the impedance of one centimeter cubes of
various salt concentrations as a function of frequency. The more rapid
variation in impedance with concentration at lower frequencies indicates that
low frequency measurements should be used if pertinent levels of pollutants
are to be discriminated. Although several commercial instruments can measure
audio-frequency conductivity, none is suitable for spill monitoring. To fill
this gap, a simple, inexpensive, telemetering device was developed. The
instrument (Figure 3) consisted of a unijunction transistor (Qj) used as an
oscillator, a pulse amplifier (Q2)> the dual of a Wien bridge used as a
filter (RiCjR2C2), a silver electrode (UG-1094 connector) which sensed
resistance in the path between it and the instrument case (a one pint tin
can), a rectifier (C3C4D1Do), and a transistor (Q3) which controlled the
frequency of oscillation of the UJT(Qj). This frequency, ranging from 200
*The Siemen (S) equals and replaces the mho unit of conductance.
-------
n
OSCILLATOR
ป 10 MHz
tt
tt
SIGNAL
PROPORTIONAL TO |
iRF POWER LOSS
I
I
__, I
PROTECTIVE ENCAPSULATION
STAINLESS
STEEL
COIL
PLASTIC
FOAM
SEALANT
PLASTIC
PIPE
STAINLESS
STEEL
COIL
Figure 1 RADIO-FREQUENCY OIL SENSOR
-------
to
x
o
LU
O
1
a
IU
a
10 ppm NaCI SOLUTION
50 ppm NaCI SOLUTION
200 ppm...L._
NaCI SOLUTION
! I
600 ppm NaCI SOLUTION
100
FREQUENCY (MHz)
Figure 2 SOLUTION IMPEDANCE VS FREQUENCY
-------
en
SHORTING
PLUG
D
;nj
00
HE
\\-
SENSING PROBE
R2 C2 C3
Figure 3 EARLY CONDUCTANCE METER
-------
to 3000 hertz as a function of resistance, was carried to shore on a two-
wire, transformer-coupled line.
i
Figure 4 shows a record obtained with this conductivity meter during a
full-scale simulation of a heavy metal spill. Five moles of ferrous sulfate
solution (about 280 grams of the salt) had been added to a creek five meters
upstream of the conductivity meter. At the time of the simulation, the slug
was about 25 meters long, 2 meters wide, and 0.2 meters deep. It was esti-
mated that the concentration of Fe++, averaged 28 ppm, ranging between 30
and 100 ppm. Flutter and wow in the small, battery-operated tape recorder
used to obtain this record contributed the broad ribbon of noise, and its
tape skip caused the spikes seen in the figure.
After this test, an improved tape recorder was obtained and the circuit
of the conductivity probe was redesigned. The functions of the pulse
amplifier (Q2) and frequency modulator (Qg) transistors were replaced by
passive elements and the conductivity probe appeared as shown.in Figure 5.
The transistor (Q) operates as a blocking oscillator with a transformer (T)
which also couples output signal to the shore cable. Frequency is determined
by the voltage across the base resistor R3. This voltage is obtained
partially from a prebias resistor R4 and partially from a rectifier, probe,
and Wien bridge dual as in the earlier unit. Total power consumption is
only a few milliwatts from the single 3-V battery. The cost of this unit is
negligible compared to that of the recorder or meter readout ($20) and a
waterproof case with stainless steel electrodes ($10).
A third version of the contacting conductivity probe (Figure 6), using
integrated circuits, was also constructed. Although this circuit is more
complex, containing four active gates rather than one transistor, most parts
have been joined in the integrated circuit, and assembly is very simple.
Power consumption is even lower than for the earlier circuits and a direct
meter readout is practical. This circuit is useful for taking manual read-
ings, typically in tracing an existing spili, while the circuit of Figure 5
is more suitable for recording and telemetry applications, such as for
detecting the approach of a spill at a preexisting sensor. For this reason,
it is the circuit of Figure 6 that has been used in the spill tracking kit
described later in this report.
The efficacy of conductivity as an indicator of pollutants, unfortu-
nately, is not matched by its performance as an indicator of treatment
efficiency. For example, laboratory measurements of conductivity during the
titration of mercuric chloride with sodium sulfide showed no readily detec-
table changes in slope or magnitude either at the correct stoichiometric
ratio for complete precipitation of the mercury as sulfide or at the point
where the mercury redissolves in excess sulfide ion concentration. The
measurements did show, however, that redissolution of the mercury requires a
tenfold excess of sulfide ion concentration over that required for correct
stoichiometric treatment, a rather favorable margin for treatment error.
-------
R
(Kfl)
1 min
2 min
Figure 4 CONDUCTIVITY METER RECORD - SIMULATED SPILL
SENSOR PROBES
Rg *
2.7K.1
c -
.005 uf
r
^
V
w
. \
-
1
"?
fo
2N3640 1 1 2-1N270
O A ^ ^
rx^ ฐ3 R ^ r T
S [Q J R2 Co C, r-hD.
' \\\ 1N270 ./K ,, 3 S ! 1
-+ U\ NH - ^ .1.. II0-1 uf .k-05uf|| ^ nxi ,
. 3V HJ T ^ SW^ (I ' If* LN *
R1 ^ J_ C1 D2
2.2K T "Tb.1 uf
1 S A 1.AAA (
R4
^ 220K
sn T
UT
.05 uf
Figure 5 SIMPLIFIED CONDUCTANCE METER
10
-------
+9
PROBE
1M
M* F
1
560K
2200 pF
HI
-9
?
GND7
4 8.2K
|-vw^Hp\/w-^
0.047^ F 1000 pF
RCA CD 4001 E
QUAD 2 INPUT
NOR GATE
IN914
rฑ:0.1// F
ฑ
_r
1 [IN914
56K 10K
Figure 6 HASP CONDUCTANCE METER MOD "C"
-------
SPECIFIC ION ELECTRODES
Electrodes for pH measurement have advanced from the laboratory to on-
line control capability and have found wide application in pollution monitor-
ing. They have been used in oceanography, where compensation for high ambient
pressure is necessary (4), and nonfouling probes are commercially available.
(5) Specific ion electrodes, which are quite similar, were used in this
program to trace the concentration of sulfide and bivalent metal ions in
spill simulations. Figure 7 shows profiles of sulfide ion taken at four
stations along a stream during sulfide precipitation of the ferrous sulfate
spill referred to earlier. The readings were made visually from commercial
specific ion meters, and later plotted versus elapsed time. All four curves
were made with two instruments; as the trailing edge of the spill passed the
upstream instrument, that instrument was moved to a position downstream from
both the second instrument and the leading edge of the plume.
Data from laboratory simulations indicated that specific ion probes are
quite sensitive to turbulence in the stream flow. Commercially-available
electronics for such probes have fixed internal time constants so short on
some instruments that turbulence noise is excessive and so long on others
that meaningful measurements cannot be made in fast flow situations. In
addition, the commercial instruments are not designed to facilitate or even
permit the use of multiple probes in a single solution. To avoid these
difficulties, new electronics (Figure 8) were designed and constructed with
selectable time constants and truly differential input and output. These
electronics were tested against commercial electronics in a 10 cm wide flow
channel. With a time constant of 0.1 second, the new electronics outper-
formed the various commercial systems in ability to reject common mode and
turbulence noise while responding much more rapidly to real changes in
pollutant level.
