600281128
DETECTION OF OIL IN WATER BY A FLAME EMISSION TECHNIQUE
by
Manfred Pragar
NUCOR Corporation
2 Richwood Place
Denville, New Jersey 07834
and
D. Stainken
U.S. Environmental Protection Agency
Edison, New Jersey 08837
Contract No. 68-03-0205
Project Officer
U. Frank
Oil and Hazardous Materials Spills Branch
Municipal Environmental Research Laboratory-Cincinnati
Edison, New Jersey 08837
MUNICIPAL ENVIRONMENTAL RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
CINCINNATI, OHIO *5268
-------
DISCLAIMER
This report has been reviewed by the Municipal Environmental Research
Laboratory, U.S. Environmental 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.
-------
FOREWORD
The U.S. Environmental Protection Agency was created because of increasing
public and government concern about the dangers of pollution to the health and
welfare of the American people. Noxious air, foul water, and spoiled land are tragic
testimonies to the deterioration of our natural environment. The complexity of that
environment and the interplay of its components require a concentrated and integrated
attack on the problem.
Research and development is that necessary first step in problem solution; it
involves defining the problem, measuring its impact, and searching for solutions. The
Municipal Environmental Research Laboratory develops new and improved technology
and systems to prevent, treat, and manage wastewater and solid and hazardous waste
pollutant discharges from municipal and community sources, to preserve and treat
public drinking water supplies, and to minimize the adverse economic, social, health,
and aesthetic effects of pollution. This publication is one of the products of that
research and provides a most vital communications link between the researcher and
the user community.
This report documents the development of a flame emission technique and
instrument that can measure low concentrations of oil in oil-contaminated water. The
technique, instrumental design, and recommendations for future development are
reported. The flame emission instrument developed in this report was proven
potentially useful as a remote detector for petroleum oils. This report will be of
interest to those individuals involved in routine monitoring of the environment, law
enforcement, and industrial wastewater research. Further information about this
report is available from the Oil and Hazardous Materials Spills Branch, MERL-Ci,
Edison, New Jersey 08837.
Francis T. Mayo, Director
Municipal Environmental Research Laboratory
Cincinnati
-------
ABSTRACT
A flame emission technique and basic instrument design is presented for
measuring low concentrations of oil in oil-contaminated water. The flame emission
instrument developed in this report would be useful as a detector for petroleum oils.
The flame emission technique utilizes the selectivity of the hydrocarbon emission
signal (at 431 nm) and oil detection is a function of the total hydrocarbon
concentration. Interference of metal ions is avoided by employing steam distillation
and condensation techniques to vaporize oil from sample solutions. The prototype
instrument successfully detected oil concentrations down to 10 ppm for oils with vapor
pressure equal to or higher than No. 4 fuel oils.
This report was submitted in fulfillment of Contract No. 68-03-0205 by Nucor
Corporation under the sponsorship of the U.S. Environmental Protection Agency. This
report covers the period December 4, 1972 to December 4, 1973.
IV
-------
CONTENTS
Foreword iii
Abstract iv
Figures vi
Tables vi
Abbreviations and Symbols vii
Acknowledgment viii
1. Introduction 1
2. Conclusions 3
3. Recommendations 4
k. Technical Dissertation 5
Concept 5
Advantages of Flame Emission Method 5
Interferences 6
5. System Development 8
Improvement of System Sensitivity 10
Oil Analysis 11
Steam Distillation 13
6. System Testing 20
Batch System Procedure 20
Continuous System Procedure 23
7. Discussion 3*f
References 35
Appendices
A. Design Goals 36
B. Description of the Instrumentation used in Flame Emission
Spectroscopy 39
-------
FIGURES
Number
1 Conventional steam distillation cell 15
2 Horizontal steam distillation cell 16
3 Jacketed steam distillation cell 17
4 Schematic block diagram of the flame emission detector system 21
5 Steam distillation apparatus linkage to sample and condenser 22
6 Benzene and salt signal variation; signal to noise ratio vs.
Nitrogen/Hydrogen ratio 2k
7 Flame emission signal of tap water (monitored at 431 nm) 25
8 Flame emission signal of 10 ppm No. 2 fuel oil (monitored
at 431 nm) 25
9 Flame emission signal of 40 ppm No. 2 fuel oil (monitored
at 431 nm) 26
10 Sampling flow scheme 27
11 Steam distillation cell 30
12 Output signal vs. time for 100 ppm No. 2 fuel oil 31
13 Effect of steam flow rate on the ability of the system to remove
No. 2 fuel oil from the sample: steam temperature was
370°C, sample rate was 100 ml/min 32
14 Output signal of 3% salt solution vs. steam pressure 33
TABLES
1 Integrated peak areas of No. 2 and No. 4 fuel oil at various
concentrations 28
vi
-------
ABBREVIATIONS AND SYMBOLS
ppm —parts per million
nm —nano meter
A° —angstrom
ml —milliliter
cm —centimeter
g -gram
cm-3 —cubic centimeter
mm —millimeter
UV —ultraviolet
AC —alternating current
vii
-------
ACKNOWLEDGMENTS
This report has been compiled by Dr. Dennis Stainken from the original Nucor
Corporation technical and progress reports. Explicit recommendations of Uwe Frank,
Project Officer, and Leo T. McCarthy, have been added.
viii
-------
SECTION 1
INTRODUCTION
Oil pollution is a major problem affecting inland water, coastal waters and the
oceans. Hydrocarbons in the aquatic environment generally result from natural,
accidental and deliberate pollution. Pollution often results from offshore petroleum
production, pipeline leaks and ruptures and tanker transportation.
Disasters involving tankers and drilling rigs are the most spectacular sources of
oil spills. Other sources of oil pollution continuously contaminate the aquatic
environment. They include leaking oil wells, releases of emulsified oil and water
soluble leachates by industries, runoff from roads and streets, exhausts from outboard
engines, waste oil from garages, oil from the stacks of ships, bilge oil and seepage.
Hydrocarbon input from the air is also believed to be a contributory factor. Additional
oil may be derived from sewage sludge and dredge spill which may contain vegetable,
animal, and petroleum oils. The presence of biogenic hydrocarbons can complicate
analyses at minimal detection levels.
The estimated total annual influx of oil to the ocean is between 5 and 10 million
tons. It is of particular importance to analyze the oil concentration in order to avoid
adverse effects to humans and their environment. Areas of analytical interest would
include discharge outfalls, spill areas, and post spill treatment areas.
Methods used to measure oil concentration in polluted water include
fluorescene analysis, gas liquid chromatography, infrared spectrophotometry, ultra-
violet adsorption methods, and total carbon analysis. Many of these methods are time
consuming and costly. They are used primarily within a laboratory and require sample
extraction and preparation. They are often specific for fractions of oils and may be
overly selective. The value of the results may be limited due to the complex
composition of oils and the sophistication of the methods.
The flame emission method does not require prefractionation of oils before
analysis. The method is based on the principle that electrons of elements held in a
flame are raised from a ground state to an excited state by the heat energy. If the
excited electron drops back to the ground state in one jump, the radiation given off is
called a resonance line. If the electron loses its energy in steps, some of the energy
may be emitted in the ultra-violet, visible and infrared regions. The wavelength of
emitted radiation will differ for each element because of the basic differences of
charge and number of electrons. The color and intensity of the emitted radiation may
be used for qualitative and quantitative analyses.
The flame emission method has already been used for measuring toxic metals,
phosphates, etc. The method has not been used for measuring carbon concentration in
water, because of the very low sensitivity of the C-H or C-C emission signal in
1
-------
comparison to that for metal ions (i.e. Na, Li, Ca, etc.). However, the flame emission
method would be useful, for measuring oil pollution, because the hydrocarbon emission
signal does not have structure selectivity. In the flame emission method, detection is
essentially a function of the total hydrocarbon concentration (e.g., carbon atom
concentration).
The objective of this study was to develop the methodology and technology for
the detection of oil in water by flame emission spectroscopy. The study involved the
development and demonstration of this concept for use as an oil contamination meter.
Appendix A describes the design goals and Appendix B contains a complete list of
features which were sought in the oil concentration meter.
Several questions had to be resolved before the flame emission technique could
be used to detect oil in water. The interference of metal ions (i.e., Na, Fe, etc.) had
to be resolved and the sensitivity of hydrocarbons determined. Interference of metal
ions was avoided by adapting a steam distillation technique (Goulden et. al, 1973) to
vaporize oil from the sample solution. Although the steam distillation separates the
water-oil mixture, certain minute amounts of lithium salts were carried over.
