EPA-650/2-73-011
August 1973
Environmental Protection Technology Series
SrawSS^^;^^
'XOV'-V'-Vt >>'•'»*» v 1*1 v »>•< •, vv.
-------
EPA-650/2-73-011
DIRECT DETERMINATION
OF METALS IN AIR
by
Dr. J. W. Robinson
Department of Chemistry
Louisiana State Univeristy
Baton Rouge, Louisiana 70803
Grant No. 800866
Program Element No. IAAOIO
EPA Project Officer: Carole R. Sawicki
Chemistry and Physics Laboratory
National Environmental Research Center
Research Triangle Park, North Carolina 27711
Prepared for
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, DC 20460
August 1973
-------
This report has been reviewed by the Environmental Protection Agency and
approved for publication. Approval does not signify that the contents
necessarily reflect the views and policies of the Agency, nor does
mention of trade names or commercial products constitute endorsement
or recommendation for use .
11
-------
ABSTRACT
An instrument has been developed capable of the direct determination of
metals in air. No prior scrubbing or extracting of the metals from the
air is necessary. Consequently, the time necessary for analysis is a
matter of minutes - permitting real time analysis to be carried out on
small volumes of air.
The method was based on atomic absorption spectroscopy and involved the
development of a highly efficient atomizer. The sensitivity of the
method was determined to be about 10~12 g. Procedures for the direct
quantitative determination of lead, mercury or cadmium in the air were
developed.
Problems were encountered with traces of impurities in the system.
Normal 'trace* levels are unacceptable at the levels necessary for
direct metal determination. Calibration techniques were especially
difficult to develop and numerous methods were studied. The
calibration methods which were found to be useful at these concen-
trations were developed into reliable analytical techniques. These
methods are described.
Preliminary studies indicated that other elements such as Ag, K, Na,
Se, As, Cu, and Zn were also detectable in air; but calibration
techniques for these methods have not yet been developed.
This report was submitted in fulfillment of Grant Number AP 00866, by
J. W. Robinson, under the sponsorship of the Environmental Protection
Agency. Work was completed as of June 1, 1973-
ill
-------
CONTENTS
Page
Abstract iii
List of Figures v
List of Tables vi
Acknowledgments vii
Sections
I Conclusions 1
II Recommendations 2
III Introduction 3
IV Equipment 8
V Operational Parameters 17
VI The Direct Determination of Lead in the Atmosphere 20
VII Determination of Mercury in the Atmosphere 33
VIII Determination of Cadmium in the Atmosphere ^
IX Determination of Arsenic 52
X Other Elements Studied 59
XI Discussion 62
XII References 63
XIII List of Publications 6k
XIV Glossary 65
IV
-------
LIST OF FIGURES
PAGE
1 SCHEMATIC DIAGRAM OF THE ATOMIZER 7
2 SCHEMATIC DIAGRAM OF EQUIPMENT (Labelled) 9
3 RESPONSE CURVES OF THE PHOTOMULTIPLIER DETECTORS 1P28-R106 11
k DIAGRAM OF CURRENT ATOMIZER 1J
5 PURE AIR SYSTEM (Labelled) 16
6 ABSORPTION SPECTRA - METALS IN AIR 18
7 ALTSHULLER EQUIPMENT FOR CALIBRATION WITH TEL 21
8 ELECTRICAL DIAGRAM OF THE Pt WIRE USED FOR CALIBRATION 25
9 SCHEMATIC DIAGRAM OF THE Pt WIRE CALIBRATION EQUIPMENT 26
10 EQUIPMENT FOR CALIBRATION BY THE LIQUID DROP TECHNIQUE 28
11 TYPICAL TRACES OF ABSORPTION CAUSED BY METAL BLEEDOFF FROM
CARBON RODS JO
12 CALIBRATION CURVE OF Pb 31
13 PLUG INJECTION OF Hg USED FOR CALIBRATION 35
Ik CONTINUOUS INJECTION OF Hg FOR CALIBRATION 36
15 CALIBRATION CURVES OF Hg WITH DIFFERENT SOURCES 38
16 CALIBRATION CURVES OF Cd (Pt Loop) AND LIQUID DROP ^7
17 ABSORPTION TRACES Cd INTRODUCED ON CARBON 50
18 ABSORPTION BY As IN THE AMBIENT AIR USING THREE DIFFERENT
RESONANCE LINES 55
19 ABSORPTION TRACE OF As INDICATING DELAYED SIGNAL 55
20 EFFECT OF BED DEPTH ON ARSENIC ABSORPTION 57
21 EFFECT OF FLOW RATE ON ARSENIC ABSORPTION 58
-------
LIST OF TABLES
No.
1 Metals and Metal Compounds Used for Calibration Purposes
in the Vapor Pressure Technique 24
2 Concentrations of Lead in Laboratory Air in Baton Rouge 29
3 Mercury Concentrations in the Laboratory air (Baton Rouge,
La.) ^2
k Analytical Precision of Plug Injection Calibration
Technique ^3
5 Analytical Precision of Continuous Injection Calibration
Technique bk
6 Typical Cd Concentrations in the Laboratory Atmosphere
at Baton Rouge, La. ^9
Vi
-------
ACKNOWLEDGMENTS
The Principal Investigator wish to thank collaborators who have contri-
buted greatly to the progress of this project. These include
Dr. Charles Christian, Dr. Paul Loftin, Dr. Yvon Araktingi, Dr. Ian
Maines, Dr. Gary Hindman, Dr. Duane Wolcott, Dr. Paul Slevin, and
Mr. Robert Garcia. In addition, special thanks should be made to the
University glassblower, Mr. Charles Burlo; to the machine shop
personnel, in particular Mr. Ed Keel, J. Cass and Mr. L. Rogge. Also,
special thanks should go to Mr. George Sexton and S. Durkin, the
electronics experts of the department, who has helped in design of the
equipment and in repair and modification.
In addition, the PI would like to thank the EPA Project Officers,
Dr. Charles Walters and Dr. Carole Sawicki, for their continued help
and encouragement in developing this equipment. Thanks are also
expressed to Dr. Paul Altshuller for suggestions that he had made
concerning the application and design of our equipment.
VI1
-------
SECTION I
CONCLUSIONS
1. A research instrument capable of the determination of metals in the
10" 2 g level has been developed.
2. The instrument has been shown to be capable of the detection and
determination of Pb, Cd, Hg in the air in real time.
3° Other elements such as Na, K, Cu, Zn, As, Se have been detected in
air, but calibration techniques must be developed to obtain quantitative
data for the results.
k. Impurities in the reagents used pose a major problem at these levels.
So far it has only been possible to remove them in the equipment itself.
5. Calibration techniques at this level presented a problem, but
successful methods have been developed,
6. Further experimental work is necessary to extend the technique to
less volatile metals.
-------
SECTION II
RECOMMENDATIONS
Based on the results it appears that the more volatile elements such as
lead, mercury, cadmium, etc., can be determined directly in the atmo-
sphere. The results also indicate that a higher temperature furnace
would permit the detection and determination of many other less volatile
elements.
It is recommended that such a high temperature furnace be designed and
tested for quantitative determination of the less volatile pollutants
in the atmosphere.
The second recommendation is that a miniaturized instrument be developed
capable of being transported from one location to another to allow
direct analysis of the atmosphere at various sites.
These sites would include suspected sources of pollution. The sensitive
areas which may be exposed to high pollution or in which it is important
to know the pollution exposure level, such as school yards, streets,
factory workshops, inside homes, inside transport vehicles such as
airplanes, motor cars, etc., and remote areas where it is suspected that
pollution levels should be very low, could be easily measured with the
miniaturized equipment.