CATALYTIC COMBUSTION SENSORS
Several commercial catalytic combustion sensors or vapor "sniffers"
were evaluated as possible detectors of various chemicals chosen to represent
broad classes of pollutants. In one series of experiments, a Johnson-
Williams Model CD 830 was used in the configuration shown in Figure 9 to
sense the vapor of the target chemicals in equilibrium with aqueous solutions
of known concentrations. The sensor, an electrically-heated platinum wire
whose temperature is further raised by the catalytically-induced oxidation
of any combustibles in the sensed atmosphere, was mounted in a beaker. The
beaker was inverted over a second beaker containing distilled water to which
the pollutant was added in steps. Except for an adjustment of the gain
control to its maximum value, the electronics of the detector were not modi-
fied from the "as received" condition. Under these conditions, the output
voltage of the detector exhibited a drift of 0.7 mV per eight-hour "day and a
temperature coefficient of 0.1 mV/ฐK. With these uncertainties, a reasonable
estimate of the detection limit is that concentration which produces a 1 mV
output signal. Since the detector response is linear with aqueous concentra-
tion of the target, the detection limit figure is also the sensitivity in
ppm/mV.
12
-------
100
200
300
400
500 600
TIME (sec)
700
800
900
1000
Figure 7 SULFIDE ION CONCENTRATION VS TIME AT SEVERAL SAMPLING POINTS
-------
OUTPUT
GAIN ADJ.
BIAS ADJ.
M DIVALENT
\UNIVALENT
TIME CONSTANT
ALL RESISTORS 1/8W 1%, VALUES IN OHMS
ALL CAPACITORS MYLAR OR POLYETHYLENE VALUES
AMPLIFIERS ARE FROM ANALOG DEVICES
BIAS
.5V
Figure 8 SPECIFIC ION PROBE ELECTRONICS
-------
CONCENTRATED
SOLUTION
OF TARGET
BURET
BEAKER
DETECTOR
ELECTRONICS
STRIPCHART
RECORDER
THERMOMETER
SENSOR
DILUTE SOLUTION
OF TARGET
MAGNETIC
STIRRER
Figure 9 "SNIFFER" EVALUATION APPARATUS
15
-------
Table 1 presents the observed detection limits for several representa-
tive chemicals and compares them with the respective critical concentrations,
i.e., levels at which behavioral effects are noted in fish. Note that the
detection limit is within a factor of three of the critical concentration
for methyl alcohol and acrylonitrile, so the combustible vapor detector
should serve as a good spill detector for these two chemicals. The factor
of 62 for ammonium hydroxide might be tolerable, but the response is orders
of magnitude too low for the detection of chlorine, phenol, and acetone
cyanohydrin in solution.
Also listed in the table are the time constants of the system, both for
increasing concentration and for flushing with fresh water. These data show
that a spill at the detection limit would be "seen" within two minutes of its
arrival at the sensor. Spills involving higher concentrations would be
detected much more rapidly.
MULTICOLOR TRANSMISSOMETER
A substance can be defined or assigned to a group of similar substances
on the basis of measurements of its transmissivity at several discrete
wavelengths. To accomplish this discrimination, weighting factors are
applied to the transmissivity measurements which are than summed. For the
transmission spectra of the substances and filters involved, matrix analysis
can derive sets of weighting factors which effect optimal separations among
the various substances. Discrimination can be improved by using n sets of
weighting factors to separate the substances in n-space and computerizing
the processing of the transmissometer output.
The effectiveness of such an approach is shown in Figure 10, where two
sets of weighting functions applied to measurements at five different wave-
lengths are used to discriminate among seven substances (water, JP-1, brake
fluid, ethylene glycol, Shell aviation 115, aviation hydraulic oil, and 10W-
30 motor oil) for which transmission spectra were readily available. Each
"cloud" is formed of one point each for the pure target substance, the target
contaminated 10% with each other substance, and the substance contaminated
20% with each other pair of substances. The present computer program
develops five weighting functions, but, as can be seen from Figure 10, two
are sufficient to separate these substances even in the presence of the 20%
contamination.
The optical transmission spectra of the chemicals in Table 1 were
examined to determine the feasibility of identifying these pollutants with
the multispectral detector. Only the phenol spectrum exhibits sufficient
structure to enable its identification by the multispectral technique. How-
ever, colorimetric indicators which produce recognizable spectra for ammonia
and chlorine have been identified which are compatible with unattended
detector operation.
In the case of ammonia, two drops of ammonia stabilizer and one milli-
liter of Nessler reagent were added to 50 milliliters of a 290 ppm solution
of ammonium hydroxide. The test solution immediately turned orange and
slightly cloudy. The spectral transmittance curve of this solution is
16
-------
Table 1
"SNIFFER" RESPONSE TO GROUP A CHEMICALS
Critical* Detection Detection Limit
Chemical Ci
Ammonium
Hydroxide
Chlorine
Methyl alcohol
Phenol
Acrylonitrile
Acetone
Cyanohydrin
:entration
>pm)
5.0
0.03
250
0. 1
15
Limit
(ppm)
310
-3300**
700
8000
45
Critical Concentratic
62
105
' 2.8
io5
3
Response Flushing
Time Time
(sec) (sec)
1-100
2x10
10
90
120
330
60
120
60
90
60
15
15
* Concentration at which behavioral effects are observable
** Response of opposite polarity (depression of combustion)
-------
8-
4-
M
8-
-12-
-16
^ JP-1
k ^
^ A^
A
* A
SHELL AVIATION 115
X
X
X
X WATER
I ฐ-
BRAKE FLUID
10W-30 MOTOR OIL
J QAVIATION HYDRAULIC OIL
ETHYLENE GLYCOL
,
80
88
96
~l 1
104 112
FAC 2 (x 10'3J
120 128
136
Figure 10 SPECTRAL DISCRIMINATION OF SEVERAL MATERIALS BY
USE OF WEIGHTING FUNCTIONS
18
-------
easily distinguished. For chlorine, two drops of orthotolidine reagent were
added to 20 milliliters of sample solution having a chlorine concentration
of 70 ppm. The solution turned orange-pink and its spectral transmittance
curve indicated a clearly distinguishable minimum at approximately 380 nano-
meters .
For acetone cyanohydrin, a test was made using vanillin in an alkaline
solution as a possible indicator. The reaction produced a brown color in the
test solution, but the reaction proceeds slowly unless the temperature of the
sample and the reagent is raised above room temperature. Also, when the
reagent solution containing the vanillin and potassium hydroxide was added to
distilled water as a control solution, the control solution itself was un-
stable and gradually changed color. Although the color reaction in the con-
trol solution was much slower than that observed for the 1000 ppm acetone
cyanohydrin test solution, this particular colorimetric indicator was judged
generally unsuitable for field use. Suitable indicators were not identified
for acrylonitrile and methanol.
The complementarity of the "sniffer" and the multispectral detector is
demonstrated by the above-mentioned data. The "sniffer" is a very good spill
detector for methyl alcohol and acrylonitrile, two chemicals invisible to the
multispectral detector, while it cannot detect phenol, which can be easily
identified by the multispectral detector. To detect phenol and similar
compounds, the development of an unattended multispectral detector was under-
taken.
Some time ago, a two-color transmissometer, shown in Figure 11, was
evaluated by Calspan for detecting oil in sewers. (3) It operated in the
near-ultraviolet and red visible regions of the spectrum. The problems with
implementing this approach in an unattended sensor are fairly obvious; one
cannot expect precision optics to perform well for any duration under water.
This was demonstrated when the two-color transmissometer was operated for
about six weeks in a Buffalo sewer. Using the two colors, it was possible to
detect when fouling had occurred. The oils that were to be detected were
fairly opaque in the near ultraviolet and at least translucent for red light.