Condensation techniques were required to remove the large amounts of water vapor
generated during steam distillation.
Several types of sample introduction systems were considered, and a dry aerosol
method (Veillon and Margoshes, 1968) was chosen. The wavelength selected to monitor
oil concentration was at 431 nm (C-H). Another strong emission band at 512 nm
(Gaydon, 1957; Robinson and Smith, 1966) is present. However, the use of this band to
monitor the hydrocarbon was abandoned because of proximity of sodium D line
interference at 589 nm. The initial evaluation of the technique was performed using a
batch system. Modifications of this approach permitted continuous analysis of oil
water samples.
-------
SECTION 2
CONCLUSIONS
The flame emission technique is useful for the detection and quantitation of oil
in water. The flame emission instrument developed in this report was also proven
potentially useful as a remote detector for petroleum oils. In tests with fuel oils, the
sensitivity factors for No. 2 and No. 4 fuel oil appeared to be within 10% of one
another. The instrument successfully detected oil concentrations of 10 ppm for oils
with vapor pressure equal to or higher than No. 4 fuel oils. However, lower vapor
pressure material (No. 6 fuel oil) was detected at decreased sensitivities. Interference
from dissolved salts was not detected until the salt concentration approached the
concentration of sodium chloride normally found in ocean water. Even at these
concentrations, significant interferences were not encountered except at the highest
steam rate used for sparging oil with high vapor pressure from the influx water. An
instrument of high resolution operated at high temperatures (538°C) appears to be
necessary for optimum performance.
Additional development is required for the detection of a broader range of oils.
The interference from dissolved salts must be considered in the analysis of low vapor
pressure materials, (i.e, crude oils, heavy fuel oils) especially from marine waters.
The system will provide a useful means for the remote measurement of oils in both
fresh and marine waters if the sensitivity of the system is improved.
-------
SECTION 3
RECOMMENDATIONS
The flame emission technique and instrument developed within this report are
recommended for further development for use as petroleum oil detectors in oil
contaminated water. The relatively simple flame emission method does not require
prefractionation of the oils before analysis and will yield satisfactory quantitative
results at ppm levels. Further development of the instrumental design will be
necessary if it is to be utilized for analysis of heavier than No. 4 fuel oils. It is hoped
that an improved instrument incorporating developmental modifications will commer-
cially evolve from the prototype design.
-------
SECTION 4
TECHNICAL DISSERTATION
CONCEPT
Flame emission spectroscopy measures the emission produced on combustion in
a small hydrogen flame to determine the concentrations of oil in water. Although
several emission bands are excited in this process the 431 nm C-H band is the most
useful for the analysis of oil in water.
Molecular flame emission spectroscopy can provide sensitive, selective, rapid
and accurate measurement of the element to be detected. It has been used to monitor
compounds of several non-metallic elements, especially sulfur and phosphorus. Flame
emission has been adapted to automated instrumentation for pollution monitoring of
chemical warfare agents. Adequate sensitivity has been found with a larger response
for hydrocarbons, than for compounds containing other elements in addition to carbon
and hydrogen (Robinson and Smith, 1966).
It appeared possible to develop a useful field instrument based on this principle.
Flame photometry also seemed to have important advantages over competitive
methods. Several commercially available instruments for continuously monitoring
water for oil contamination are commercially available.
The most commonly used technical principles applied to oil analysis are
fluorescence and absorption spectroscopy. The attenuation of light is measured at
wavelengths ranging from the ultra-violet to the infrared. Some workers have also
measured oil in water by reflection of incident light by the oil or by determining the
ratio of reflected to transmitted light. Other methods include measuring light
scattered by oil particles dispersed in water and reduction of electrical conductivity of
water by oil.
Some of these methods are entirely non-specific, (i.e., conductivity and light
scattering). For other methods, oil must be dissolved (i.e., UV absorption spectro-
scopy) or the oil must form a film on quiescent water (i.e. reflectance methods).
Spectroscopic methods provide a high degree of specificity. This is one reason for the
extensive use of UV spectroscopy for monitoring oil in water. However, substances
other than oil that are usually dissolved or dispersed in the water will also respond at
the analytical wavelength. Flarne photometry may solve the problem, because the
extent of organic dispersions in water does not affect the measurement in this process.
ADVANTAGES OF FLAME EMISSION METHODS
The literature previously cited indicates a higher flame photometric response
for hydrocarbons than for other organic compounds. Introduction of oxygen or chloride
into the hydrocarbon structure can reduce emission intensity by a factor of 10 to 25.
5
-------
One of the requirements for flame emission detection is that the moJecuJes be
sufficiently excited to emit radiation; this light is then selectively measured, through
regulation of the available excitation energy. This is accomplished by control of flame
temperature, flow rates of fuel, oxidation rate, and fuel to oxidant ratio. This is one
reason for better specificity in emission than in adsorption spectroscopy.
In flame emission spectroscopy, the maximum intensity of the light emitted by
various species is not necessarily observed in the same region of the flame. For some
emitters (i.e., sulfur compounds in a hydrogen flame) the emission can be largely
confined to a region above the flame cones. For the other emitters, only certain
portions of the flame itself are colored. By restricting the field of view, it is often
possible to eliminate from measurement the emission from interfering substances.
This technique for improving specificity is called limited area flame spectroscopy and
is not available in conventional adsorption spectroscopy.
Techniques used in UV spectroscopy to reducing interferences can also be used
in flame emission spectroscopy. These methods include the comparison of measure-
ments at analytical and reference wavelengths or measurements made on total
samples and on samples with oil removed.
An important advantage of flame emission is that the degree of oil dispersion
does not matter because all oil will be detected. In flame spectroscopy, molecular
excitation of vaporized oil is measured. The instrument in this study uses an
ultrasonic nebulizer to convert the sample to an aerosol. Fine droplets (with a mean
diameter of approximately 5 microns) are formed by the oil and water solution in this
process. Droplets are then fed into the burner wthere they are easily vaporized. The
entire procedure requires only seconds.
The flame emission methodology and instrumentation, can detect oil concentra-
tion as low as 5 ppm. To achieve this, several innovations were required: (1)
ultrasonic nebulization of the water sample, (2) addition of nitrogen to the hydrogen
fuel gas to cool the flame, (3) viewing only a portion of the flame to achieve a high
signal above background, (4) sample conditioning whereby much of the sample water is
removed from the oil prior to reaching the burner. A detailed description of the
instrumentation is described in Appendix B.
INTERFERENCES
Particulates such as paint, wood chips, and scale, do not interfere with
detection process. However, they can clog the flow system in the instrument and
should be removed by a mechanical filtration prior to nebulization. Air and bacteria
do not interfere.
When emitted lines of inorganic species are close to the analytical wavelength,
the following methods are available for avoiding interference:
(1) The interfering substance may emit in a region of the flame different
from that in which the oil emits. (Oil emits primarily near the base of
the burner). In this case it is possible to mask that portion of the flame
in which the interference emits and still transmit the radiation emitted
by the oil.
-------
(2) When the radiation emitted by interference cannot be prevented from
reaching the photomultiplier tube, it is still possible to prevent a
response. The emission lines of inorganic ions are very narrow compared
to the broad ban C-H emission from oil. To eliminate the ionic line
emission from being measured, a chopper with two interference filters is
employed. One filter has maximum transmission at the analytical
wavelength at which both oil and interference emit. The other filter is
selected to have peak transmission at a wavelength at which the
interference but not oil emits. These filters are mounted on a wheel
between the burner and the photomultiplier tube. The wheel is rapidly
rotated by a constant speed motor so that the detector tube will
alternately view radiation with and without that from the oil. It is then
possible to electronically cancel the interfering radiation.
(3) It is possible to remove ionic species from the water by brief contact of
the water with an ion exchange resin prior to nebulization. Contacting
water with resin for 30 seconds results in 99% removal of ionic species.
NUCOR has used both filters and ion exchange resin with flame photometric
instruments. Interference has not been observed from any anions in water. Signals
from cations at high concentrations have been obtained only from calcium and sodium
when phosphorus is measured at 525 nm. Both interference could be removed with ion
exchange rsins.
Organic substances other than oils are also potential interferences. These
would primarily be non-hydrocarbons, and their emission intensity would be expected
to be much less than that of oils.