-------
SECTION III
INTRODUCTION
Scope. The scope of the work was to develop an instrument capable of
ultra sensitive determination of metals.
Purpose. The purpose was to provide an instrument capable of real time
analysis of individual metals in the atmosphere, these metals would
include these obtained in particulates and those in molecular form.
Such a device would permit the observation of the concentrations in air
of dangerous metals such as lead, mercury, cadmium, arsenic, silver,
etc. Based on this information warning systems and control mechanisms
could be created.
It was also intended to carry out preliminary work on the development of
a field method based on this technique. Such a field instrument would
be portable and capable of locating sources of pollutants. In addition,
it would provide the capability of running analyses of air which is
normally inaccessible to conventional equipment such as on top of
buildings, in machinery or in rural areas.
Background. The current methods for the determination of metals in the
atmosphere generally follow the procedure (a) scrub a known volume of
air over an efficient scrubbing agent, (b) analysis of all or part of
the scrubbing solution after extraction of the metal from the atmo-
sphere, (c) based on this analysis the concentration of metal in the
original air volume can be calculated.
This procedure suffers from several rather important disadvantages.
First, it is invariably not possible to carry out real time analyses.
The sample collection and subsequent analysis of the scrubbing mixture
always takes a finite time. In the case of some metals such as mercury,
this time is relatively short, but in the case of other metals the time
lapse may be several hours or even as long as a full day. Such long
times prevent the techniques from being used as the basis of control
systems and at best only provide data on average concentrations during
that time period. The principal usefulness of such data is the long
term evaluation of various inputs into the atmospheres. This is a
valuable contribution but does not get to the heart of pollution
control.
Second, the use of scrubbing solutions in analytical procedures always
leaves open to question the efficiency of the scrubbing mixture. Any
particular scrubbing mixture is rarely a 100$ efficient even when used
under the correct operating conditions. If the operating conditions are
less than optimum the scrubbing efficiency decreases and the analytical
error is increased.
-------
In addition, the scrubbing mixture itself may include as an impurity the
metal which is being determined. Although it is quite often possible to
correct for this source of error there is always the distinct likelihood
that the correction will not be made or that an inaccurate correction
will be made. In either event the analytical data obtained may be
inaccurate.
It can be seen quite clearly that scrubbing techniques can easily lead
to inaccurate analytical data. However, it should be pointed out that
the reason that scrubbing techniques have been used is because of the
very low concentrations of metals generally encountered in polluted air.
For example, it is common for the concentration of lead to be about one
microgram per cubic meter of air. This means that if one cubic meter of
air were scrubbed and all the lead extracted, then one microgram of lead
would be extracted and must now be determined. This in itself is quite
a feat and challenges the sensitivity of most conventional analytical
techniques. It can readily be understood that impurities in the solvent
or any inefficiency in scrubbing can lead to a major percentage error in
the final determination. For this reason it is frequently necessary to
scrub several cubic meters of air in order to get sufficient metal to
complete an analysis. This takes time and leads to the problems of real
time analysis referred to earlier. It was our belief that approximately
50-100 cc of air could be viewed at any one time and that determination
of the metal in that quantity of air would necessarily require high
sensitivity, i.e., in the order of 10~10 g. At the time of inception of
this research proposal no such analytical techniques were available, but
it was felt that the advantages of such methods justified the research
effort.
Theoretical Approach. It was decided to devote our primary attention to
the development of a technique based on atomic absorption spectroscopy.
The reasoning behind this decision was that it had already been shown
that atomic absorption spectroscopy was capable of reaching sensitivity
limits in the order of 10~7 g and in the case of a few elements 10"8 g.
Thus, it was felt that the method was already capable of high sensitiv-
ity, although not sufficiently high for direct determination of metals
in air but within striking distance.
Secondly, there is ample evidence in the literature that atomic
absorption spectroscopy enjoys a high degree of freedom from analytical
interferences and is therefore more accurate than most other analytical
techniques. The method is based on absorption of radiant energy by free
atoms. The degree of absorption is a function of the number of free,
unexcited atoms in the system and therefore to their concentration.
Thus, within the first approximation it is independent of the tempera-
ture of the system. Secondly, since many metal compounds are decomposed
in flames or other atomizers, the percentage of atoms reduced to
free atoms is fairly constant and independent of the chemical form.
Unfortunately, there are numerous exceptions to this generalization
particularly where the metal forms very stable compounds with the flame
products. There is a proportional decrease of atom population and
-------
consequently a lap in absorption signal for a. given sample concentra-
tion. This gives rise to chemical interference. It is one of the
major problems of the technique. It would be necessary to develop a
procedure which was independent of the chemical form of the metal. The
analytical signal from any one metal such as lead in the atmosphere
would not depend on whether the lead existed as tetraethyl lead, lead
bromide, or lead oxide or any other lead compound. It was felt that
atomic absorption spectroscopy could be capable of providing the basis
for such a procedure.
The proposed technique, therefore, has two very strong points of
recommendation: (l) high sensitivity and (2) a high degree of freedom
from interference effects and therefore a high analytical accuracy.
Factors Controlling Sensitivity. The mathematical relationship
relating the degree of absorption and the number of atoms in the light
path is given by
= -ns Hf
me
o
CD
Where JlCvdv equals the total amount of light absorbed at frequency v,
2 o
ire
— have the conventional meanings, c is speed of light, e is change of
the electron, N is the number of atoms in the light path and f is the
oscillator strength of the energy transition involved in the absorption.
Generally, it can be assumed that function of the atomizer is to charge
the sample element from an ionic form or an organometallic form to a
, . , . , atomizer ,. _ .
free atomic form, i.e., sample > free atoms N.
The efficiency of producing free atoms from the sample is controlled
entirely by the atomizer used. At the time that this work was
initiated, the standard atomizer system used in atomic absorption
spectroscopy was the flame. Sensitivities in the region of 10"6 to
10~8 were common using flame atomizers.
If time is taken to envision the situation of a free atom liberated in
a flame atomizer it can be readily noted that the free atom is in an
extremely hot plasma surrounded by very reactive chemicals. It is not
surprising, therefore, that the free atoms rapidly combine with some
other chemicals present in the flame and form a compound such as an
oxide or hydroxide, etc. When that happens the atom is removed from
the free atomic population. The latter is, therefore, rapidly lost.
Earlier calculations have shown that approximately 1 atom in 10a is
actually reduced to the free atomic state in the flame.1 It was felt
that the atomizer is a very weak link in the process used in atomic
absorption spectroscopy.
-------
Early work by L'vov2 had shown that a carbon tube could be used as an
atomization system and that high sensitivities in the order of 10"11 g
were achievable. However, all his work indicated that quantitative
analyses were extremely difficult because of the poor reducibility of
the system.
L'vov's work was neglected for many years, but it was felt that it
could serve as the basis for a quantitative method of analysis.
The approach taken in this work was to use an atomizer composed of
heated carbon in which atomization took place, the free atoms then
issuing into an absorption tube at right angles to the atomization tube
as indicated in Figure 1. Indications were that separation of the
atomization chamber and the absorption chamber would eliminate many of
the causes of interference encountered in the L'vov process. For
example, the combustion products when organic compounds were analysed
would be burned in the atomization stage and reduced to carbon monoxide.
This complicated procedure always produces a high background; however,
with the process used, this would take place out of the absorption tube
and would not be a source of interference.