If the red light signal that was periodically radio-telemetered to the labo-
ratory at Calspan disappeared, it could be assumed that the device was
fouled. During the second week of the test, scale from hard water fouled the
apparatus. After cleaning the dilute hydrochloric acid it remained usable
for over one month, when the UV emission from the argon lamp became too weak
for further use. During this time, only one oil spill was detected. From
the calibration of the instrument, a concentration of tens of parts per
million was estimated; a sample taken shortly after detection contained
hundreds of parts per million of oil. Improvements in calibration and resis-
tance to fouling were clearly needed. A design that keeps precision optics
out of the water was developed. Figure 12 shows an arrangement with a
diffuse reflector under water and all other optics above the surface. The
cone shape of the reflector enables it to shed sediment through the hole at
the bottom. The argon and neon lamps used in earlier designs have been
replaced with germicidal, fluorescent black-light, and incandescent lamps,
emitting at 254, 360, and about 600 nanometer wavelengths. Although the
device is capable of detecting a wide variety of organics, only its
19
-------
ANTENNA
POWER AND ELECTRONICS
ARGON
LAMP
PHOTO
RESISTOR
WATER OIL
Figure 11 NEAR-ULTRAVIOLET OIL SENSOR
20
-------
ELECTRONICS
Figure 12 MULT1SPECTRAL TRANSMISSOMETER
21
-------
sensitivity to various oils has been demonstrated to date.
CYCLIC COLORIMETER
A major effort of the task on detection and monitoring centered on the
development of the "cyclic colorimeter"* (Figure 13), an instrument designed
TRANSMISSOMETER
DUMP
OUTPUTS
V
DUMP
Figure 13 CYCLIC COLORIMETER - BLOCK DIAGRAM
for automatic detection of pollutants. (7) In this instrument, an indicator
(for example, sodium sulfide) is injected periodically into a flowing sample
of the water to be analyzed. Downstream of the point of injection, varia-
tions in the optical transmittance of the water are observed by means of a
light source and photodetector. When the pollutant is present, cyclic
variations in the optical transmittance of the sample stream occur at the
indicator injection frequency, and are detected by the electronic subsystem.
The electronics can be configured to yield either a quantitative indication
of the pollutant or a simple alarm when a threshold pollutant level is
exceeded. The various subsystems of the colorimeter are discussed in the
following paragraphs.
* U. S. Patent 3,992,109.
22
-------
Three different schemes for indicator injection were considered. In
the original concept of the colorimeter, a pulsatile pump was envisioned as
the indicator injector. Further consideration of the system led to the
conclusion that the same effect could be obtained with much less power
consumption, greater system simplicity, and improved reliability by replacing
the pump with a so.ler.oid valve moduating the flow from a pressurized or
gravity feed indicator reservoir (Figure 14B). Laboratory experiments showed
that any indicator remaining between the valve seat and the outlet to the
sample transport tube at the end of the injection cycle leached into the
sample stream very slowly, lessening the contrast in the sample stream and
degrading the instrument detection limits. To avoid the leaching problem, a
configuration (Figure 14C) in which the drops of the indicator solution fall
through an air space within the reaction chamber to reach the sample stream
was selected. This delivery scheme has the added advantage of a complete
lack of moving parts in the injection system, with a resultant further
improvement in reliability and power consumption.
Reaction Chamber
In the first prototype of the reaction chamber for the free falling
droplet injector scheme, the chamber arm containing the indicator dropper
was sealed to trap an air bubble in the arm. With this configuration, small
air bubbles in the sample stream tended to accumulate in this arm, augmenting
the trapped air bubble. The trapped bubble would grow until the pressure
drop across it overcame surface tension, then a large section would break
off and be carried downstream with the sample flow. This bubble would cause
erratic readings from the transmissometer as it passed through. To eliminate
this problem, the sample injection arm was vented to the atmosphere and the
sample input and output lines were valved. With the valves properly adjusted,
the Bernoulli effect just cancels the sample stream head, and a free surface
at atmospheric pressure can be maintained within the injection arm of the
mixing chamber.
A prototype of the reaction chamber was constructed from a standard,
circular cross section, glass tee, but it was discovered that appreciable
light was conducted from the light source to the detector of the trans-
missometer through the glass wall of the tee without passing through the
sample stream. This stray light, of course, reduced the sensitivity of the
instrument. To avoid the stray light problem, a second chamber with 3/8-inch
square cross section was constructed from plexiglass. Although this chamber
had dimensions similar to those of the circular cross section chamber, it
was discovered that the opacity of the sample stream continued to increase
for several centimeters beyond the output of the reaction chamber. (For
proper operation of the colorimeter, there must be sufficient turbulence
within the chamber to ensure adequate mixing of the indicator and the sample
stream but not so much turbulence that the cloud of reactants spreads
appreciably parallel to the stream flow.) Visual inspection of the flow
conditions within the reaction chamber showed that the mixing within the
square cross section chamber was quite different from that observed in the
earlier, circular cross section chamber. Only a small fraction of the
reagents actually underwent reaction within the chamber and the major portion
of the resultant precipitate flowed out of the chamber without intercepting
23
-------
SAMPLE
INLET
SAMPLE
INLET
INDICATOR
RESERVOIR
LAMP
/VALVE
\w
PHOTORESISTOR
OUTLET
INDICATOR
RESERVOIR
VALVE-
B
LAMP
SAMPLE
INLET
OUTLET
INDICATOR
RESERVOIR
//frv^frfrs^^^fr^vyx/s^sw^x/x/x^^w.
PHOTO RESISTOR
AIR
^~-
WATER
OUTLET
'//////77/////////////////////7///////77/S/.
PHOTORESISTOR
Figure 14 REACTION CHAMBER CONFIGURATIONS
24
-------
the detector beam. When the flow rate was decreased to improve the reaction
efficiency, accumulations of the precipitate rapidly fouled the chamber.
To induce mixing of the sample stream and the indicator and to prevent
the accumulation of the precipitate, the reaction chamberjwas, rotated 90
degrees and used in the configuration shown in Figures 15;and 16. The
combination of the sharp corner in the sample channel and the occlusion
formed where the exit tube joins the reaction chamber induces severe turbu-
lence which ensures that the sample stream and the indicator are well mixed
within the chamber and that the precipitate does intercept the detector beam.
The air bubble prevents back diffusion of the indicator and, thus, increases
the target-background contrast. .With this configuration, the modulation of
the photocell output for a given concentration of pollutant is increased by
about three orders of magnitude over that observed with the earlier configur-
ation. The sample stream flow rate is 0.5 ml/s with this plumbing configur-
ation.
Figure 16 also shows the other significant aspects of the plumbing for
the fieldable cyclic colorimeter. The filtered input sample stream is
delivered to a constant head reservoir. The outflow from the reservoir is
regulated by a. needle valve and fed to the reaction chamber. The reaction
chamber is vented to the atmosphere to prevent accumulation of air in the
chamber and its outflow is regulated by a valve identical to that regulating
the reservoir outflow. Since the valves regulating the input and output from
the reaction chamber are identical, input and output rates can be made to
track with variations in temperature and viscosity by choosing the respective
heads to be equal. Any slight mismatch between the valves causes the fluid
level within the chamber to shift slightly, varying the head and the output
flow to compensate for mismatch. With this configuration, the cyclic color-
imeter remains operable as the input sample stream temperature is varied
between 5 C and 45 C. In one -test, terminated after 72 hours, the colorimeter
operated unattended at an indicator drip rate which would consume 2.5 liters
of indicator per weeka rate certainly compatible with field installation of
the instrument. The plumbing system, therefore, although extremely simple,
is effective and reliable.