If necessary, the previously described techniques for treating such interferences
may be used with a very low temperature flame or a dual channel filter system.
In this study, oil monitors were designed to incorporate custom made interfer-
ence filters with sharp cut-offs in order to restrict transmitted radiation to wave-
lengths close to the anaytical band. This also prevented transmission of radiation from
species other than carbon compounds.
Further improved discrimination was expected with the use of custom made
photomultiplier tubes. The sensitivity of photomultiplier tubes to radiation passed by
the interference is a function of wavelength. Peak transmission of the RCA-1P21 tube
used in the NUCOR photometric oil monitor is approximately ^00 nm. At the time of
this study, NUCOR was currently negotiating with a photomultiplier tube manufac-
turer, EMR Inc., for fabrication of a custom made tube that would peak near the
analytical wavelength, (approximately 430 nm) and cut off more sharply at longer
wavelengths.
-------
SECTION 5
SYSTEM DEVELOPMENT
The fabrication of the laboratory study instrument used in the fundamental
studies of the flame emission analytical method employed interference filters with
peak transmittance at ^300 A and 51^5A for measurement of C-H and C bands
respectively. An RCA 1P28/V1 photomultiplier tube (to measure the analytical peak
emission intensity) and a Mistogen Model EN 142 electronic nebulizer, (to present oil in
water samples to the burner) were also used. A burner was assembled from
components on hand. It was installed with other available and necessary components
(aspirator, ignitor, etc.) in a cabinet for use in initial evaluations. A NUCOR
fabricated variable voltage power supply and a Keithley Model 600B electrometer
were used to supply voltage to the photomultiplier tube and to measure photomulti-
plier output current. A Cast Model 1531 oil-less compressor was chosen to provide the
driving air for the aspirator.
An electrically heated glow ignitor was fabricated from a coil of platinum wire.
However, this ignitor did not function reliably. In earlier work, a spark ignitor was
used. A high voltage was applied between the capillary tube which supplies the
hydrogen fuel and the stainless steel nozzle that the hydrogen fuel and the stainless
steel nozzle that separates the combustion chamber from the emission zone of the
burner. A spark jumping across a gap of a few millimeters between these two pieces
ignited the flame. This procedure was suitable for a phosphorus or sulfur detector in
which the capillary was placed behind the emission zone for purposes of shielding the
photomultiplier tube from the flame. For oil monitoring however, it was necessary to
view the flame directly which would require that hydrogen capillary be advanced into
the emission tub. Therefore, the distance between the nozzle and the capillary would
be too great for the ignition to occur. A spark was therefore used for ignition while
the capillary was in a retreated position. After ignition, the capillary tube was
advanced into the emission zone. Although inconvenient, this method of igniting
sufficed until optimum burner geometry was established.
Initial measurements were made using benzene vapors in air as a source of
carbon containing radicals. These samples were used to measure C-H emission of ^315
A° (or C emission at 5165 A°) and to investigate the effect on emission intensity of
such operating parameters as fuel and combustion air flow rates, analytical wave-
length, and photomultiplier tube supply voltage. At the same time, it was possible to
avoid complications (flame noise, cooling of the flame) caused by water droplets.
The following method was used for introducing benzene vapors into the air
stream supplied to the burner. A piece of filter paper was rolled up and inserted into a
short length of 6 mm Pyrex tubing. A few drops of liquid benzene were put on the
paper. One end of the tube was attached to a cylinder of compressed air, and the
other end was attached to a hypodermic needle. During measurements, the
8
-------
hypodermic needle was placed Into the sample inlet. Air was passed at a low flow rate
through the tube and entrained benzene vapors. At the sample inlet, this benzene
containing air was diluted by the much larger air flow introduced into the burner to
support the combustion of the hydrogen. It was assumed that the air from the
hypodermic needle was saturated with benzene vapors. From knowledge of the two air
flow rates and the ambient temperature, it was possible to calculate the benzene
vapor concentration.
Some measurements were made from observation of the flame characteristics.
A blue-green emission was observed in and beyond the flame when benzene vapors
were added to the air stream, and the burner was set a hydrogen flowrates of about
150 to 1600 ml/min with air flow rate of 600 ml/min.
Measurements were made with approximately 10 ppm benzene in air, which
could be easily detected. At a constant air flow, the best response was obtained at a
hydrogen, flow rate of 350-^00 ml/min. Measurements designed to determine the
effect of operating parameters on performance were continued with approximately 15
ppm benzene vapor in air.
Using hydrogen at 350-^50 cc/min. the air flow rate for combustion was varied
between 700 and 1700 ml/min. At high air flow rates, it was necessary to increase the
hydrogen flow above the normally used value of 350 ml/min in order to maintain a
stable flame. The background current which depends greatly on hydrogen flow was
also insignificantly affected by changes in air flow rate.
The burner was usually cooled by passing a portion of the air from the
compressor over the exterior burner surface in the vicinity of the flame. When the
burner was operated in this manner, a thermocouple on the outside of the burner
indicated a temperature of approximately 70°C. When the burner was not cooled, the
temperature rose to 120°C. Cooling was beneficial, dampening the signal noise.
Consequently, the signal to noise ratio and detectability were improved.
Measurements were made with an interference filter exhibiting peak transmis-
sion at 4300°A. The response was significantly greater when the lower wavelength
filter was used. This filter was designed to measure C-H emission which is most
intense at 5145°A.
Photomultiplier currents were measured with a Keithly electrometer. In the
instruments under development, an electrometer circuit with range switch, which is
used in a NUCOR radiation monitor was to be included. The necessary components
capable of measuring currents between 10~^ and 10~9 amps were assembled. The
range switch permits selection of full scale meter ranges of 10~-5, 10~^, 10~7, 10~^, and
10~" amps. The circuitry was tested with currents applied from a battery. At high
currents of ID"-* to 10~7 amps, the circuit behaved correctly. At lower currents the
same signal applied to different ranges could not be measured correctly. It was
eventually determined that AC pickup interfered with the measurements at low
currents. This was eliminated by appropriate modification of the circuits, and
satisfactory performance was subsequently observed.
Preliminary emission measurements were made with hydrocarbons dissolved in
water. In the first experiment, a saturated solution of pentane in water was used. A
-------
large signal was obtained on introducing an aerosol sample into the burner. This signal
lasted only a few seconds, although aerosol generation was continued.
When the nebulizer was disassembled it was observed that the original strong
hydrocarbon odor had completely disappeared from the solution left in the aerosol
generator. Because of the violent agitation accomplying aerosol generation, the
hydrocarbon was apparently removed from the nebulizer at a faster rate than water.
This resulted in an increased pentane concentration during the aerosol phase.
Subsequent experiments were performed with solutions of 50 ppm heptane or
100 ppm benzene in tap water. Small signals that did not decrease to the background
value were obtained with these solutions. However, nebulization reduced the intensity
of the odor of these solutions.
The effect on hydrocarbon emission intensity of singularly varying several
operation parameters (one at a time) was studied. Responses were best when a
relatively large and hot flame was used. This was achieved by increasing the size of
the flame jet, heating the exhaust aspirator, and not cooling the flame. In all cases,
the hydrocarbon emission represented a small increase above that of the background
observed with water without added hydrocarbon. Improved performance with a hotter
flame may be caused by more efficient volatilization of the sample. A comparison of
data obtained with air and water samples containing benzene showed the system to be
almost twice as sensitive to a given amount of benzene in air compared to benzene in
water.
Limited area measurements were also made with an interference filter with
peak transmission at 51^5 A. However, the response was much smaller than was
observed with the lower wavelength (i.e. 4300A) filter.
The response of various concentrations of benzene in tap water were also
examined by the limited area technique. Although the signal obtained was much
improved, it was not steady because of a gradually decreasing hydrocarbon concentra-
tion in tap water during nebulization.
IMPROVEMENT OF SYSTEM SENSITIVITY
The C-H sensitivity was highly improved by measuring the limited area of the
flame, but further enhancement of the signal to noise ratio was needed for the
detection of hydrocarbon concentrations of less than 10 ppm in water. Therefore, the
following modifications of the burner system were made:
(1) With the H2 flame pointing downward, the sample was introduced upward
into the flame.
(2) The sample was introduced from the top of the flame.
(3) A non-shielded oxidizing flame without glass chimney was used.
(4) The NT2~H2 flame was diffused.