Atomization Step. The atomization of the free atoms was based on the
use of hot carbon over which the air sample was drawn. The air sample
was composed, naturally, of nitrogen plus oxygen and the following
reaction with carbon took place: C + N2 + 02 -> CO + N2
It
CO 2
The carbon monoxide formed was in equilibrium with carbon dioxide and
at temperatures above 900°C the equilibrium was thrown heavily over to
the direction of carbon monoxide. Hence, at temperatures greater than
900°C the atomization step would involve the use of hot carbon +
carbon monoxide. The system which is illustrated in Figure 1 used an
rf coil. This coil induced heat in the carbon rods or pieces which
rapidly attained a temperature of 1200-1400°C, according to the power
of the radio frequency (rf) generator used. The advantage of using any
rf generator as a power source was that the carbon bed was evenly
heated across its diameter and that it was rapidly brought to tempera-
ture by turning on the rf generator.
In practice ambient air was pulled over the hot carbon bed and reduction
to free atoms of metal pollutants in the atmosphere took place. These
free atoms issued into the absorption tube. By using the correct
system of hollow cathodes and detectors, concentrations of the metal
in the atmosphere was calculated from the measured atomic absorption
signal.
-------
t
EXHAUST PORT
I
^3 m
R. F. COIL-
TEFLON PLUG-
HEVI DUTY HEATERS
' 1/4 in. CARBON RODS
TEFLON CAP
•AIRSAMPLE IN
Figure 1. Schematic diagram of the atomizer.
-------
SECTION IV
EQUIPMENT
Optical System and Components. The major components and the optical
layout of the equipment are illustrated in Figure 2.
Source. Barnes Demountable hollow cathode
Monochromator Jarrel Ash, Model 82-500, half-meter monochromator
Detectors 1P28, 1P22, R106
Atomizer built during the investigation
Rf generator Lepel 5 kilowatt generator
Amplifier system, PAR amplifier, Model 126 equipped with PAR photometric
preamp Model 18^.
Recorder Beckman Model 10005 10" potentiometric strip chart.
Optical System. The optical system used was essentially that of single
beam optics. Although there is a strong tendency to work with double
beam systems in most spectroscopic work, it has been shown that there
is very little advantage to the double beam system over the single beam
system in atomic absorption. It is seldom possible to put a "reference"
sample in the reference beam. The only commercial instrument effec-
tively employing double beam optics employs pseudo-double beam optics
where the reference beam is an empty path. Under these circumstances
it was felt that there was not sufficient advantage to the double beam
system to justify its incorporation in the system.1
Construction of the Instrument
Hollow Cathode. The Barnes Demountable Hollow Cathode was preferred to
sealed hollow cathodes for several reasons. The first was that having
bought the Barnes system for approximately $1,000, new hollow
cathodes can be purchased for approximately $10 each. This is an
important consideration when it is intended to examine many elements in
the periodic table. The alternate route of using sealed hollow cathodes
and the necessary power supply would present a smaller initial cost.
The power supply can be bought for approximately $500, and the individ-
ual hollow cathodes cost approximately $150 each. But, a study
involving twenty elements would necessarily require the purchase of
twenty different hollow cathodes to provide at least one of the
cathodes for each element. It is well known that sealed hollow cathodes
have a finite life which may end without warning. This breakdown may
hold up work for weeks and perhaps months under some circumstances. The
problem was avoided by using demountable cathodes. These can be
fabricated in the lab. Hence, these hollow cathodes are cheaper and
better to use than sealed lamps.
-------
DETECTOR
POWER SUPPLY
MONOCHROMATOR
DETECTOR
PLANO-CONVEX LENS
AIR INLET
SYSTEM
R.F. COILS
o o o o o
- CONVEX LENS
Figure 2. Schematic diagram of equipment.
9
-------
Monochromator. A Jarrell Ash Model 82-500 half-meter monochromator with
grating and adjustable slits was purchased from Fisher Scientific. The
monochromator was fitted with an inlet and exit port to allow flushing
with nitrogen or argon. This was an important feature if work on
elements with resonance lines at wavelengths less than 2100 A is antic-
ipated j because removal of oxygen from the monochromator was necessary
for all such elements. These elements included mercury, arsenic,
selenium and sulfur. Our intention was to study some of the elements.
Detector. The detectors used were the 1P28, 1P22 and R106. The
response curves of two of these detectors is shown in Figure 3- As can
be seen, it was an advantage to use the 1P28 for elements with resonance
lines at wavelengths greater than 2100 A and to use the R106 detector
for studies involving elements with resonance lines shorter than 2100 A.
The 1P22 was used for studies of potassium. The housing for the
photomultiplier in the Jarrell Ash instrument was completely surrounded
by metal and it was felt that this would shield it from any stray
radiation from the rf generator and coil. It was found later that this
was an erroneous assumption and special precautions had to be taken to
shield the monochromator from this radiation.
Amplifier. The amplifier used was a PAR amplifier Model 126 equipped
with PAR Photometer Preamplifier Model 18^. The unit could be operated
as a "Lock-In" amplifier or as a wide band AC amplifier over the range
0.2 Hz to 210 kHz. Since the output from the detector and amplifier
system was in the range 0-10 volts, this necessitated the use of a
helipot to match the PAR amplifier to our recorder which has a maximum
input level of 100 mV.
Read-out System. The read-out system used was the Beckman Model 10005
10 in. potentiometric Strip Chart Recorder.
Atomizer. The most important instrumental development to be accom-
plished was the construction of an efficient atomizer. As has been
pointed out, the flame atomizer is capable of analytical sensitivities
in the range of 10"s to 10"8 g. However, as described earlier,
indications were a long path system using carbon as the atomizing agent
offered the promise of a significant improvement in the efficiency of
atomization. It was felt that when air was pulled over the hot carbon
rods that this would produce a system of carbon plus carbon monoxide.
The hot carbon monoxide, being a powerful reducing agent with chemical
energy of approximately 11 electron volts, would probably be sufficient
to reduce the metals in most, but not all, inorganic compounds. The
first system used is shown in Figure 1. It consisted essentially of
[CO]Vd = 11.1 eV
two quartz tubes with half-inch internal diameter joined in the.center
of one tube to form a T-piece. The short side section had the carbon
rods.
10
-------
a
cc
_i
-------
In the initial work, the quartz tubes were held in a horizontal plane.
The carbon rods used were approximately six inches long and laid inside
the side piece so as to fill it's crossection as far as possible.
Numerous types of spectroscopically pure carbon rods were utilized in
this experiment. The analytical data will be presented later in this
report.
Much of the preliminary data on lead was obtained using this design.
However, it was found later that the use of a considerably smaller
system was at least as effective as the system described above.
One of the difficulties with the horizontal loading system was that the
flowing air sample tended to channel across the top part of the side
tube and to burn out that part of the carbon bed preferentially. This
meant that the carbon bed was burned out while much of the carbon was
still unused, resulting in a reduced lifetime for the carbon bed.
In later models the size of the carbon rods was reduced to small pieces
J" long. The current equipment is illustrated in Figure k.
It was also found that using the system in a vertical plane held
several advantages. For example, it was easier to mount and screen the
rf coil thus reducing stray radiation; secondly, since top loading was
possible it was much easier to load the bed initially and to refill it
during operation since dismantling was not necessary. Thirdly, it was
found for later work using liquid samples, that injection of the
sample was greatly facilitated by a vertical loading system. When the
change was made from a horizontal atomizer to a vertical atomizer, it
was also decided to change from carbon rods approximately six inches
long to carbon pieces approximately J" long. As previously indicated
the use of long carbon rods increased channeling between the carbon
rods as well as at the top of the atomizer system by the incoming air.
This caused uneven burning of the carbon and inefficient reduction of
the air. When the change was made to carbon pieces the bed became much
more uniform and channeling was greatly reduced.