After being modulated by the injection of an indicator, the transmis-
sivity of the sample stream must be measured. In keeping with the desired
multipollutant response of an automatic detector, an unfiltered light bulb
and a photocell were used to determine transmissivity. With this system,
absorption in any portion of the optical spectrum will cause an;instrument
response. With sodium sulfide as the indicator, for example,.sulfide pre-
cipitates of any color can be detected. Of course, such a wide optical band-
width system will also see any coloration or turbidity in the sample stream
(mud, algae, etc.) and any deposits on the walls of the reaction chamber
(rust). The first scheme to eliminate the sensitivity to interferents was to
AC couple the electronics to the photocell with a bandpass tuned to the indi-
cator injection frequency. For a strictly AC coupled detector, the output is
proportional to the fraction of the time during which indicator is present
in the sample stream, making a duty cycle of about 50% desirable. With a
synchronous detection system on the other hand, the duty cycle can be reduced
almost arbitrarily without reducing the output signal. Such a reduction in
25
-------
\
Figure 15 BREADBOARD OF CYCLIC COLORIMETER
26
-------
DRIP RATE
VALVE
DRIPPER
SYNCHRONIZING
PHOTOCELL
SIGHT GAUGE
LIQUID LEVEL
SENSING
PHOTOCELL
SODIUM SULFIDE SUPPLY
INPUT SAMPLE STREAM
OVERFLOW
OUTFLOW
RESERVOIR
OUTFLOW VALVE
Figure 16 CYCLIC COLORIMETER PLUMBING
27
-------
duty cycle saves considerable indicator solution and, thus, increases the
field maintenance interval. Furthermore, with the freely dropping indicator
injection system, injection frequency varies with temperature and time, very
likely wandering outside the bandpass of the amplifier. A synchronous
detector, however, can be cued to the falling of the indicator drop (sensed
by another lamp-photocell combination) and maintain arbitrarily small band-
width while tracking the injector frequency. The readout scheme selected is
indicated in Figure 17.
^OUTPUT
Figure 17 CYCLIC COLORIMETER ELECTRONICS - BLOCK DIAGRAM
The absorption signal from the photocell is applied to a logarithmic
amplifier to expand the dynamic range of the instrument. The output of the
logarithmic amplifier feeds a. steerable integrator (8) which accumulates the
absorption during the presence of the indicator with one sign and then sub-
tracts from that signal an equal time integration of the sample stream
absorption without indicator. The integrator is synchronized with the
indicator drop by a lamp-photocell detector and suitable electronics. At the
end of the integration cycle, the integrator .output is .accepted by a sample-
and-hold circuit and becomes the output of the colorimeter. Note that this
output is the difference between the logarithms of the sample stream absorp-
tions with and without indicator, the logarithm of the ratio of these two
absorptions. It is precisely this logarithm which is proportional to the
concentration of the indicated substance by the, Beer-Lambert law.
-; . .' "" J *; ^ _ - ซt ' ;J ; ;
Electronics
The circuit diagram of the cyclic colorimeter electronics appears in
Figure 18 and is discussed in detail in the following paragraphs.
'28
-------
SYNCHRONIZATION
CELL
N)
ABSORPTION
CELL
ALL DIODES IN9U.
ALL CAPACITANCES IN M T. CERAMIC
UNLESS NOTED.
ALL RESISTANCES IN Q. 1/ซW. 10ป
UNLESS NOTED.
POWER FOR AMPLIFIERS AND GATES ปซV.
GATES ARE 1/4 OF THE RCA PACKAGE
CD40-AE,
O OUTPUT
J
Figure 18 CYCLIC COLORIMETER ELECTRONICS - SCHEMATIC DIAGRAM
-------
Consider first the analog portion of the circuit, the lower half of the
schematic diagram. Resistor R^4 and diode 04 maintain a 0.7 V bias across
the pollutant sensing photocell RIS- At constant voltage, the current from
this photocell is nearly proportional to the input light intensity. Since
the operational amplifier (ideally) draws no current and diode 05 is
ordinarily reverse biased, the entire input current becomes the collector
current of (>2. The voltage output from this stage is simply the emitter-base
voltage of Q2, which is proportional to the logarithm of its collector
current (9). A bucking voltage developed by Q3 (supplied with a constant
collector current through R^y) compensates for the major temperature depen-
dence of the emitter-base voltage of Q2ซ In the breadboard colorimeter, Q2
and Q3 are individual 2N3904's. There were some problems with matching the
two transistors adequately and a dual transistor such as the 2N2920 is
recommended for future models. With clear water flowing through the colori-
meter sample channel, RJJ is adjusted to give zero output from the follower
stage, which serves only as a buffer.
The output of the logarithmic amplifier drives an integrator whose
direction of integration is controlled by analog gate Gj2 (8). With gates
G12 and 613 closed and Gj4 open, input current flows only through R2Q and
is accumulated on C^, as in a normal integrator. When Gj2 opens, however,
both input terminals of A3 follow the output from the logarithmic amplifier
and the or;!y input current that flows is through R^g and Rig. Since this
current has the opposite sign from that which flows through R2Q with G]^
closed, the integration continues in the reverse direction. If integration
proceeds in each direction for the same duration, the output from the
integrator at the end of the cycle is proportional to the difference in the
average values of the input over the time interval. This output should be
sensed only with Gj2 closed to avoid picking up an offset due to the instan-
taneous value of the input voltage. 6-^3 turns the integrator on and off and
Gj4 resets it after an integration cycle has been completed. After the
completion of the integration cycle and before reset, the output of the
integrator is passed on to a conventional sample-and-hold circuit which
provides an output approximately proportional to the concentration of the
pollutant.
The appropriate array of pulses for the four analog gates is derived by
the digital circuitry in the upper portion of Figure 21. The passage of an
indicator drop between photocell R2 and its illumination lamp causes a
momentary increase in the resistance of the photocell. The resultant
positive voltage pulse is AC coupled into the base of Qj, turning this
transistor off and allowing the voltage on its collector lead to drop. This
negative pulse sets the integrator sense flip-flop, resets the analog
integrator, and triggers the 60 ms monostable multivibrator. Capacitively-
coupled positive feedback (10) is used to sharpen the fall of the multi-
vibrator output pulse. This pulse triggers a 14 second monostable, which
also has capacitively-coupled positive feedback to sharpen up its transitions.
The output of the slower monostable turns on the integrator which, with Gj2
held open by 62, integrates its input in the noninverting sense. The fall of
the 14 s monostable, and integration resumes in the inverting sense. With
the low output from Gj satisfying one input of GH, the fall of the 14 s
30
-------
pulse triggers Gn and opens G15> updating the output and completing the
cycle.
Laboratory Tests
Calibration curves obtained with the cyclic colorimeter for the eight
most common sulfide-treatable heavy metals and mercury are presented in
Figures 19, 20, and 21. Detection limits are quite good for all the metals
which form darkly colored sulfides. The poor detection limits for the
lighter colored sulfides (ZnS -- white, MnS -- peach, and CdS -- yellow)
could probably be improved by sensing light scattering rather than absorption.
A line of +1 slope has been included on each graph to indicate the
theoretical dependence predicted by the Beer-Lambert law. The deviations
of the actual data from this predicted behavior are not surprising con-
sidering the facts that the reactants are not very well mixed in the reaction
chamber and that they are swept from the chamber in a time not very long
compared to the reaction rates expected. It is interesting to note that
despite the less than ideal chemical conditions, only in the case of the
0.1 M solution of Co++ is the carryover from one cycle to the next sufficient
to cause a decrease in sample-to-background contrast with an increase in
target element concentration.