10
-------
The results obtained from arrangements (1), (2), and (3) were not as good as
those with the flat burner. A N2-H2 diffusion flame, however, produced a 'f-fold
increase in the emission signal even when a shielded burner was used.
To prevent the loss of sensitivity when benzene in water aerosol was directly
injected into the burner, the benzene aerosol sample was passed through a glass tubing
preheated at approximately 250°C and then cooled in an air condensor. This had a
trap at the bottom for the removal of condensed water. The detection limit for
benzene in distilled water was improved to 30 ppm by the combined modifications of
nitrogen diluted hydrogen flame and aerosol sample heating-condensing techniques. It
was clear that the method used to introduce the aerosol sample into the burner greatly
affected C-H sensitivity.
In order to optimize the C-H signal, a combination of (1) N^-H2 diffusion flame,
(2) a flat burner, and (3) plane mirrors around the burner to reflect additional light to
the photomultiplier tube were investigated. In addition, the hydrogen flame was
shielded by a 0.6 cm slit as previously described, permitting a limited area of the
flame to be viewed by the photomultiplier tube.
Generally, the injection of a 15 ppm benzene in air sample into the burner,
caused a doubling of the output signal under the limited area measurement. The
output signal doubled again when a flat burner was used, and tripled if three flat
mirrors were mounted around the burner. The most important factor in the flame
detector was the use of a N2-H2 diffusion flame which decreased the background of
the flame and improved the signal ten times. These modifications improved the
sensitivity 120 times compared to earlier measurements of benzene in air.
OIL ANALYSIS
An oil dispersion was prepared by mixing 0.2 ml of No. 2 fuel oil in one liter of
distilled water, followed by agitation with an ultrasonic probe for 10 minutes. The
colloidal mixture was allowed to stand for 15 minutes in a separatory funnel before the
lower part of the dispersion was collected in a stock bottle. A small sample of the oil
dispersion was then extracted with Freon 113 and the oil concentration determined
with an IR double beam spectrometer (Gruenfeld and Frederick, 1977).
In modification of the technique of Veillon and Margoshes (1968), aerosol
samples generated in the Mistogen Electronic Nebulizer were passed through a glass
tube wrapped with heating tape and heated to 250°C. The samples were then passed
through a Friedrichs condenser and into the burner. Using this arrangement with the
instrument, modified with a N2-H2 diffusion flame, flat burner, plane mirrors and a
0.6 cm slit, it was possible to detect 4 ppm benzene or 10 ppm of No. 2 fuel oil in
distilled water. Measurements with a 5 ppm oil solution also showed a small positive
signal. We began work on eliminating responses from interfering substances present in
non-distilled water.
An oil in tap water solution was introduced into the burner under the same
conditions as previous tests with distilled water solutions. Even a No. 2 fuel oil in tap
water sample was not detectable, presumably due to emission interference from
metals in tap water. When an oil in tap water solution was introduced into the burner,
11
-------
the signal of the C-H band at 431 nm was masked by these metals. Experiments to
remove interfering rnetals without affecting the oils were attempted by passing the oil
in tap water solution through an ion exchange column. This failed because the organic
ion exchange resin had a strong affinity for the oils.
A modification of this metal ion removing procedure was performed by
connecting a dry cation exchange resin column between the sample injection port and
the condensor of the sample introduction system. It was found that a dry benzene
vapor sample was able to pass through the column without losing its sensitivity.
However, some signal from the No. 2 fuel oil in tap water sample was still lost to
adsorption of the oil on the resin. Only 70 ppm of No. 2 fuel oil in tap water was
detectable by this method. We also tried using Bio-Rad Zp-1 (100-200 mesh) inorganic
cation exchange crystals instead of the organic resin in the column between the
condenser and burner. In order to increase the mobility of oil in the crystal column,
the crystal column was preheated to 250°C. The detection limit was lowered to 50
ppm using this method.
A small portion (1-2 g) of the cation exchange resin was also mixed with the
sample solution and stirred on a magnetic stirrer, prior to nebulization. It was found
that after stirring for one minute, 99% of the metals were removed in water, but also
about 50% of the oil was removed from a 50 ppm oil in distilled water emulsion. The
detection limit of oil in tap water was lowered by this technique to 20-30 ppm.
It was then thought that by reducing flame temperature through addition of
nitrogen to the fuel gas, the emission of metallic ions (i.e. sodium and calcium) could
be reduced and their interference minimized. Measurements of C-H, tap water and
sodium response were made in flames in which the N2 to H2 ratio was varied from 2 to
5. Within this range, the C-H emission was twice as great at the highest N2 to H2
ratio than at the lowest. In contrast, sodium and tap water responses were about 25%
greater at the lowest N2 to H2 ratio than at the highest one. Therefore, operating the
flame at a high N2 to H2 ratio (low flame temperature) increased the C-H response
and lowered the metal ion response.
These results indicated that C-H response and discrimination against metal ions
could be improved by operating under cool conditions. Unfortunately, even at the
highest N2 to H2 ratio at which stable flame conditions could be achieved (a N2 to H2
ratio of about 5), the sodium response was still very strong.
Both narrow band interference filters and broad band pass filters were tried
individually and in combinatin to occlude sodium emission. A significant improvement
was not observed over the commonly used Baird-Atomic filter. The use of a Baird-
Atomic filter in conjunction with a didymium filter designed to block the sodium line
at 589 nm also did not result in improvement. The data indicated that sodium was a
much more serious interference than calcium.
The following steps were taken to eliminate sodium interference. A custom
made filter with sharp cut-off was ordered from Baird-Atomic. Although the standard
filters used previously transmitted 0.1% outside the band pass, the custom made filter
transmitted 0.001% outside the band pass (at the expense of some loss of transmission
at the analytical wavelength). In addition, negotiations were started with EMR, Inc.
for the fabrication of a custom made photomultiplier tube with a cut-off at a much
12
-------
lower wavelength than the 650 mm cut-off of the RCA 1P21 tube used previously. It
was thought that the intensity at 590 nm could be decreased 100 to 1000 times. If
these gains could be realized, the sodium emission would be blocked from detection
and measurement of oil in sea water would be feasible.
Measurements with the improved instrument using 6 concentrations between
No. 2 fuel in distilled water showed a linear response over the entire range. It was
also found that the same response was obtained for equal concentrations of No. 2 and
No. k fuel oils in distilled water.
The methods used to decrease the metal (sodium) emission interference from
tap and salt water were as follows:
A. to use a non-commercial special wavelength selective photomultiplier
tube with a sharp decrease in response beyond 500 nm.
B. to use a high transmission but very narrow band optical filter with
specially sharp transmisson cut-offs.
C. limited area measurement of the flame
D. to separate the oil from salt water by either physical or chemical
methods.
The responses of RCA-1P21, RCA-1P28, and several specially made EMR
photomultiplier tubes were then compared for a better oil signal in tap water. A
variety of Corning and Kodak optical filters with sharp long wavelength cut-offs to
prevent detection of the 589 nm sodium line were also tested. Combinations of these
optical filters and photomultiplier tubes decreased the sodium interference in tap
water but did not eliminate it. The sodium emission spectrum still exhibited a small
band in the reducing flame at ^0 nm which appeared in the inner cone of the flame.
This was apparently the principal interference, rather than the 589 nm line.
STEAM DISTILLATION
The strongest C-H flame emission signal was found at a height of 6-7 nm above
the burner tip in the outer cone. Construction difficulties prevented the complete
separation of the inner and outer cone of the diffusion flame. A steam distillation
method which resulted in the successful determination of 5-10 ppm of oil either in
distilled or tap water was therefore devised. Steam was passed through the oil
containing water sample. The sample was then passed through a preheater-condenser
to eliminate much of the water prior to reaching the burner, as previously described.
With this arrangement, the same background response was obtained from distilled
water, tap water, and a 1% sodium chloride solution; and 5 ppm of oil in tap water
gave a positive response above the background.
Plotting the area under signal peaks against concentrations generally yielded a
linear relationship for the No. 2 and No. ^ fuel oils. Some difficulty was encountered
in working with a No. 6 oil because it has a high boiling point and therefore more time
was required to remove No. 6 oil from solution. Apparently, the steam temperature
greatly influences the rate of distillation and to accomplish this efficiently a higher
13
-------
temperature steam was required. A high temperature steam would presumably
vaporize oils much faster and a narrow signal band would be obtained. The peak height
rather than its area would then be used for the oil concentration determination.