Rf Coil and Generator. The rf generator was a standard Lepel 5 kilowatt
generator operating at a frequency of 3 megacycles. The rf coil was
hollow copper tubing through which water was pumped. The coil consisted
of 10 turns, diameter 35 mm with the total coil length of 120 mm.
Absorption Tube Heater. The absorption tube was made of quartz and,
therefore, did not couple to the rf generator. It was necessary to heat
it by some other means. The method used was to wind it with nichrome
JOO W resistance wire and to cast this in a plaster of asbestos. This
was then dried out leaving the side arm heated by an electrical fila-
ment. The filament itself was powered off 110 V line together with a
variac in order to control the voltage to the filament.
Rf Shielding. In practice it was found that the rf generator and rf
coil emitted a considerable amount of rf stray radiation. This
12
-------
AIR INLET
36C
1
mm
25 mm
C
b
INNER SLEEVE
R F COIL
RESISTANCE WIRE
\
ASBESTOS \
X \
1 O O O O O O O O O O (J O
>
IWW-WQW^ SFAI AfsSFMRlY
Av^
Mt
o
»o
o
0
o
o
o
o
o
Oi
pp
iV?
MA?
FIBfW
„„.
l8^
VI Ml IV
.CARBON CHUNKS
/
O
O
0
o
0
o
o
£ QUARTZ FLAT'
2 EXHAUST PORT ^
O \^
LIGHT PATH 0
300 ao o a o a o oo ooo
-------
radiation was picked up by many components of the system and led to a
very erratic and unstable read-out. Abrupt changes in base line took
place for no apparent reason. Also there were slow swings in base line
drift and short swings in base line drift producing an inoperable
system.
The first approach to correcting this problem was to replace all
electrical leads with shielded electrical leads between the various
components of the system. This led to some reduction in the background
variation, but was not sufficient in itself. Secondly, all components
were grounded individually. This, again, led to an improvement in the
problem but anytime the rf generator was turned on there was invariably
a major change in the signal level of the recorder.
This was finally overcome by building a copper box to completely enclose
the atomizer and the rf coil, which prevented stray radiation from the
coil from reaching other components of the equipment. In addition, the
leads from the coil to the rf generator were enclosed in copper casing.
This, again, caused a significant decrease in the stray radiation
intensity.
An extra piece of shielding was put around the detector housing. This,
again, caused a reduction in background variation. Indications were the
normal metal housing of the detector acted as a shield and picked up
stray radiation from the rf system, but then became a transmitter, at a
lower intensity, of the radiation that it picked up. The problem was
overcome by putting a second shield outside and completely around the
metal housing system.
Pure Air System. In practice, it was necessary to compare ambient air
with some other standard, preferably, a 'pure' air. It was not possible
to buy pure air, so it was necessary to provide some system whereby we
could produce our own pure air on demand and at a reasonable price.
Numerous scrubbing agents were tried, including nitric acid, hydrochloric
acid, aqua regia, silica gel, and alumina. None of these were completely
successful in removing the metal contaminates from air pulled" through
them.
The best purifier was found to be activated charcoal. This effectively
removed, as far as could be determined, all of the metal impurities
from the air when the system was used under the correct operating
conditions.
Unfortunately, it was found that the system rapidly became saturated
with water vapor in the atmosphere and that its capacity was not great.
In practice, this was a severe disadvantage.
The first attempt to overcome this problem involved heating the carbon
to a temperature above 110 C at which point water would not condense
and would therefore not saturate the scrubber.
-------
Further studies showed that this system was not satisfactory. As
predicted the water was not held very tightly by the carbon and the
latter did not become saturated. Unfortunately, other volatile
compounds such as mercury salts were also retained by the carbon
scrubber at this temperature. The scrubber was not completely satis-
factory in providing pure air. The effect of the increased tempera-
ture, therefore, was to prolong the life of the carbon purifier, but
also made the bed ineffective in removing certain metals from the air.
The problem was later overcome by putting a tube of calcium chloride
approximately three feet long in line before the activated charcoal
bed. Ambient air was passed through the calcium chloride bed, to
remove excessive water and then passed through the activated carbon bed
where the metals were removed. Finally a millipore filter stage was
added to trap any particulates or carbon dust. This system, in practice,
was found to be very successful and provided as easy supply of pure air
free of water vapor and metal contaminates. No studies were made on
the removal by this train of contaminates such as carbon monoxide or
carbon dioxide, because these impurities would not interfere with the
system. The system is illustrated in Figure 5-
15
-------
NHLLIPORE
FILTER
CONSTANT
PRFSSIIRFHFAD
CALCIUM CHLORIDE
DRYING COLUMN
COAL
MN .
1 1
—
o
o
o
o
o
OfftQ
|P^8
*&$
t$S(l
J?jjj
o EXHAUS"
0 I
o
0
o
o
0
ABSORPTION TUBE
COMPRESSED AIR TANK
Figure 5. Pure air system.
16
-------
SECTION V
OPERATIONAL PARAMETERS
The Feasibility of the Method. The first information to be sought was
if the system was capable of detecting various metal-containing
impurities in the air.
Nine elements were chosen based on their present importance as metal
pollutants and their future possible importance in this area. These
metals were lead, mercury, cadmium, arsenic, selenium, sodium,
potassium, copper, and zinc.
The pure air system was set up so that it could be connected to the
entrance port of the atomizer section as desired by turning a glass
tap. Alternately, ambient air could be drawn into the system and the
signal measured for the particular element being determined.
First, the system was set up with pure air flowing through it. The
base line which was presumably zero absorption or at least
pseudo-zero absorption was recorded continuously. The source of air
was then switched from pure air to ambient air and the absorption
signal traced out after a sufficient time to establish a steady
absorption signal. The sample was then switched from ambient air to
pure air and the signal was measured continuously until it again
stabilized.
The data obtained were very encouraging. Traces of the absorption
spectra are shown in Figure 6.
These data indicated that the method was capable of determining
various metals in air directly and it was felt at this time that
further studies on producing a reliable analytical system were
justified.
One very important observation was that these data were purely
qualitative. They indicated that indeed these metals were present
in air and could be detected and measured, but the data gave no
indication of what these signals represented in terms of metal
concentration or how much of these metals were present in the air.
The most important problem to be overcome was that of calibration,
i.e., translating the absorption signal into concentrations of the
particular metal.
It was felt that the problems of calibration were difficult and varied
from one metal to another. It was decided that the best approach
was to study calibration techniques for one element at a time,
thereby learning the analytical problems involved in the determination
of that element. After several studies it was anticipated that a
general calibration technique applicable to all metals would be
developed.
-------
Figure 6. Absorption by Metals in Air (Directly)
o
2138 2288
y*vu f*~4
fc ,\*sfaf \t\im jtt^tf *-M/j*J
2537
A^A//'"'
Zinc
Cadmium
Mercury
1931
2833
O -
5895
Potassium
Sodium
Lead
1961
Copper
Arsenic
Selenium
Time
18
-------
For this reason, it was decided to study the element of greatest
interest in the air pollution field at this time, i.e., lead.
19
-------
SECTION VI
THE DIRECT DETERMINATION OF LEAD IN THE ATMOSPHERE
Choice of absorption line. There were two resonance absorption lines
for lead. These were at 28J3 A and 2l?0 A. The most commonly used
line was at 28J3 A. The forte of this line was that it is very
stable and easy to work with. From an analytical point of view, it
was subject to less interferences from other absorbants in the system
and was therefore relatively free from interference, thus generating
accurate data. Also the line had a sufficiently high oscillator line
to generate high analytical sensitivity for the analysis of most
samples.