The immunity of the cyclic colorimeter to background murkiness of the
host stream was demonstrated using Fe++ ions as the target and powdered
carbon as the interferent. Figure 22 shows recorder traces of the output
of the colorimeter for slugs of 0.005 M Fe++ solutions in clear and dyed
water. Note that, not only is the slug of pollutant easily detectable in the
stream having a visibility of 1 cm, but also that there is no offset in the
output baseline due to the carbon. Figure 23 presents calibration curves for
Fe4"1" in clear water as well as- in waters darkened with powdered carbon to
give light transmissions of 50% and 5% centimeter (visibilities of about 4 cm
and 1 cm, respectively).
These laboratory tests showed that the cyclic colorimeter had the
potential of detecting multiple pollutants, even in extremely murky waters.
To verify these laboratory results, a field evaluation was conducted,
Field Evaluation of the Cyclic Colorimeter
The laboratory tests outlined above were successful in detecting low
levels of heavy metals in turbid water. The next logical step was to
determine whether the concept could be implemented in the field. This effort
was performed under a separate contract (No 68-03-0287) and consisted of four
phases: Phase I, apparatus modification during which the laboratory proto-
type was made field-ready; Phase II, field installation during which the
modified cyclic colorimeter was installed at a stream in the Buffalo, New
York area; Phase III, field testing during which spills of a heavy metal
simulant were made to determine whether the system could operate in an
unattended mode for two weeks; and Phase IV, performance evaluation. Details
of this work are presented below.
31
-------
1000
100
o
N>
-.--i ---{ !--<-:-
SLOPE
BEER-LAMBERT
10
,-5
10"
10"
10
,2
10"
CONCENTRATION (MOLES PER LITER)
Figure 19 CYCLIC COLORIMETER CALIBRATION FOR IRON, NICKEL, AND ZINC
-------
1000
100
>
J
H
D
O
SLOPE FROM
BEER-LAMBERT LAW
10"'
CONCENTRATION (MOLES PER LITER)
10"
Figure 20 CYCLIC COLORIMETER CALIBRATION FOR LEAD, COPPER, AND MANGANESE
-------
1000
o-i
100
>
E
Q_
o
SLOPE FROM
BEER-LAMBERT LAW
10
CONCENTRATION (MOLES PER LITER)
Figure 21 CYCLIC COLORIMETER CALIBRATION FOR COBALT, MERCURY, AND CADMIUM
-------
O.OOSM Fe*
CLEAR WATER
-TIME
tn
O.OOSM Fe
IN DYED WATER*
Figure 22 CYCLIC COLORIMETER RESPONSE
-------
800
700
600
~ 500
J
H
Q.
D
0 400
300
200
100
0
O CLEAR WATER
50%/cm (~4 cm VISIBILITY)
5%/cm <^ cm VISIBILITY)
10
Fe++ CONCENTRATION (MOLES PER LITER)
Figure 23 CYCLIC COLORIMETER CALIBRATION - TURBIDITY IMMUNITY
36
-------
Apparatus Modifications
The cyclic colorimeter comprises optical, electronic, and hydraulic
components. Although satisfactory operation in a laboratory environment was
demonstrated in 1971, improvements in electronic and hydraulic structure were
needed for an instrument to be used in the field. The hydraulic structure
that was originally contemplated used a sample stream to which indicator was
added by a valve or pump at intervals sufficiently long to permit development
of the color and flushing of the stream to restore the background condition.
With this approach the large quantity of indicator that was used in a few
days presented storage problems and might lead to incidental pollution.
Indicator injection with a dropper mechanism permitted reduction of the size
of hydraulic components but introduced new problems with drop rate control
and regulation of the sample stream flow rate. These problems were overcome,
and the hydraulic system that finally evolved is shown in Figure 24. It
consists of a reaction chamber where optical transmission is measured by a
lamp and photocell, two level control chambers, an orifice for stream flow
regulation, and a distributed filter to control indicator flow rate. A
second lamp-photocell combination senses the fall of the drop that initiates
the measurement cycle. After the drop falls into the reaction chamber,
transmission with indicator is measured for about twelve seconds, the time
required to flush most of the reactants from the chamber. The background
transmission without indicator is then measured during the interval from
twelve to twenty-four seconds after the drop.
Sample stream flow is controlled by regulating the head, or pressure,
above an orifice feeding the reaction chamber with a simple overflow chamber.
The height of fluid in the reaction chamber is regulated by connecting it
through a passage to another overflow chamber. All effluents are gathered in
a single pipe that returns to the stream. In case a spill is detected all
effluent could be stored for further analysis or reactants might be treated
if a particularly toxic indicator were being used.
Field Installation
The cyclic colorimeter was installed next to a creek about 45 kilometers
south of Buffalo, New York, and received its input via a pipe that extended
from the instrument installation to the central, most turbulent part of the
creek (Figure 25). The pipe (A) about 3 cm in diameter, contained and pro-
tected a smaller tube (D) with an inside diameter of 6 mm which was connected
to a foot valve (B) in the stream and a motor-driven pump (M, E) near the
cyclic colorimeter. The pump was an automotive fuel pump especially treated
for handling insecticides. The return flow is discharged into the 3 cm pipe
and returned to the creek via a saddle tap and a piece of tygon tubing that
trailed downstream and prevented recycling of samples.
Since the modified fuel pump consumed more power than all of the rest of
the system, smaller pumps were tried but found to have insufficient suction
head for operation at low water level. Power consumption was reduced, thus
allowing battery operation during power failures, by cycling the pump to be on
for three to four seconds every thirty seconds. With this duty cycle of
ten percent, the pump flow rate of ten liters per minute was reduced to about
one liter per minute. As a consequence of the cycling, considerable fluctua-
tion in sample stream flow was expected and this led to the inclusion of an
37
-------
INPUT HEAD
REGULATOR
FLOW-
ORIFICE
CHAMBER
HEAD
REGULATOR
SAMPLE
DISTRIBUTED
FILTER
150 MM
REACTION
CHAMBER
100 MM
OUTFLOW
Figure 24 HYDRAULIC SYSTEM
38
-------
CM
Figure 25 SAMPLING MODULE
-------
additional intake level control chamber (Figure 26) that also acts as a
settling basin for coarse sediment. The entire shore installation was
installed in a utility cabinet (Figure 27). A pump, originally, in
the lower left corner, fed the first level control chamberan inverted
plastic bottle--in the upper left corner whence the stream proceeded through
the cyclic colorimeter in the center. Indicator was stored in a Florence
flask in the upper right, while power supply and a recorder occupied the
lower right of the cabinet. A 12 V storage battery was mounted in a separate
case on shore. It was trickle charged and provided the surge of current for
the pump as well as standby power for interruptions of line power of up to
48 hours.
Electronic components were mounted next to the cyclic colorimeter. The
first of two 10x15 cm boards include the drop sensor, a logarithmic
amplifier following the transmission photocell, and an integrator that was
fed directly during the first 12 seconds following a drop and with an invert-
ed input during the following 12 seconds. In this fashion the logarithm of
the ratio of transmission with indicator to that without indicator is calcu-
lated. This number is related to the logarithm of pollutant concentration
by the Beer-Lambert Law. It is stored for display on the recorder and
updated 24 seconds after every drop. Timing is initiated by the drop sensor
and performed in LSI counters by familiar techniques.