In order to detect the oil concentration in a continuous flow system, a
conventional distillation cell, shown in Figure 1, was made for this purpose.
A preheated sample solution was introduced into the distillation cell (in Figure
1) through the bottom (A) and heated steam was fed through a tube (C) to the bottom
of the cell. The port at (B) served to maintain constant sample volume in the cell.
The sample flow rate was adjusted to about 0.3 cc/sec. The vapor outlet (E) was
connected to a water condenser, which in turn was connected to the H2-N2 burner.
This method was sensitive only at the 20-30 ppm level for No. k fuel oil. The solution
from the drain (D) was collected and some oil was still left in the waste suggesting
that more efficient oil volatilization was required for the continuous measurements.
It seemed reasonable that a long horizontal cell might serve this purpose. An example
of a horizontal steam distillation cell is shown in Figure 2.
The cell had to be long enough to permit vaporization of all the oils during
sample flow from A to B.
In reporting the linear relationship between the area of the recorder response
and the oil concentration for a batch system, the average response for the No. 2 fuel
oil was about 25% higher than for the No. 4 fuel oil. The data indicated the
importance of increasing the temperature in the distillation cell in order to more
efficiently volatilize the oils.
A horizontal distillation cell 23.75 cm long and 3.12 cm in diameter was
prepared, and a preheated solution was introduced into the sample cell along with hot
steam at the flow rate of 0.^ cc/sec. Using this conventional method, approximately
10 ppm of No. 4 oil in tap or salt water solution was detected.
The passage between this distillation cell and the condenser was too small to
pass the oil vapor and the steam which resulted in a somewhat noisy signal. Extra
large holes were then made on the top of the distillation cell so that a steady and
homogeneous steam-air mixture could be obtained. The whole distillation cell was
covered with a 53 mm diameter cylinder jacket (Figure 3).
With this system, a 1 ppm benzene in air sample was easily detected, but a 5
ppm oil in water sample flowing continuously could not be detected. Detectability of
No. k fuel oil was still at 10 ppm. Examination of the solution flow out of the drain
showed that the No. 4 fuel oil was not completely removed from the water by steam
distillation. The temperature of the inlet steam was measured to be 32^.7°C which is
far below the boiling point of the No. <4 and No. 6 fuel oils.
In general, detection of No. 2 and No. 4 fuel oil was possible down to 10 ppm
when the steam temperature was 318°C. Number 6 fuel oil, having a lower vapor
pressure than either No. 2 or No. 4 fuel oils, required an increased steam temperature
in order to achieve the 10 ppm detection level. The stearn distillation rates are
dependent on the vapor pressure of the components to be removed from solution.
-------
Steam -H
(Preheated)
Condenser and Burner
Sample
(Preheated)
D
Drain
Figure 1. Conventional steam distillation cell.
-------
Condenser and Burner
Sample
A
Steam
o\
Drain
Figure
2. Horizontal steam distillation cell.
-------
Burner
* Ai r
Steam-
Cylinder Jacket
f-t-t —t-t—r-t
1 t t t i t t t—i t t r t i r
i i
4> i i — i i 4 i 4 4 i
I
Drain
Jl
To
Condenser
f
Sample
Figure 3. Jacketed steam distillation cell.
-------
Although most of the hydrocarbons could be steam distilled at temperatures lower
than their boiling points, the steam distillation rate was highly dependent on the
component vapor pressure at the temperature of the applied stearn. Therefore, a
steam generator and superheater were utilized to increase the steam temperature to a
range of 486°C to 542°C.
It was observed during operation with the high temperature steam that an
unexpected noise signal resulted when high temperature steam (greater than 318°C)
was passed through a 0.5%-3% salt solution. Temperature scanning of the 3% salt
solution showed that a small amount of salt was driven out of solution when the steam
temperatue exceeded 318°C. The noise level from the salt depended on the salt
concentration. Therefore, although a 542°C steam temperature removed oil in
solution more rapidly than 318°C steam, oil concentrations of less than 30 ppm were
not readily detected due to the increased background (salt) signal.
It was therefore necessary to remove the salt from the aerosol generated by
steam distillation at higher temperature before introducing the aerosol into the
burner. It was found that either a gas trap half filled with glass beads or a screen
mounted slightly above the sample solution significantly reduced the noise from the
belt. Salt could be removed by the above methods because:
1. Salt is soluble in water but oil is not. When a mixture of oil and salt
containing vapor is passed through a water trap, salt is removed by a
dissolving process. However, small amounts of oil are also removed by
condensation and adsorption.
2. When tap water was dropping onto a hot plate, a very large signal was
obtained from the atomization of salt. A screen prevented the water
from splashing onto the hot steam pipe and thus the formation of salt
aerosol was reduced.
NUCOR subsequently developed the following techniques toward fabrication of
the required instrument:
1. Establishment of a continuous sample flow system employing the steam
distillation technique.
2. Optimization of system parameter.
3. Assembly of prototype instrument.
A continuous sample introduction system was established using a small peristal-
tic pump for introducing the sample into the distillation cell. Additional improve-
ments were made in the control and measurement of the steam flow rate and
temperature, and the sample flow rate. With the continuous system established, a
series of sensitivity and linearity measurements using No. 2 and No. k fuel oils were
conducted.
This series of measurements resulted in linearity within 10% attained for No. 2
and No. 4 fuel oil. Samples ranging from 13 ppm to 250 ppm were used, and favorable
linearity and sensitivity resulted.
18
-------
A more efficient vaporatior. of oil contained in the sample was sought.
Reasonable performance was obtained with high temperature steam, but the higher
heat content of the resulting vapor necessitated the use of a more efficient condenser
for optimum performance. The system was modified to include a high performance
condenser.
Results from the steam distillation cell were encouraging. To evaluate the
system more precisely, attempts were made to improve sample preparation and to
develop a simple calibration technique. A laboratory stirrer was found to be
convenient for mixing No. 2 and No. 4 fuel oils, and a sonifier was used to prepare
homogeneous No. 6 fuel oil samples.
The set point precision of the flow controller regulating the detector flame was
not satisfactory when distillation cell efficiency was evaluated. Small variations in
hydrogen flow rate caused variations in detector sensitivity; a calibration scheme was,
therefore, required to evaluate system sensitivity before distillation cell efficiency
was evaluated. The calibration technique consisted of a metered flow of air through a
wick saturated with benzene. The air carried the benzene to the detector
reproducibly. Calculations involving the air flow rate, the benzene vapor pressure and
temperature enabled an accurate determination of system sensitivity.
19
-------
SECTION 6
SYSTEM TESTING
EXPERIMENTAL SYSTEM BATCH SYSTEM PROCEDURE
The flame emission oil detector system is illustrated in Figure <4. Sulfuric acid
acidified tap water was used as the source of steam in this procedure. The tap water
was acid treated and boiled before each measurement to remove any carbon dioxide or
bicarbonate. The sample flask contained W-50 ml of sample per measurement. A
Baird-Atomic narrow band filter with maximum transmission at 431 nm, band width at
10 nm and 85% transmission at peak wavelengths was used.
The tubings and glassware were insulated with asbestos. The steam distillation
was performed in a 125 ml Erlenmeyer flask fitted with a 5 cm diameter round glass
trap (Figure 5). The steam vapor was connected to a Friedrich condenser shown in
Figure 5, to allow ambient air to flow into the condenser without difficulty.
The stainless steel burner was supplied by a mixture of hydrogen and nitrogen
gas. The gas flow rate was regulated with two Matheson flow meters. The 1 x 7 mm
flat burner was centered in a 1.9 cm diameter quartz tubing chamber. The flame was
viewed 1 mm above and 6 mm below the burner tip with the flame upside down for
maximum sensitivity.
The tests were conducted to evaluate the technique and system utilizing
experimental benzene and oil standard solutions. These standard solutions provided a
means of calibrating and evaluating the system.
The Benzene Standard
A 12 cm long, 5 mm diameter glass tubing was filled with cotton (to a depth of
about 10 cm) and wetted with benzene. The benzene standard tubing was connected to
a flow meter which was attached to an air tank. Air, flowing over the benzene-soaked
cotton became saturated at a concentration corresponding to the vapor pressure of the
compound at room temperature. The flow meter was adjusted so that a constant
amount of benzene was vaporized at any temperature.