In contrast, the 21JO A line had a higher oscillator strength and
therefore should be capable of giving better sensitivity. Inasmuch
as we felt that we would be striving for sensitivity, it was decided
to direct our efforts to the use of this latter line. It was found
in practice, that water vapor in the air produced H2 and CO when
passing through the carbon bed. The hydrogen molecules produced
absorbed over a broad wavelength range including the 2170 A line of
lead. This absorption produced a direct interference to the method
but later it was found that the molecular absorption at wavelengths
coincidence with atomic absorption lines could be corrected for
and could be eliminated as an analytical problem.
Calibration of Lead. Altshuller Method.3 The first method used for
calibration was that developed by Altshuller. The schematic diagram
of the equipment is shown in Figure 7.
Tetraethyl lead (T.E.L.) was placed in the container and weighed
accurately. It was then maintained at a temperature of 35.5°C so
that tetraethyl lead evaporated off into the air stream which
flowed into the equipment. The evaporation rate was such that the
lead gave an absorption signal easily measured by the equipment.
The constant temperature was maintained until sufficient tetraethyl
lead had been evaporated off to cause a measurable loss in weight
of the T.E.L.
In the early experiments, this took approximately k or 5 days. However,
as the equipment was improved by increasing atomizer temperature, etc.,
the sensitivity improved by an order of magnitude. This meant that it
was necessary to control evaporation of the sample of tetraethyl lead
for a period of lj-0-50 days in order to evaporate off a sufficient
20
-------
DILUENT
GAS OUT
DIFFUSION
PATH
DILUENT
GAS IN
TETRAETHYL
LEAD
Figure 7. Altshuller Equipment for Calibration with tEL
21
-------
quantity to weigh. After further improvement by another order of
magnitude the time period necessary for evaporation also increased an
order of magnitude to between JOO and kOO days.
Clearly the method was now not suitable for calibration of this
instrument when it was operated to give high sensitivity.
Another problem with the method was that all material evaporated off
during the process was weighed as tetraethyl lead. This included
light hydrocarbon impurities present, some of which were evaporated
off preferentially but weighed as T.E.L. This generated an error
in the procedure which was difficult to evaluate. The technique
was not used for further studies.
Calibration Using Compressed Air. The basis of this technique was
to pass compressed air from a cylinder into the instrument and measure
the degree of absorption by the lead entrained in this compressed air.
It was assumed that the sample was homogeneous and that this air
measurement would be constant throughout the entire cylinder.
After taking the absorption measurement, the air was passed through
a dilute HN03 scrubbing solution in which the entrained lead was
scrubbed out the sample. The scrubbing solution was then analyzed
using spectrophotometrie analysis* and the lead concentration of
the scrubbing solution determined. From this determination the lead
concentration in the compressed air was calculated.
One major difficulty with this procedure was that it was time
consuming, i.e., it took a whole day to scrube out sufficient air
to measure the lead entrained. Also only one datum point was
obtained and that was the concentration of the lead in compressed
air. A calibration curve could only be constructed by drawing a
line from this point through the point of origin. This technique
is notoriously dangerous and is not in anyway accurate at
concentrations significantly different from those in the compressed
air.
Lead Vapor Pressure Techniques. The Altshuller technique for the
calibration of lead was based on the fact that tetraethyl lead is
a volatile compound. There are a number of other volatile lead
compounds available and it should be possible to use the Altshuller
technique with these compounds.
22
-------
Further, it was felt that since the sensitivity of the method appeared
to be very high that it might be possible to simply heat up many pure
metals and measure the metal that evaporated off into the air stream.
Based on literature data on vapor pressure it should be possible to
calculate the vapor pressure at the temperatures used of the metal or
the metal compound and from that calculate the concentration of metal
evaporated during the process.
The compounds tested are listed in Table 1.
The data obtained were very discouraging. First, it was found that
many of these compounds decomposed at increased temperature. Many
reacted with the hot air which had to be used in order to generate
carbon monoxide in the instrument. When the compound decomposed or
oxidized in this, the original compound changed chemically and the
vapor pressure data were no longer valid. Under these conditions it
was impossible to calculate the real vapor pressure of the compound
used.
This was even true when pure metal was used because frequently a thin
film of molecules was on the metal surface. These molecules were
chemically different from the pure metal itself and with a different
vapor pressure. Also, the pure metal frequently reacted with the
nitrogen or oxygen or other impurities in the air used to pass over the
heated metal and again changed chemically, vitiating vapor pressure
data.
This calibration method could have been pursued further and used
successfully. However, it appeared that the difficulties involved
would be extensive and study may have developed into a whole new
research project for which little time was available.
It was therefore decided to abandon the use of this technique after
preliminary studies.
The Platinum Wire Technique. In this method a platinum loop was
incorporated into the entrance of the atomizer system. The equipment
is illustrated in Figures 8 and 9-
In this technique a known volume of a solution of the metal under test
was spotted onto the platinum wire with a microsyringe. The volume used
was in the order of 1 or 2 (J.1. Any volume greater than this tended to
drop off the wire and to provide incorrect data. Any volume less than
this was difficult to measure accurately.
After loading the platinum wire it was lowered into the furnace. The
current was then turned on and the platinum wire heated to approximately
2000°C. The heating step took approximately 2 seconds during which time
the liquid sample evaporated as did the metal components. The vaporized
metal was then swept into the carbon bed and atomized by the flowing air
-------
TABLE 1
Metals and Metal Compounds Used for Calibration Purposes in the
Vapor Pressure Technique
Compound
Comment
Arsenic metal
Arsenic trioxide
Cadmium chloride
Cadmium sulfide
Cadmium sulfate
Copper metal
Copper oxide
Cuprous chloride
Cupric chloride
Cupric iodide
Lead chloride
Lead phosphate
Lead carbonate
Mercuric chloride
Mercuric sulfate
Mercuric nitrate
Insufficient temperature control*
Insufficient temperature control
Successful volatilization
Successful volatilization
Successful volatilization
Reacted with trace 02 in N2 to give CuO
Unable to volatilize
Preferential volatilization of impurities
Decomposed to CuO
Decomposed to CuO
Successful volatilization
Successful volatilization
Successful volatilization
Successful volatilization
Successful volatilization
Successful volatilization
* Note: Very small changes (e.g., 10°C) caused significant
changes in vapor pressure. We were unable to control the tem-
perature this well for the extended time periods necessary to
get reliable results.
-------
6
D
Platinum Loop Injector:
A) ammeter
B) variable transformer
C) filament transformer
D) Platinum loop
Figure 8. Electrical Diagram of the Pt Wire Used for Calibration
-------
1 f lj
o
o
0
0
j
1
1
1
I
|
•* «.
j
^A
>•
^4 !
$f 5V_ ,rj
^^
^i^r~j-JTJJC/
1
O
jo
o
o
B
The platinum loop is at point A immediately above the carbon bed
B. The electrical inlets are at point C.
Figure 9. Schematic Diagram of the Pt Wire Calibration Equipment
26
-------
stream in the normal way. It was assumed that all the sample that was
put onto the platinum wire ended up in the furnace and contributed to
the absorption signal.
This technique worked well for a number of elements and gave some very
informative results. One problem involved was that volatile metals
such as lead tended to evaporate at temperatures much lower than that
of the very hot platinum wire. Lead was sometimes lost when the
platinum wire was lowered into place because of the heat radiated by
the bed. Unless the technique was carefully controlled low results
were obtained because of this loss of sample by exposure to the hot
carbon bed prior to heating the Ft wire. No calibration curves for
lead were completed using the Ft wire technique.
Liquid Drop Technique. In this technique a small volume of sample, i.e.3
1 to 5 jj,l was injected onto the hot carbon bed using a microliter
syringe. The solution immediately evaporated and the vaporized compo-
nents were swept through the carbon bed atomizer in the normal way. The
equipment is illustrated in Figure 10.