Additional amplifiers and electronic switches were mounted on a second
board acting as an interface with a Rustrak recorder. This one-channel
device was modified to record two channels, by sensing motion of the print
bar with a mechanical switch that drove a binary divider or flip-flop. The
flip-flop alternately connected either log transmission (density) signals or
the stored pollutant concentration signal to the recorder meter movement.
As a result, odd-numbered dots represent one signal and even-numbered ones
the other, and since the dots of each moiety merge, two traces were generated.
Further separation of the traces was provided by the use of the raw integral
of the density signal which superposed a sawtooth modulation on it and left
the pollutant record as a simple line.
Field Testing--
The apparatus was installed near the creek in August 1973, removed in
January 1974 (Figure 28)--when the creek froze preventing acquisition of
samples--reinstalled in May 1974, and removed in June 1974. Although some
difficulties were encountered initially the device soon operated reliably,
requiring maintenance only every two weeks. It was calibrated, by injection
of a pollutant simulant, every week and blind tested several times by a
simulated spill of ferrous sulfate.
Initially difficulties were experienced with the dropper control
mechanism, a needle valve that swelled shut in the alkaline indicator solu-
tion. It was replaced with a distributed filter, generated by packing
shredded fiberglass filter material between nylon cross hairs in a tygon
tube. This type of filter, with no other obstacles in the indicator supply
line, reliably produced a drop every forty to seventy seconds, could be
trimmed slightly by squeezing and would last for about two months before
enough glass fibers dissolved in the indicator to necessitate replacement.
40
-------
Figure 26 INTAKE LEVEL CONTROL
41
-------
Figure 27 EXPERIMENTAL CYCLIC COLORIMETER
-------
Figure 28 TEST SITE
43
-------
In addition, an annoying problem arose with the drop detection sensor when it
began to fail intermittently but only between 1 AM and 4 AM with the cabinet
door closed. The problem was traced to AC ripple in the power supply that
caused problems only when there was moisture condensation in the cabinet and
the internal resistance of a Nicad battery was increased by temperatures
below freezing. The electronic circuits were coated with hot paraffin to
resist condensation, a ripple filter was installed, and this problem was
eliminated.
Throughout the evaluation, scale was found to deposit in the reaction
chamber and effluent tubes by the reaction of calcium bicarbonate in the
stream and OH ions in the indicator to form calcium carbonate. The scale
would build up to the point where it interfered with measurements in about
two weeks and had to be removed by flushing about 20 cc of concentrated
hydrochloric acid through the colorimeter. In addition, paper was replaced
in the recorder every week, the system was calibrated with known concentra-
tions of ferrous sulfate solution and battery electrolyte levels and power
voltages were checked.
A few moles of ferrous sulfate in solution were injected into the
creek about 25 meters above the sensor on several occasions. Detection of
these spills on the recorder traces was in each case doubtful, primarily
because concentration at the sensor was in the vicinity of one to three
parts per million, that is, in the noise level. A separate test of the
dissipation of a simulant spill indicated that both turbulence and precipita-
tion of ferric hydroxide decimated the spill too quickly and that one should
have made much larger and potentially harmful spills. A typical spill record
is shown in Figure 29 and compared with the noise produced by the growth and
decay of runoff turbidity before and after a rainstorm.
Performance Evaluation--
The cyclic colorimeter is a useful instrument for field monitoring of
spills. An evaluation of its characteristics showed that it maintained
adequate sensitivity of a few parts per million of heavy metal ion for a
period of about two weeks without maintenance and despite noticeable fouling
due to scale buildup and stream turbidity. It proved difficult to make a
simulated spill that would not be suppressed by natural processes in the
creek without spilling unconscionable quantities (tens of moles) of simulant.
The cyclic colorimeter used in these tests was delivered to EPA at Edison,
NJ, in 1975. A third generation version, modified slightly to facilitate
production, is now available on the commercial market.
44
-------
Figure 29 SPILL AND STORM RECORD
I; .! RELATIVE
"f
ryr DENSITY
LL LOW-HIGH
2
-I"
i r
. .-
0
:;:!":
---
*;'
jt^^L
"'"*>
..J...|..-
, i . . .
-4-
i
1
1
--^pH
':.:.:!' ,7
! 4- .
*"*"- .
1
-
-* -
. .. .
.
:- ' i
' : 4 !":';
r:
<=->-'
"r: .
.....
>
--51
. :' -v
.
-
' ... i : - ^ . .'.
r t- L -1 -J '" ~
:1 .^"H-7.
:--j"~23ฃ
'-rJ " ""7
--"-I
^..,:jT.^v|-~-----.
,-^-^w.
- . .
- 1 .
" \
_ I ""
-.,. ._,
: '-".-^ : "";:.
k-~
"-'."
.:
^--M-"^-
.-: "
_::.
.:.:-..:
-*
:-;::! -::::-J--_-)^-ir -:--
^- "n.Trj^
..-:..
.:" ::
;:"
:--:
l:-r-
-:-
r .-. ,
.u .|\. :
".":.'.:.
.-_-:
-::.:
~'-.'
-. r.
1
t"" '
" " I _
- \ -
ป*ป*t^
."^
L-T-
,
:~. ...-::
t.-J !
~4 -
->-
~::;v:"
RUSTRAK ! ..
;::^r"^r -..:;-' -
.:..-(-::.-.
-.:-
:.~I
~-:-':
'.-."::
'.'~
:;H
.-(r-^r::
--"-'
-----
-:-.-
".-".. '
..: -
":v:
'--
..- .
-.' "-
-
'---
~
^
; -vj
H- r
" ;-
-..-:. .-_-
::.
- ..:
;
-
,
> "-
H~-M>
-. -.'
.:-
.::
-
-
-L-*^.'
SPILL
RELATIVE
-i
1 DENSITY
LOW-HIGH
1-
o
.lii.i::-:! .
-
-^~^J -I
r
1
.- -7
~'r
:-v
.1-- ::
-.:'.
- .:"
;:..::
!____ ^4
i * . i : ' *
I ... - '
' ||V
^^
-.
-:.;
_,.
SI
ST
'L~r
QRI
AR1
.F*US7"rtAKi
'-
VI
rs
:.~TW---
".'-
::
1
j i
w-,'ป.^j ปf*rrr
.-:
""'
-:
7 "Z. '
.." .:-
.." .:/
";
j
:-.--
~.-.
'...::.
.'
... '.-n~-~Li'-':'
:rr~
H.:!
.r.!..'.
"--
v '"
:.;.-
.;._-""
- '"-:.
... .i^ .....
V--.
..." '
~ V^IL 1
' - 1 .-.-
-..'.
'.*
:^-
T\:..:
.:
:- ~
:"-:.'
^_j_ .^-r,"
.vwu
:'.:..j"
"..-.
;-":.
;
:; '
. .. .
~" "
r- -
'.. i^-irti .i^.nrปซ*wfcj-\;-~'^-j"V!>^
.-.
.
~' -.
::.ฃlr
:'".
T'.. .
"' :.':
....
=^-:.
;.:..-
;-!r .
...
..:'.-
.:-. "
-.