The Oil Standard Solution
Standard oil solutions (of No. 2, 4, and 6 fuel oils were prepared fresh and salt
water. Stock oil emulsions were prepared by sonification with a Branson sonifier cell
disruptor (Model W-185) and quantified by infrared analysis (Gruenfeld and Frederick).
20
-------
ro
A. Steam generator (500 ml Erlenmeyer Flask)
B. Trap
C. Sample flask (125 ml)
D. Friedrich condenser
E. Burner (1x7 mm, flat stainless steel)
F. Aspirator
G. Baird atomic narrow band filter
H. RCA IP-21 Photomultiplier tube
I . High voltage DC supply
(J. Fluke Mfg. Co., Inc., Model 412A)
J. Keithley 61 OB Electrometer
Figure 4. Schematic block diagram of the flame emission
detector system.
-------
Ambient
Air
Steam
IV)
ro
Friedrich Condenser
To Detector
Sample
Waste
Figure 5. Steam distillation apparatus linkage to sample and condenser.
-------
RESULTS
Flame sensitivity was measured with a 2 mm diameter cylindrical burner. The
sensitivity near the burner tip was about four times higher in comparison to the
sensitivity near the top of the flame. The sensitivity of the diffusion flame was also
examined. The N2 to H2 ratio did not have much effect on the most sensitive flame
area. However, the benzene detectability was increased with an increasing N2 to \\2
ratio which optimizes the C-H signal. At a \\2 to N2 ratio of 5:1, a maximum response
for C-H emission was obtained. The emission response of No. 2 and k fuel oils was
linear and equal for both oils (from 5-90 ppm in distilled water).
Benzene and salt signal variation with N2 to H2 flow rate and the signal for tap
water and 120 ppm sodium ions atomized with a Mistogen Nebulizer are shown in
Figure 6. The aerosol was preheated before injection into the burner compartment.
This data indicates that salts must be removed before injection into the burner. The
burner must use a diffusion flame to minimize interference. A nitrogen flow rate of
350 cc/min and hydrogen flow rate of 100 cc/min gave the best performance. A
quartz viewing tube was required to prevent interference signals from the sodium
normally found in glass.
The effect of steam distillation of tap water is shown in Figure 7. The curves
were obtained with 40 cc of tap water in a 125 ml Erlenrneyer flask. The sharp peaks
are caused by desorbed carbon dioxide. The sensitivity of the flame emission detector
to carbon dioxide could be explained by the formation of a C-H radical which was
observed repeatedly. The signal of tap water was significant and reproducible.
Figures 8 and 9 illustrate the characteristic signals of No. 2 fuel oil. The
integrated peak areas of various concentrations of No. 2 and No. 4 fuel oils are shown
in Table 1.
SAMPLE PREPARATION - CONTINUOUS SYSTEM PROCEDURE
Oil-water samples were prepared by placing a measured quantity of oil in
approximately 200 mi of tap water to prepare 100 ppm stock dispersions. A syringe
was used to dispense the oil. Oil-salt water samples were prepared using 3.5% table
salt, tap water and oil. The oils employed were Nos. 2, >4, and 6 fuel oils, and a Waring
Blender was used to emulsify oil-water mixtures. Replicate samples of the resultant
dispersions were analyzed to determine their homogeneity and reproducibility.
Analysis revealed a difference of approximately + 4% between replicates. Subsequent
dispersions were not analyzed, although the same procedure was used in making the
samples.
EXPERIMENTAL SYSTEM
A peristaltic pump metered the oil sample into the steam distillation cell.
Figure 10 illustrates the sampling flow scheme. Flow rates from 50 ml/min to 150
ml/min were obtained by varying the rotational speed of the pump. Precise control of
flow was not possible with this setup, and during critical experimental runs the flow
was monitored continuously with a burette. The sample was added to a 50 ml burette
which discharged into the peristaltic pump. Measurement of the time interval to
inject 20 ml gave an accurate indication of the flow.
-------
HO-
NK)-
90-
80 —
70 —
co
tr
1 60
o
u.
03
^
O
50-
03
c
0)
c/5
40 —
30 —
20 —
10 —
SALT
BENZENE
-• TAP WATER
\ I I I
1234
Nitrogen/Hydrogen Ratio
Figure 6, Benzene and salt signal variation: Signal to noise ratio vs
Nitrogen/Hydrogen ratio-
-------
Figure 7. Flame emission signal of tap water
(monitored at 431 nm).
Figure 8. Flame emission signal of 10 ppm No.2 fuel oil
(monitored at 431 nm).
25
-------
Figure 9. Flame emission signal of 40 ppm No. 2 Fuel oil
(monitored at 431 nm).
26
-------
IV)
-J
Distillation
Cell
Steam
Generator
Peristaltic
Pump
a—E>
Q_
Cooling Water
s
Detector
Connector
50-150 ml/min
Figure 10. Sampling flow scheme.
-------
TABLE 1. INTEGRATED PEAK AREAS OF NO. 2 AND NO. 4 FUEL
OILS AT VARIOUS CONCENTRATIONS
Concentration of sample
Blank
10 ppm
15 ppm
20 ppm
30 ppm
40 ppm
50 ppm
60 ppm
70 ppm
110 ppm
1 40 ppm
220 ppm
No. 2 fuel oil
1.03
2.13
3.53
7.65
10.40
13.20
13.00
16.50
20.10
32.00
41.60
64.70
No. 4 fuel oil
1.13
1.63
3.15
6.31
9.70
12.20
13.00
13.80
19.50
30.70
42.00
63.50
28
-------
The steam distillation cell is indicated in Figure 11. Steam was supplied from a
commercially manufactured generator. Approximately 5-30 ml/min (condensate) were
used to heat and strip the oil from the sample. The amount of steam supplied was
altered by varying the power to the generator. Greater extraction efficiencies (90%)
were obtained at flow rates of 10 ml/min and higher.
The oil laden steam was cooled in a Friedrichs condenser and detector as
described in earlier sections of this report. Steam rates higher than 30 ml/min could
not be handled by this condenser.
RESULTS
Typical output data is indicated in Figure 12 at a steam input rate of 30
ml/min. A concentration of 100 ppm of No. 2 fuel oil was used in the sample. Similar
results were obtained with No. 4 fuel oil at the highest steam rates, but the sensitivity
of the instrument for No. 6 fuel oil was considerably lower.
Figure 13 illustrates the effects of steam flow rate on the ability of the system
to remove No. 2 fuel oil from the sample. The background signal of Figure 10, may be
due to carbon dioxide or bicarbonates present in the water. The oil signal is directly
proportional to the steam flow rate up to approximately 30 ml/sec. At this point the
increase in signal is minimal at the higher flow rates. Similar data with No. k fuel oil
was obtained. Number 6 fuel oil required extremely high steam rates, (30 ml/min), to
obtain any signal at all.
Operation of the instrument was also tested employing the 3.5% water samples.
The salt signal obtained at high steam rates is shown in Figure 14. This data was
obtained from a distillation cell without a baffle. The baffle illustrated in Figure 11
suppressed the salt signal to very low values until steam rates in excess of 50 ml/min
were used. The sensitivity of the instrument for No. 6 fuel oil did not increase at
these flow rates, and therefore they were not utilized.
29
-------
Vapor Outlet
Perforations
Sample
Figure 11. Steam distillation cell.
30
-------
ro
c
to
Q.
.tw
3
o
567
Minutes
10 11
Figure 12. Output signal vs Time for 100 ppm No. 2 fuel oil .
Sensitivity was 10 x 10~9 amps full scale with 100 ml/min
of sample flow - steam distillation.
-------
.4 —
UJ
a.
CO
«? .3 —
to
.§.2-
•j
Oil Signal
and
Background
Background
.1 .2 .3
Steam Flow Rate, ml/sec
Figure 13. Effect of steam flow rate on the ability of te system
to remove No. 2 fuel oil from the sample: steam temperature
was 370° sample rate was 100 ml/min.
-------
5H
l-v
° 4-
X
a.
•*.-«
3
o
CO
3-1
2-
1-
T
.7
1.05 1.40 1.75 2.10
Kg/cm^ Steam Pressure
1.45
Figure 14. Output signal of 3% salt solution vs steam pressure.
No baffle was used in the distillation cell, steam temperature was 370°
and sample flow was 110 ml/min.