It can be calculated that the 2 jil liquid sample after hitting the bed
evaporated to a volume of approximately 10 to 20 cc. This expansion
took place extremely rapidly and amounted to a miniature explosion.
However, it was found that very little sample was lost during this
time and the method was quite reliable and reproducible.
Calibration data were obtained by assuming that all the lead introduced
in the sample as a liquid entered the atomizer. Hence, the concen-
tration of lead could be calculated and the degree of absorption
correlated with this concentration. It was also assumed that the liquid
sample rapidly became gaseous and so approached that of an actual gas or
air sample as it entered the atomizer. We believe that this eliminated
error caused by different physical types of sample.
The limiting factor of the technique was reproducing the volume of the
solution introduced to the sample atomizer. It was difficult to
introduce samples of 5 M-l reproducibily, i.e., to within an error of 5$
using a conventional microliter syringe. Frequently, part of the sample
would drop off the syringe tip, leaving a small part of it which stayed
on the syringe tip and did not drop off. Instead it evaporated leaving
some of the metal behind; in this case a low result was obtained. At
other times a sample drop would not fall straight down the column but
hit the side of the atomizer. Again, because of the low temperature of
the side of the atomizer, slow atomization took place and low results
were obtained.
However, the problem was largely overcome by using a Brummond micro
pipet modified for positive full displacement. This syringe was capable
of reproducing the volume delivered to the atomizer very precisely.
Sample volumes of 2 ^1 -were easily handled and reproducible data
obtained.
27
-------
o
o
o
o
o
o
o
o
o
o
o
o
Figure 10. Equipment for Calibration by the Liquid Drop Technique
28
-------
Contamination on Carbon Rods. The carbon rods as delivered from various
manufacturers were tested for metal impurity. It was found that in all
cases the lead concentration was sufficiently high to give a full scale
deflection immediately after the carbon bed was heated. However, after
prolonged heating, i.e., about an hour at llKDOoC, the lead evaporated
off and the carbon rod became pure enough to give a steady absorption
signal which approached zero.
A tracing of typical absorption curves derived from heating the carbon
rod is shown in Figure 11. Various types of carbon rods from different
manufacturers were tested. The data indicate that the conventional
spectrographic carbon rods were far too contaminated to be used under
normal circumstances. However, high purity carbon rods were quite
satisfactory after suitable cleanup in the equipment for a number of
elements tested. Unfortunately, we were not able to cleanup zinc or
copper reproducibly.
One other problem with this technique was that frequently lead was
reabsorbed from the atmosphere after the rods had been cleaned. This
was particularly bad if the instrument was not used for several hours
or days. However, the problem was easily corrected by reheating the
carbon rods up to temperature and evaporating off the reabsorbed lead
until the absorption signal was again steady and approached zero.
Typical Calibration Curves. A typical calibration curve is shown in
Figure 12. As can be seen the dynamic range was 0-60 p,g m3.
Data for the Baton Rouge Area. Based on the calibration techniques
which were developed and the absorption signals for ambient air in
laboratories, the following data were obtained for lead concentrations
in Baton Rouge.
Table 2
Concentrations of Lead in Laboratory Air in Baton Rouge
Date Absorbance Concentration Pb
8-5-69
9-2-69
10-14-69
11-7-69
.040
.300
.068
.050
1.4 fig/m3
15.0 |ig/m3
2.8 )J,g/m3
2.0 |xg/m3
29
-------
120
TIME, minutes
150
180
210
240
Figure 11. Typical traces of absorption caused by metal bleedoff from carbon rods.
-------
100
80
60
u
CQ
(£.
I 40
20
10 20 30 40 50
MICROGRAMS PER CUBIC METER
60
Figure 12. Calibration curve of Pb.
31
-------
It can be seen that reliable data can be obtained in the concentration
range encountered in normal atmospheres. This illustrated that the
technique was capable of real time analysis without prior concentration
or normal air samples.
Based on the calibration data obtained the sensitivity of the procedure
was 10"X:L g of lead for air samples and based on the calibration data
using the liquid drop technique sensitivity for liquid samples was
10~12 (ig of lead in a 2 Rl sample size. The sensitivity in this case
is defined as the concentration that gives 1$ absorption. The precision
of the technique (2 a 95$ confidence) at the 6 u-g/m3 concentration level
was 0.2
-------
SECTION VII
DETERMINATION OF MERCURY IN THE ATMOSPHERE
The next element chosen for study was mercury. This element was
chosen for three reasons: first, it is a very important metal
pollutant in the atmosphere, it is known to be highly toxic, even
at low levels. Second, it is known that mercury has been detected
by atomic absorption at low levels and it therefore appeared
feasible that this instrument could be applied to reducing the
level sufficiently low to enable direct analysis of mercury in
the atmosphere. All current methods involved in the determination
of mercury require a pre-concentration step followed by analysis
by one of several conventional methods. The third reason was that
the best resonance line for mercury was at 18^9 A. Hitherto, most
of the work carried out on the determination of mercury has been
made by using the resonance line at 2533 A. The oscillator strength
of the 18^9 A line is approximately 50 times as strong as the 2533 A
line. It is, therefore, reasonable to assume that the detection
limit at 18^9 A would be at least an order of magnitude better than
the detection limit at 2533 A.
In addition to this, the investigation of the use of the line at
18^9 A would provide the extra challenge of having to work in the
vacuum ultraviolet region of the spectrum. A number of elements
have very intense absorption lines in the vacuum ultraviolet. These
elements include arsenic, selenium and sulfur, all of which are very
important in the field of air pollution. Their determination has
been difficult in the past because of lack of sensitivity. It was,
therefore, considered to be an attractive challenge to work in the
vacuum ultraviolet and study the problems involved in the use of
this part of the spectrum. It was hoped that by successfully
solving these problems the determination of arsenic and selenium
would be much easier than the present status of affairs.
Direct Determination of Mercury in the Atmosphere. By drawing
ambient air through the instrument it was possible to detect
absorption signals caused by mercury present in the air. Typical
absorption traces are shown in Figure 6. The data showed that
it was experimentally possible to measure the absorption signal
caused by the presence of mercury in the air. This data proved
that mercury could be determined directly using this equipment.
It remained to calibrate the signal by one of the techniques
already developed for lead and to correct the signal for molecular
absorption in the vicinity of the mercury line.
33
-------
Calibration Procedure. Mercury was one of the easier metals to
handled inasmuch as it is readily volatile at low temperatures and
it is available in fairly pure chemical form. Two approaches were
taken to the calibration problem, one was to enclose mercury metal
in a well-stoppered bottle and allow the metal to come to equilib-
rium with mercury vapor above it. The temperature of the partially
filled bottle was rigidly monitored. After a suitable length of
time a known volume of the air above the mercury was extracted using
a syringe. The weight of mercury in the volume extracted was
calculated based on the vapor pressure of mercury metal at the
temperature used. The air withdrawn from the bottle was then
injected as a slug into the equipment by way of the pur air stream
entering the carbon bed. This is illustrated in Figure 1J. The
mercury metal travelled through the carbon bed and into the
absorption tube. Here the absorption signal was measured in the
normal way.
By varying the volume of mercury and air withdrawn from the bottle
and injected into the sample train the weight of mercury injected
into the equipment could be varied. Using this technique the
relationship between the absorption signal and weight of mercury was
determined by the standard procedure.
The second technique used was to inject the mercury vapor contin-
uously over a relatively slow time period of several minutes into a
flowing air stream. The mixture of Hg and air provided a continuous
absorption signal. By varying flow rates the mercury concentration
was varied. The equipment is illustrated in Figure 1^.