"-':'
'~-~ :.:::':::'
,"
! " ' 1
- r -
\
\
i
'
~
":'~..
i
-. '
.":
.;.. i---1
: _ r
^^ .. e- --- ^----^ -~ฎ0^- ซ--._ซ
"_-
j
1
-
.i ปnm. H
.'-'
;
45
-------
SECTION V
SPILL TRACING TECHNIQUES
A review of the history of hazardous spills compiled by EPA and a report
on potential hazards compiled by Battelle Memorial Institute under EPA
contract (6), showed that the many hazardous spills could be detected by pH,
conductivity, odor, or sulfide or chromate precipitation.
pH
In developing a kit for use by laymen (e.g., volunteer firemen) to use
in identifying spilled hazardous materials, a colorimetric indicator for pH
was decided upon. Many citizens are already familiar with such pH measure-
ment through their experience with swimming pools. Bromothymol Blue the most
common indicator, has a very narrow range of sensitivity and one with a more
useful range was sought. Such a pH indicator, superior to several universal
paper indicators in readability and reproducibility, was described by
Yamada (11). It contains four indicators and goes through the colors of the
spectrum in the pH range from 4 to 10 (Table 2). Furthermore, it remains red
for pH less than 4 and violet for pH greater than 10. The change in color
with one pH unit, with the possible exception of the change from pH 9 to 10,
is sufficiently clear to override the masking effects of even fairly murky
water.
Table 2
pH INDICATOR
5 mg THYMOL BLUE
12.5 mg METHYL RED
60 mg BROMOTHYMOL BLUE
100 mg PHENOLPHTHALEIN
(+100 ml H2O. 100 ml EtOH
TITRATE TO GREEN WITH NaOH)
pH
4
5
6
7
8
9
10
COLOR
RED
ORANGE
YELLOW
GREEN
BLUE
INDIGO
VIOLET
46
-------
ODOR
The potential of odor as a hazardous spill indicator was investigated
by a team of Calspan psychologists. They concluded that smell tests are of
little use if the spilled substance is already known by type and quantity,
and are quite dangerous if the fumes are of toxic concentrations. The main
purpose of a smell test kit would be to verify the identify, and perhaps
concentration, of already-smelled odors. In the event that an odor is
detected in the absence of any previous information that a spill has occurred,
the observer may be alerted to a potential problem and initiate action as
warranted. Odors as indicators may also be used to follow the geographic
progress of spilled material in those cases where the dilution of the aqueous
solutions produce fumes of nontoxic concentrations.
Tests were conducted to measure the sensitivity of the sense of smell in
detecting several pollutants. Solutions in various concentrations, of
chlorine, methanol, ammonia, acrylonitrile, and phenol were used to fill
small, covered jars to approximately three-fourths capacity. These jars
were given a coded designation on the lid. For each chemical, three or four
measured concentrations were arbitrarily selected for testing, as well as
several jars containing distilled water to serve as controls. A jar was
shaken briefly and then uncapped for presentation to the subject. The subject
rated the intensity of the odor as none, faint, moderate, or strong (recorded
as 0, 1, 2, or 3, respectively). Random orders of presentation were used for
each subject. Six subjects were presented with phenol (50,100, 300, 1000 ppm
in water), acrylonitrile (100, 300, 1000, 4000 ppm), and four jars of
distilled water. Six different subjects were presented with chlorine (20,
50, 100 ppm), ammonia (5, 10, 50, 100 ppm), methanol (4000, 10,000, 30,000
ppm), and three jars of distilled water.
The results indicated fairly reliable detection of the following con-
centrations :
phenol 100 ppm
acrylonitrile 1000 ppm
chlorine 20 ppm
ammonia 5 ppm
At lower concentrations of these chemicals and at all levels of methanol,
the ratings tended to overlap with ratings given to water. The averaged
results are shown in Table 3.
INDICATORS
The psychologists also undertook a study to define a simple method and
set of colorimetric indicators with which a layman could make a preliminary
identification of heavy metal pollutants. They found that sulfide precipita-
tion could be used to identify a considerable variety of heavy metals and
that precipitation of barium chromate, by using either barium chloride or
potassium dischromate as an indicator, would serve to identify chromate and
barium ions. These particular indicators also show what treatment should be
used to alleviate the hazard. The ranges over which easily observable and
47
-------
Table 3
SMELL INTENSITIES
F*.
00
AVERAGE
RATING
3.0 -
2.5 -
2.0 -
1.5 -
1.0 -
0.5 -
0 -
CHEMICAL
AMMONIA
METHANOL CHLORINE PHENOL ACRYLONITRILE WATER
RELIABLE
X5
X50
X100
1
UNRELIABLE
X30.000
X 10,000
X4.000
X100
X50
X20
XI000
X100
X50
X1000
X4000
X300
X100
X
XX
X
X
X
CONCENTRATIONS ARE IN PARTS PER MILLION.
-------
and readily describable differences in color and opacity allow the discrim-
ination of the concentration of various heavy metals are indicated by the
solid lines in Figure 30. The dotted lines indicate the extensions to the
ranges possible under ideal circumstances. Below the stated concentration
limits, the solution with indicator plus pollutant is indistinguishable from
pure water containing the indicator alone, and above the upper limit the
solution turns so opaque that no furhter changes with concentration are
observable. This study also showed that benzidine dihydrochloride in 5% HC1
could be used as a chlorine indicator with a clearly distinguishable response
at 10 ppb concentration and a laboratory detection limit of 2 ppb.
SPILL TRACING KIT
A pollution detection kit (Figure 31) based on the above studies was
assembled containing smell samples, a pH indicator, the heavy metals
colorimetric indicators, and a conductivity meter of the type shown in
Figure 6. A manual with instructions for using the kit was prepared. It
was designed to require only a minimum knowledge of spill problems, chemistry,
or engineering. A brief explanation of the principles involved in the use of
each component of the kit is followed,by instructions for using the component.
Uses of the various tests are described in the sequence most likely to occur
in the real world situation. A one-hour training presentation was found to
be adequate for instructing a group of 21 volunteer firemen in Golden, New
York, on detecting hazardous spills using: 1) the sense of smell; 2) the
conductivity meter; 3) the pH indicators; and 4) the heavy metal indicators.
Methods for estimating the rate of movement of spills were also covered in
the training session.
Qualitative identification of actual substances is clearly difficult and
should properly be left to a chemist. This does not say that treatment can-
not be initiated with the limited information derived from the above kit.
Frequently, one has an estimate of the quantity of material lost. For
example, many times no more than a few pounds of a poisonous heavy metal
compound are spilled. If the addition of a sulfide precipitation indicator
shows such a compound to be present, one can safely apply a small amount
of treatment. If one still discovers the metal ion downstream, one can
continue with this approach until a chemist and a hydrologist can estimate
the exact quantity of residual treatment needed. The treatment of an acid
or base spill is even simpler. One can add enough neutralizing agent to
bring pH into the range from six to eight. If one considers the complexity
of first aid procedures mastered by fire and police department employees, it
seems likely that fairly simple treatment procedures can be carried out as
described above.
IN SITU COLORIMETRY
The possibility of using colorimetric indicators added directly to the
polluted body of water to assess the extent of the pollutant and the
progress of the treatment was also investigated. It was envisioned that a
stripe of an appropriate indicator solution might be laid down across the
suspected spill area from a low-flying aircraft, and the dimensions of the
spill determined from the aircraft on a second pass at a higher altitude.
49
-------
CONCENTRATION (ppm)
ION
10,000 1000
100
10
Ba
Cr207
Pb
++
-> 1
Figure 30 COLORIMETRIC DISCRIMINATION RANGE
50
-------
Figure 31 SPILL TRACING KIT
5 i
-------
Laboratory experiments showed quite promising results (Figure 32). A field
test with a pH indicator stripe applied from a boat across an industrial out-
fall of pH 4, however, failed to provide adequate contrast to permit the
delineation of the outfall from the ambient pH 7 water. The dark background
of the river bottom made reliable determination of the indicator color
impossible.