33
-------
SECTION 7
DISCUSSION
The flame emission technique was evaluated as a means of detecting oil in
water. The experimental method and instrumentation developed during this study was
found to be sensitive to oil concentration as low as 10 ppm for oils with vapor
pressures equal to or higher than No. 4 fuel oils.
The system tested during the Batch system procedure successfully detected
concentrations of No. 2 and No. 4 fuel oils but the No. 6 fuel oil was not satisfactorily
detected. The vapor pressure of No. 6 fuel oil was lower than No. 4 fuel oils and the
steam distillation procedure incompletely removed them from the water sample.
Much of the higher molecular weight compounds and asphaltenes remained in the
water sample after steam distillation. Background interferences were successfully
removed by steam distillation and use of a narrow band filter.
The continuous system procedure successfully detected No. 2 and No. 4 fuel
oils. The instrument was less sensitive for No. 6 fuel oil. Steam flow rates were
determined for the system. These are utilized for No. 2 and No. 4 fuel oil detection.
Greater steam rates were employed to obtain a signal for No. 6 fuel oil. However, the
system was still relatively insensitive to the No. 6 fuel oil. The physical
characteristics and chemical composition of the No. 6 fuel oil made it undetectable.
Interferences were removed from the system by use of a distillation cell with baffle,
when steam rates less than 50 ml/min were used.
The batch system and continuous system procedure reported will detect and
quantitate No. 2 and No. 4 fuel oils. Further development of the system may improve
the sensitivity of the instrument to fuel oils and crude oils. It may be necessary to
include a sample pre-treatment or extraction procedure to obtain maximum use of the
system.
-------
REFERENCES
1. Gaydon, A.G., 1957. The Spectroscopy of Flames, John Wiley & Sons, Inc., New
York.
2. Goulden, P.D., P. Brookshank, and M.B. Day. Determination of Submicrograms
Levels of Phenol in Water. Analytical Chem. 45 (14), 2430-2433, 1973.
3. Gruenfeld, M. and R. Frederick. The Ultrasonic Dispersion Source
Identification and Quantitative Analysis of Petroleum Oils in Water. Rapp. P.-
v. Reun. Cons. Int. Explor. Mer. 171, 33-38, 1977.
4. Robinson, 3.W. and V. Smith. Emission spectra of organic liquids in hydrogen
flames. Anal. Chem. Acta. 36, 489-498, 1966.
5. Veillon, C. and M. Margoshes. An evaluation of the induction-coupled, radio
frequency plasma torch for atomic emission and atomic absorption spectro-
photometry. Spectrochim. Acta. 238,503-512,1968.
35
-------
APPENDIX A
DESIGN GOALS
The U.S. Environmental Protection Agency wanted to develop an instrument
that couJd be used to measure, on an absolute basis, the concentration of oil in oil-
contaminated water. The source of the oil-contaminated water could be from on-
shore oily waste disposal facilities, off-shore oil production platforms, accidental
discharges of oil (tank farms, railroad tank cars, trucks, pipelines, etc.) and from
similar sources and facilities where water is oil-contaminated, and where restoration
and recycling of the water was to be achieved by gravity settling, or by mechanical,
chemical or physical processes.
Several methods have been developed and are proposed for restoring and
recycling the oil-contaminated water. Some of the methods used to separate the bulk
of the oil contamination prior to discharge of the water to the environment are gravity
separation, adsorption, centrifugal separation, flotation, coalescence, filtration,
pressure, vacuum and other related methods.
An instrument which could measure low concentrations of oil in water (i.e., five
to ten parts per million) on a continuous basis and in a reliable and economical manner
was required. Oil concentration meters currently known to EPA involved the use of
ultraviolet fluorescence, particulate measurement (i.e., Tyndall effect) or IR analysis.
An instrument based on these techniques may or may not have the potential to
measure low oil concentration in water. Although several instrument systems using
methods and techniques based on electro-optical or electro-acoustical principals have
been considered, no one system is presently favored over another.
Development of an instrument that could measure low concentrations of oil in
oil-contaminated water was desired, according to the following basic objectives:
1. The instrument should be able to operate with minimum interference or
obstruction by rust particles, pH, salinity (marine, brackish or fresh
water), etc.
2. The instrument should be able to operate with a minimum amount of
auxiliary equipment.
3. The instrument should be able to reject solids and debris which would
interfere with the efficiency of the system or damage the instrument.
4. The instrument should be compact and readily movable from one site to
another.
5. The first cost should be reasonable.
36
-------
6. The operating expense should be minimal.
7. The instrument should need minimum maintenance.
8. Repairs should be able to be made easily.
9. There should be readily available replacement parts.
SPECIFIC DESIGN GOALS
The instrument was to approach the following design goals.
A. Continuously measure and record, on a real time basis, the concentration
of oil in treated water discharged from either marine, brackish, or fresh
water contaminated by oily waters. The limit of detection of total free,
emulsified, and dissolved oil was to range from five parts per million
(ppm) to 500 ppm oil concentration in water.
B. Sensitivity +_ 0.5 ppm of oil (5 ppm to 10 ppm) and 10% of absolute
concentration from 10 ppm to 500 ppm oil, or limit of instrument.
C. Full scale accuracy to + 5%.
D. Digital or analog readout in ppm.
E. Automatic ranging.
F. Self calibrating.
G. Self zeroing.
H. Rugged, all-weather construction.
I. Pre-set alarm and manual alarm test button.
3. Minimum false alarm.
K. Short warm-up and stabilization.
L. Manual and automatic on/off control.
M. Self purging.
N. Self cleaning.
O. Go/no-go indicator or operating condition of instrument.
P. Power: AC 110 Hz, 5 amp maximum
DC 12 or 2>4 VDC, 5 amp maximum
37
-------
Q. Foolproof hook-up — no electrical damage to instrument.
R. Time required for assembly and disassembly shall be one hour or less by a
person not skilled, but able to read instructions.
S. No special tools for assembly and disassembly — prefer wing nuts and
weather proof electrical hook-ups.
T. Must function normally on non-level surfaces or mountings — up to
from level.
-------
APPENDIX B
DESCRIPTION OF THE INSTRUMENTATION USED
IN FLAME EMISSION SPECTROSCOPY
A typical flame photometric monitor employed in this study was shown
schematically in Figure 4. Hydrogen fuel is fed into the burner at a rate of about 100
ml/min through a capillary tube and burns at the end of it. The air required for
combustion and the aerosol water sample are supplied through a stainless steel tube at
right angles to the capillary tube. The UV emission (431 nm) developed on combustion
of the oil is measured by a photomultiplier tube. The 431 nm analytical band is
isolated by a custom made narrow band interference filter with a half band width of
about 10 nm and special sharp cut-off. Radiation of different wavelengths due to
other substances in the water is thereby prevented from detection by the photomulti-
plier tube. The photomultiplier tube signal is amplified, displayed on a meter and may
be recorded on a 10 rnn potentiometric recorder.
The flame is ignited by a spark drawn between the capillary tube and the outer
burner housing. The burner is oriented vertically with the flame pointing downward
for rapid elimination of any condensed moisture. A compressor supplies air to an air
aspirator connected to the burner. The sample inlet is therefore at a slightly reduced
pressure allowing combustion air and aerosolized sample to pass through the burner.
The aspirator is heated electrically with a cartridge heater to prevent condensation of
water vapor.
There are important advantages to the use of an aspirator rather than a pump
for moving air and sample through the burner. The system is always open to the
atmosphere, not subject to sudden pressure fluctuations which can add to instrument
noise, and there is no contamination of a pump.
It has been observed that oil sensitivity is improved by operating the burner at a
relatively cool temperature (a surface temperature of about 100°C). This is achieved
by blowing some of the compressor air over the outside of the burner to cool it and
also by addition of nitrogen to the fuel gas.
The oil containing water is aerosolized with an ultrasonic nebulizer. Smaller,
more easily vaporized droplets are obtained in this manner, and consequently, greater
sensitivity. Significant improvements in response to oil have been made recently by
inserting a heated tube and condenser between the nebulizer and the burner. This
modification serves to condense out much of the water but not the oil, prior to
introduction into the burner. There is another advantage to the ultrasonic nebulizer
important in this application. The violent agitation in the nebulizer results in an
intimate dispersion of the oil in the water and a more complete aerosolization.
Consequently, a more accurate measurement of oil concentration can be made.