Results of Calibration Experiments. The early results of this
experiment were very disappointing inothat they indicated the
absorption by mercury using the 18^9 A line was not particularly
good. Moreover, it was not better than the absorption found using
the 2533 A line, which was in conflict with theoretical predictions.
It was concluded at this time that we were operating under some
incorrect experimental conditions; therefore, we continued to study
this problem with the hope of elucidating the reason for the low
sensitivity.
One of the salient features of the calibration curves was the very
flat relationship between the degree of absorption and quantity of
mercury introduced into the AA instrument.
This kind of data was typical of that obtained when the line source
of the hollow cathode was self-reversed.
-------
CLEAN
Xl//?
SEPTUM
i tin/I a i',\
?/// t > t ' 11 / >\
I/// ' r < i /1 >i\
V i t 11111 / >-it\
TO
ATOMIZER
^MERCURY VAPOR
"PLUG"
Figure 13. Plug Injection of Hg Used for Calibration
35
-------
DILUENT G/4S—1
SEPTUM
4— DILUENT GAS
PLUS
MERCURY VAPOR
SATURATED
'MERCURY VAPOR
SYRINGE
DRIVE
Figure lU. Continuous Injection of Hg for Calibration
-------
Under these conditions, the hollow cathode itself generated a cloud of
free mercury atoms which float in front of the cathode discharge. Free
atoms absorb the very center of the emitted resonance line.
The linewidth of the emitted resonance line was somewhat less than
l/10th of an angstrom. Not all of this line can be absorbed by free
atoms in the light path. In fact, normally only a linewidth of about
.01 of an angstrom would be absorbed by free atoms. It should be noted
that the actual linewidth absorbed depends on the temperature of the
absorbing atoms; increasing temperatures causes an increase in the
linewidth absorbed. However, the heart of the emission line is that
part which is most easily absorbed by free mercury atoms. If this part
of the emission line is absorbed in the hollow cathode, then it is not
possible for it to be absorbed by free metal atoms in the sample.
The net result was that although the hollow cathode appeared to be
emitting at its resonance line, in fact the center of the emitted line
had been absorbed in the lamp itself, i.e., the line was reversed.
Under these circumstances, the mercury atoms in the sample could only
absorb a little radiation no matter what was the concentration of the
metal atoms. The net result was a flat calibration curve similar to
those obtained in our first experimental studies.
Several other types of lamps were used for preparing calibration curves.
It was found in each case that if a sealed lamp was used that the cali-
bration curve was very flat and not particularly useful for the analyt-
ical determination of mercury.
However, when the Barnes demountable hollow cathode was used, a much
steeper relationship between absorption and metal concentration was
obtained. The Barnes demountable hollow cathod was not sealed. Rather
the filler gas flowed steadily through the cathode itself, when in use.
Under these circumstances any mercury vapor which was generated by
sputtering of the hollow cathode was not allowed to accumulate inside
the cathode itself, but was swept out with the filler gas during its
normal functioning operation.
The degree of self-reversal was very much reduced and the heart of the
resonance line was left relatively unabsorbed by the lamp itself.
Typical calibration traces using different types of mercury lamps are
illustrated in Figure 15- It can be seen that the Barnes demountable
hollow cathode was much more successful in application than any of the
other sources routinely used. It should be noted, however, that this
difference in calibration curve was not noted to the same degree when
the 2533 A resonance line was used. This was because the oscillator
strength of the 2533 A line was much less than the lQk$ A line and
self-reversal was much less of a problem because entrapped mercury metal
in the sealed hollow cathode absorbed out a much smaller fraction of the
resonance line used.
37
-------
LU
O
1
cc
O
LAMP USED
FUSED DHC - DEMOUNTABLE HOLLOW CATHODE USING
A FUSED Hg SALT IN THE CATHODE
AMAL DHC - DEMOUNTABLE HOLLOW CATHODE USING
AMALGAMATED Hg IN THE CATHODE
OZ - MERCURY LAMP
PEN - Hg PEN LAMP
100 200
CONCENTRATION, jig/m3
Figure 15. Calibration curves of Hg with different sources.
-------
It was concluded that if mercury were to be determined with the maximum
sensitivity that the 18^9 A line should be used because of the higher
oscillator strength of this line. But, an. unsealed hollow cathode such
as the Barnes demountable hollow cathode must be used in order to avoid
self-reversal problems encountered in sealed mounts.
Correction for Molecular Absorption. At 18^9 A there were a number of
compounds which might absorb quite strongly. Included in these were
water, oxygen and hydrogen. Under normal operating conditions, all
oxygen in the air was burned to carbon monoxide. All water in the air
was converted to carbon monoxide and hydrogen, but any organic material
in the air was converted to carbon monoxide and hydrogen leaving a
significant amount of hydrogen in the gas phase. It was important,
therefore, to correct for the variable amount of hydrogen present after
the wet air had passed through the atomizer.
Correction for Molecular Absorptions at 18^9 A. The procedure used for
making this correction was to take a hydrogen lamp or a deuterium lamp
which emitted broad band radiation at the wavelength of the mercury
resonance line l8M? A.5 The instrument was tuned to the resonance
wavelength using the Barnes demountable mercury cathode. This cathode
was then replaced with a deuterium lamp and the wavelength setting held
constant. The spectral line width of the monochromator slit system was
kept at 1.25 A, hence this was the width of the wavelength range falling
on the detector. The absorption line width of mercury atoms was approxi-
mately 0.01 A wide. Because the absorption line width was extremely
narrow compared to the emission line width, the mercury atoms could
only absorb a small amount of the light falling on the detector. Even
if the system was saturated with mercury atoms, the total amount
of light falling on the detector would be decreased by an exceedingly
small amount.
For example, suppose the spectral line width was 1 A wide and the
mercury absorbed out of this a line of .01 A wide, then with complete
absorption of that narrow band this signal would decrease by 1$. This
is a very small change in signal and probably within the accuracy of
the equipment being used.
On the other hand, molecular absorption is invariably broad-band and
any absorption by hydrogen at 18^9 A would take place over the entire
band width falling on the detector, in this instance 1 A.
39
-------
Absorption of Broad-band Radiation (1 A) by Molecules and by Atoms
Initial signal Io 100 units
$ Absorption by H2 molecules (molecular
absorption) 20$
Bandwidth 1 A
Net signal (I0-mol absorption) 80 units
$ Absorption by atoms 30$
Bandwidth 0.01 A
i.e., J0$ of the 80 units is absorbed over a
bandwidth of 0.01 A, but none of the 80 units
is absorbed from the remaining 0.99 A
.*. Net reduction in broad band signal by atoms is
80 x J0_ x 0.01 = .021)-$
100 1.00
.'. Net signal after absorption by molecules and
atoms = 80 - 0.021*-
i.e. 79.976
It can be seen that even with relatively strong atomic absorption that
the absorption of radiation from the deuterium lamp was to a first
approximation, a fairly accurate measurement of the molecular absorption
at that wavelength. This means that by measuring the background
absorption using the deuterium lamp, molecular absorption could be
measured directly without significant interference by atomic absorption
even at the same wavelength.
In contrast, when the mercury hollow cathode lamp was used, then not
only did the mercury atoms absorb strongly at the resonance line, i.e.,
18^9 A, but also molecular absorption by hydrogen took place.
Under the conditions indicated in the illustration above, the degree of
absorption by the hydrogen molecules would be 20$ of the resonance line
and of the mercury lamps would be J0$ of the remaining radiation.