DYE TRACER EXPERIMENTS
'In some spill situations, one may be able to add an identifying dye to
the spilled chemical before or shortly after it enters a waterway. The
reactions of dyes which might be used to tag spills of certain heavy metal
compounds were investigated. Table 4 displays the results of an experiment
performed with the indicated chemicals and the dyes, rhodamine B and
fluorescein sodium. One drop of dye was introduced into (1) a 50 ml, 0.1 M
test sample of the chemical, and (2) a 50 ml test sample of water. The two
samples were than compared under an ultraviolet light.
As indicated in the table, the majority of the heavy metal compounds
exhibited only moderate suppression of the rhodamine B and flourescein
sodium fluorescences or showed no apparent change. On the other hand, silver
nitrate and cobalt chloride changed the color of fluorescein sodium and
completely suppressed its fluorescence. On the basis of these observations,
rhodamiiie B is usually preferred as a tracer because of less interference
from the metallic ions involved in the spill.
As little dye as possible should be used to minimize cost and harmful
side effects of the dye, while assuring detection of the spill boundary at
a significant pollutant concentration. Rhodamine B and fluorescein sodium
can be detected visually at dilutions of about fifty parts per billion, and
at lesser concentrations with fluorimeters. To apply one of these dyes to a
chemical that should be detected at a level of one part per million, one
pound of dye would be needed for every twenty pounds of pollutant. If
several barrels of pollutant are spilled, the cost of the tracer is probably
acceptable; if tank car quantities are spilled, it is not.
52
-------
A. PERPETRATING THE SPILL
B. ADDING AN INDICATOR STRIPE
C. THE SPILL SPREADS
D. ADDING THE TREATMENT
'
E. SPILL PARTIALLY NEUTRALIZED
F. ALMOST TOTAL NEUTRALIZATION
Figure 32 COLORIMETRIC MONITORING OF TREATMENT EFFICACY
LABORATORY SIMULATION
.
-------
Table 4
ATTENUATION OF DYE FLUORESCENCE BY SELECTED HEAVY METAL SALTS
SALT
DISODIUM FLUORESCEIN
RHODAMINE B
AMMONIUM MOLYBDATE
ZINC CHLORIDE
SILVER NITRATE
SODIUM CHROMATE
SODIUM ARSENATE
CADMIUM CHLORIDE
MANGANOUS SULFATE
COBALT CHLORIDE
NICKELOUSSULFATE
BARIUM ACETATE
MODERATE
MODERATE
COMPLETE-ARMY DRAB GREEN COLORATION
MODERATE
NONE
MODERATE
MODERATE
COMPLETE-BUFF COLORATION
MODERATE
NONE
MODERATE
MODERATE
NONE
MODERATE
NONE
NONE
NONE
MODERATE
MODERATE
SLIGHT
-------
SECTION 6
SUMMARY
This investigation has shown that it is possible to detect a wide variety
of spilled hazardous materials with simple, fieldable instruments. While
the limits of detection are not always below the limits of toxicity, minor
discrepancies are not critical for two reasons. First, the target of the
automatic detectors is the acute spill. Quoted toxicity limits are usually
based on prolonged exposure and are usually lower than the minimum concentra-
tions damaging in acute exposure. Detection of chronic pollution near the
toxicity level can be implemented by more sensitive laboratory determinations
made periodically on a sampling basis. Second, the detectors are to be
placed in likely spill locations where they will intercept the spill before
it undergoes much dilution. Little of the environment will experience the
pollutant at or near this concentration.
Simple and effective procedures to enable laymen to identify spilled
hazardous materials have also been developed and demonstrated. Only minimal
instruction in the performance of these procedures is required.
Finally, dyes suitable for tracing heavy metal spills have been sug-
gested.
Improvements in detection limits, equipment reliability, maintenance
intervals, and economy are certainly desirable in many instances, but a
quite significant start has been made in providing the technology for a
nationwide hazardous spill detection network.
55
-------
REFERENCES
1. Standard Methods for the Examination of Water and Wastewater, American
Public Health Association, Inc., New York, 1966.
2. Bock, D.H. and Eckert, E.H., "Detection of Oil in Sewers," IEEE
Transactions on Geoscience Electronics, 119, GE-10, April 1972.
3. Olson, R.A. and Lary, E.G., "Electrodeless Plasma Conductivity Probe
Apparatus," Review of Scientific Instruments, 12, 33, December 1962.
4. Durst, R.A., "Ion-Selective Electrodes in Science, Medicine, and
Technology," American Scientist, 3, 59, May-June 1971.
5. e.g., from Leeds ง Northrup, North Wales, Pennsylvania.
6. Dawson, G.W., Shuckrow, A.J., and Swift, W.H., Control of Spillage of
Hazardous Polluting Substances, Batelle Memorial Institute, FWQA,
Department of the Interior, Program 1508, November 1970.
7. Bock, D.H., Cyclic Colorimetery Apparatus, Calspan Patent Disclosure
Number 1146, December 1970, U.S. Patent 3,992,109.
8. Sullivan, P.P., Steerable Integrator, Calspan Patent Disclosure
Number 1241, June 1972.
9. Gibbons, J.F. and Horn, H.S., "A Circuit with Logarithmic Transfer
Response Over 9 Decades," IEEE Transactions on Circuit Theory, 378,
CT-11, September 1964.
10. Sullivan, P.P., Chatterless Zero-Crossing Detector, Calspan Patent
Disclosure Number 1172, June 1971.
11. Lange, B., Kolorimetrische Analyse, Verlag Chemie, Berlin, 1944.
Also: Foster, L.S. and Grundfest, I.J., "Demonstration Experiments
Using Universal Indicators," Journal of Chemical Education, 274,
14, 1937.
56
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
REPORT NO.
EPA-600/2-79-064
3. RECIPIENT'S ACCESSION>NO.
TITLE AND SUBTITLE
Selected Methods for Detecting and
Tracing Hazardous Materials Spills
5. REPORT DATE
March 1979 (issuing date)
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
D. Bock and P. Sullivan
8. PERFORMING ORGANIZATION REPORT NO.
S. PERFORMING ORGANIZATION NAME AND ADDRESS
Calspan Corporation
P.O. Box 235
Buffalo, New York 14221
10. PROGRAM ELEMENT NO.
IBB610
11. CONTRACT/GRANT NO.
68-01-0110 & 66-03-0287
12. SPONSORING AGENCY NAME AND ADDRESS
Industrial Environmental Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Cincinnati. Ohio 45268
13. TYPE OF REPORT AND PERIOD COVERED
Final
14. SPONSORING AGENCY CODE
EPA/600/12
15. SUPPLEMENTARY NOTES
16. ABSTRACT ~~
Detection of hazardous chemicals by a wide range of phenomena including
electrical conductivity, catalytic combustion, and colorimetry was investigated.
This study showed that simple, fieldable instruments are available or can
readily be made available for detecting spills of most common, hazardous
materials at or near the threshold for deleterious biological effects.
Several applicable commerical instruments were identified. A novel apparatus
employing chemical indicators was developed for the early warning of spills
of a wide range of pollutants in natural water bodies. A prototype spill
tracing kit was designed and fabricated for use by laymen and its effective-
ness demonstrated with volunteer firemen as operators.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS
COSATI Field/Group
Hazardous Materials, Detection, Kits,
Chemical Analysis, Water Pollution,
Water Analysis
Spills Tracing,
Spills Detection
68D
8. DISTRIBUTION STATEMENT
Release to Public
19. SECURITY CLASS (This Report)
Unclassified
21. NO. OF PAGES
65
20. SECURITY CLASS (Thispage)
Unclassified
22. PRICE
EPA Form 2220-1 (9-73)
57
USGPO: 1979 657-060/1631
------- |