39
-------
METHODS OF SAMPLING
The nebulizer can operate on only a few ml/min of sample. This small amount
may be continuously fed into the nebulizer sample chamber after abstracting from the
main stream of the water to be monitored. The few milliliters of sample required are
quickly aerosolized, and the instrument will respond rapidly to changes in oil
concentration. However, care must be taken that the sample is moved quickly to and
through the monitor. This will be accomplished by abstracting the sample from the
main stream at a fairly rapid rate, (approximately 100 ml/min) and passing this
through tubing of small volume to the analyzer. In addition, only a small portion of
this abstracted sample will be fed to the nebulizer with the larger portion being
returned to the main stream. In this manner, water from the main stream reaches the
detector quickly which rapidly senses changes in the composition of the water to be
monitored. However, the sample abstracted must be sufficiently large to be
representative of the main stream.
MAINTENANCE
During many years of development and use of flame photometric monitors,
NUCOR personnel have determined that these instruments require very little
maintenance. In the past, hydrogen was supplied by cylinders of fuel gas. A standard
commercial No. 1A cylinder of hydrogen, at the present rate of combustion of 100
ml/min, can supply fuel for 38 days of operation. For added safety, however, it is
proposed to replace the hydrogen cylinder by a solid polymer electrolyte hydrogen
generator. This is desirable for safety. These require only the periodic addition of
water (not electrolyte) to the generator.
In laboratory use, flame photometric monitors have required very infrequent
maintenance. A glass emission tube in the burner allows the UV emission developed on
combustion of the oil to be viewed by the photomultiplier tube. During many months
of daily use in the laboratory, the emission tube has remained clean.
The ultrasonic nebulizer system has been used for several years by NUCOR for
flame analyses of batch samples in water. No maintenance has ever been required
during this time. No other significant maintenance operations have been encountered
in the operation of flame photometric instruments.
FALSE ALARMS
The principal causes of false alarms would be response from interfering
substances or a drift in base line. Drift is due primarily to changes in flow rates of
burner temperature. Long term base line drift in NUCOR photometers are less than 3
percent and would be unlikely to result in false alarms.
CLEANING OF OPTICAL COMPONENTS
In order to maintain a fuel-rich, cool flame, the flame is shielded from the
surrounding air by a short length of Pyrex tubing. The emitted radiation is transmitted
by this glass tube to the interference filter and photomultiplier tube. Deposits on the
wall of this tube could reduce the amount of radiation transmitted to the detector.
During several years of laboratory investigation and development of flame
photometers for water monitoring, such deposits have never been observed. During
1*0
-------
operation only a few mlJIiliters of sarpple water are aerosolized each minute and much
of this water which may contain various impurities condenses in the preheater-
condenser never reaching the burner. Combustion of the hydrogen fuel in the burner
results in the formation of clean water vapor which can actually clean the glass tube.
If necessary, it would be possible to clean the burner without disrupting operation by
passing clean aerosolized water through it.
RESPONSE TIME
Response by flame emission spectroscopy is very rapid. Oil is detected in Jess
than two seconds after introduction of the aerosolized sample into the burner. The
preheater and condenser were inserted between the nebulizer and burner to condense
much of the sample water prior to introduction into the burner improving sensitivity.
The sample requires about 6 seconds to pass through the preheater and condenser.
Almost 8 seconds are required for detection following nebulization. No attempts were
made to modify this sample conditioning system in order to reduce response time.
The time for the sample to reach the burner depends on the lengths of the
preheater and condenser (approximately 30 cm each) and the air flow rate carrying the
sample aerosol to the burner. The distance between nebulizer and burner are kept as
short as possible, and the sample flow rate as rapid as possible.
TEMPERATURE FLUCTUATIONS
Performance is not affected by temperature changes in the influent. Sample
abstraction and nebulization are not temperature sensitive. To remove much of the
water prior to detection, the ultrasonically-generated aerosol is heated and cooled
condensing much of the water. Sufficient heat is applied so that even cooler than
average water is vaporized in the preheater. Above average influent temperature will
also have no significant effect since the cooler has sufficient capacity to remove any
excessive heat.
The only temperature effect observed is that large changes in burner tempera-
ture (approximately 30°C), result in base line drift. Such great temperature changes
are caused by large intentional variations in hydrogen combustion air flow or cooling
air flow over the burner.
SAFETY
Flame photometric detectors are safe to operate. There are three potential
hazards associated with the type of flame photometer used by NUCOR: (1) hydrogen,
(2) the flame, (3) the ignitor. The flame is very tiny and is reached only by the
successive removal of three separate enclosures. Ignition of the flame is accomplished
by drawing a spark between the hydrogen jet and the outer housing of the burner. This
is done by applying a potential difference between these burner components. The
burner and the ignitor circuit are completely enclosed and are not accessible during
operation of the instrument. The ignitor is actuated by means of a toggle switch and a
cabinet interlock and is provided with status light location on the front panel of the
instrument. Following ignition, this switch is turned off and its use is no longer
required during operation of the instrument. Turning on the ignitor switch will actuate
the ignitor only if the main power switch (provided also with a status light) is also
-------
turned on.
Hydrogen from an electrolytic generator is fed directy into the burner where it
is completely burned. As long as the hydrogen is combusted, it presents no hazard. A
potentially dangerous situation could arise only if the flame went out and hydrogen
continued to pass into the instrument without being burned. If ventilation is poor in
the compartment in which the detector is kept, the hydrogen concentration could
eventually increase to levels that could burn or explode if ignited. Measures should
therefore be taken to assure shut-off of the hydrogen supply in case of a flame-out. A
temperature sensor mounted on the exterior surface of the burner in the vicinity of
the flame will rapidly sense any flame-out. When the temperature drops to a preset
point (20°C) below the operating value (far beyond normal temperature fluctuation
during operation) an electrically operated shut-off valve in the hydrogen flow to the
burner and shut down the hydrogen generator. In this way, it is possible to cut off all
flow of hydrogen within one minute of a flame-out. The instrument will be provided
with an automatic re-ignition circuit to light the flame again following an accidental
flame-out. NUCOR has routinely included such circuits in earlier flame detector
monitors and flame-outs have rarely occurred during operation of NUCOR flame
photometers.
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing}
REPORT NO. 2.
TITLE AND SUBTITLE
etection of Oil in Water by a Flame Emission
echn i que
AUTHOR(S)
anfred Pragar and Dennis Stainken
PERFORMING ORGANIZATION NAME AND ADDRESS
UCOR Corporation U.S. EPA
Richwood Place Edison, NJ 08837
enville, NJ 07834
. SPONSORING AGENCY NAME AND ADDRESS
unicipal Environmental Research Laboratory ,Ci ,(
ffice of Research and Development
.S. Environmental Protection Agency
incinnati , OH 45268
3. RECIPIENT'S ACCESSION-NO.
5. REPORT DATE
6. PERFORMING ORGANIZATION CODE
8. PERFORMING ORGANIZATION REPORT NO.
10. PROGRAM ELEMENT NO.
1BB610
11. CONTRACT/GRANT NO.
68-03-0205
13. TYPE OF REPORT AND PERIOD COVERED
IH Final
14. SPONSORING AGENCY CODE
EPA/600/14
. SUPPLEMENTARY NOTES
Project Officer: Uwe Frank (201) 321-6626
.ABSTRACT
A flame emission technique and basic in
or measuring low concentrations of oil in o
lame emission instrument developed in this
etector for petroleum oils. The flame emis
electivity of the hydrocarbon emission sign
ion is a function of the total hydrocarbon
if metal ions is avoided by employing steam
.echniques to vaporize oil from sample solut
lent successfully detected oil concentration
apor pressure equal to or higher than No. 4
strument design is presented
il-contaminated water. The
report would be useful as a
si on technique utilizes the
al (at 431 nm) and oil detec-
concentration. Interference
distillation and condensation
ions. The prototype instru-
s down to 10 ppm for oils wit
fuel oils.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
hemical Analysis
ils
rude Oil
ater Pollution
lame Photometry
DISTRIBUTION STATEMENT
el ease to Pub! i c
b. IDENTIFIERS/OPEN ENDED TERMS
Oil Detection
Oil -Contaminated Wat<
Petroleum Oils
Flame Emission
19. SECURITY CLASS (This Report}
Unclassified
20. SECURITY CLASS (This page}
Unclassified
c. COSATI Field/Group
;r
21. NO. OF PAGES
51
22. PRICE
Form 2220-1 (9-73)
43
------- |