Initial signal 100 units
Absorption by molecules 20$ unabsorbed signal
(I0 - mol absorption) 80 units
Absorption by atoms 30$
i.e. 80 x 30 units = 2k units
100
UO
-------
Final signal 80 - 2^ 56 units
Total absorption by atoms and
molecules 100 - 56 =
It can be seen, therefore, that the absorption of the resonance line
emitted by the hollow cathode lamp was attenuated by atomic and mole-
cular absorption whereas the absorption of the deuterium lamp was, to
a first approximation, only that of the molecular absorption. By
subtracting absorption of the deuterium lamp from that of the hollow
cathode the atomic absorption at that wavelength could be calculated.
This permitted a direct correction for molecular absorption to be made.
Equipment Used. In order to measure the absorption of the resonance
line in the vacuum ultraviolet at 18^9 A, it was necessary to remove
oxygen from the light path. Oxygen absorbs at wavelengths shorter than
2000 A. Also, it is known that nitrogen begins to absorb at wavelengths
less than 1850 A.
In order to eliminate most of the problem due to air absorption,the
monochromator was purged with Argon. Indications were this would
completely eliminate any absorption by oxygen, which was significant,
and any small absorption by nitrogen.
The Jarrell Ash instrument used was fitted with purging inlet and outlet
taps and Argon was blown through the system for approximately ten minutes
before measurements were taken. This purging resulted in a significant
decrease in absorption due to air and a consequent increase in the
intensity of radiation falling on the detector. This is always an
advantage since under these conditions the amplifier and the photo-
multiplier can be operated at lower currents with a consequent
improvement in the noise level of the signal generated.
It should also be noted that the absorption cells used contain hot
nitrogen from the air and carbon monoxide derived from oxygen in the air.
It was found in practice that the molecular absorption was approximately
15 to 20^ and that we could operate under these conditions.
There were small air gaps in the equipment between (a) the hollow
cathode and the absorption tube and (b) the absorption tube and the
inlet system of the monochromator. In practice it was found that the
degree of absorption by these air gaps was sufficiently small to permit
reasonable signals to reach the detector. No steps were taken to
eliminate these air gaps.
Results. Calibration data was obtained using the slug injection system.
After making corrections for molecular absorption due principally to
water in the atmosphere, the following data were obtained for mercury in
the laboratory atmosphere on the dates cited.
-------
Table 3
Mercury Concentrations in the Laboratory Air (Baton Rouge, La.)
Mercury Concentration
Date (|J.g/ni3)
3-3-72
3-7-72
3-7-72
3-8-72
3-9-72
3-13-72
3-1*4-72
Ik
30
8
15
10
5
1
It should be noted that these mercury levels were quite high,
significantly higher than that which is normally obtained in outside
atmosphere. However, it should be remembered that the "ambient
atmosphere" which was analyzed was that of a laboratory and it is
quite common for the mercury levels in chemical laboratories to be
significantly higher than that found in outside air.
Conclusions. The procedure developed clearly indicated that mercury
could be determined in the atmosphere directly and in real time.
Care must be taken to correct for atmospheric humidity variations
and positive steps must be taken to use the correct type of hollow
cathode in order to avoid self-reversal problems. However, neither
of these complications are difficult to overcome in practice.
The sensitivity (l^ absorption) obtained for mercury was 10"12 g
based on 0.1 n-g/m3 of air and sampling 3° ml- Typical data are
shown in Table k and Table 5.
-------
TABLE k
Analytical Precision of Plug Injection Calibration Technique
Injection
Number*
1
2
3
k
5
6
7
Absorbance
0.18U
0.185
0.175
0.178
0.176
0.162
0.170
Mean = 0.176
Deviation
0.008
0.009
0.001
0.002
0.000
0.011*
0.006
Standard Deviation = 0.009
* 10 cm3 of saturated vapor at 20°C
-------
TABLE 5
Analytical Precision of Continuous Injection Calibration Technique
Injection Rate
1 cm3/min @ 20°C
(13.2(ig/m3)
0.66 cra3/min @ 20.8°C
(9-3 Hg/m3)
0.25 cm3/min @ 22°C
(3.9 Hg/m3)
Absorbance
0.452
0.608
0.487
0.463
0.354
0.367
0.416
0.526
Mean = 0.459
0.410
0.386
0.506
0.438
0.413
0.390
Mean = 0.424
0-392
0.348
0.227
0.363
0-435
0.250
0.256
0.496
Deviation
0.007
0.149
0.028
0.004
0.105
0.008
0.043
0.067
Standard Deviation = 0.076
0.014
0.038
0.102
0.014
0.011
0.034
Standard Deviation = 0.052
0.046
0.002
0.119
0.017
0.089
0.096
0.090
0.150
Mean = 0.346 Standard Deviation = 0.096
44
-------
SECTION VIII
DETERMINATION OF CADMIUM IN THE ATMOSPHERE
The next element chosen for study was cadmium. This element is known
to be highly toxic and is of increasing importance in pollution studies.
It was also known that the concentration levels in the atmosphere were
extremely low, i.e., in the order of .01 to .1 micrograms per cubic
meter of air. The current methods of analysis were to scrub out the
cadmium from the atmosphere and to determine the metal by conventional
colorimetry or flame atomic absorption spectroscopy.
The research approach taken was the conventional approach now developed
by this technique, i.e., the air was drawn into the atomizer and cadmium
absorption at 2288 A was measured. Samples of ambient air produced a
cadmium absorption signal indicating that the procedure was sufficiently
sensitive to measure the cadmium in the air directly by this technique.
EQUIPMENT. The equipment used was similar to that illustrated in
Figure 2. The components were as follows: RF Generator - Lepel Model.
Monochromator - Jarrell-Ash ^ Meter Ebert. Detector and Read-out
System - IP28A photomultiplier tube. Jarrell-Ash broad band AC
amplifier and Beckraan 10" stripchart recorder. Optics - Suprasil
Phano-convex Quartz Lenses, 10 cm focal length, and suprasil -|" diameter
quartz Flat Cell Windows. Optical Pyrometer - Leeds and Northrup
Company. Filament Transformer - General Electric 3^J2^3 HO V input -
6-3 V output. Variable Transformer - Staco Inc. 2PF751 0-8 amps rated.
Air Sampling Network - compressed air tanks, input rotameter, calcium
chloride scrubber, activated charcoal scrubber, millipore filter,
and constant-pressure sampling head.
Platinum Loop Liquid Injector - variable transformer, ammeter, filament
transformer, and 20 gauge platinum wire loop.
Direct Drop Liquid Injector - Hamilton 10 microliter syringe with 3"
needle.
Chemicals and Solutions. Cadmium stock solution (1000 p-g/ml) was
prepared by dissolving 1.000 g Cd metal (A. D. Mackay and Co.) in the
minimum volume of concentrated hydrochloric acid and diluting to
1000 ml with 1$ (V/v) HC1. More dilute solutions were made up daily
by dilution of 1000 (ig/ml stock solution with distilled water. All
solutions were stored in polyethylene bottles and vials. No
solutions were stored overnight in order to avoid plating out
problems.
-------
Calibration Techniques. Several calibration techniques were investi-
gated in order to correlate the absorption signal from cadmium in the
atmosphere and the concentration of cadmium in the air. Although
cadmium itself is relatively volatile compared to most metals, its
salts are not particularly volatile. In addition the metal is quite
chemically reactive at elevated temperatures. This precluded the
method of evaporating cadmium or cadmium salts into the air stream as
the basis for a calibration technique.
The successful techniques developed were (a) platinum wire technique
and (b) the liquid drop technique.
Platinum Wire Technique. A platinum wire was built so that it could be
inserted into the air stream immediately before reaching the furnace.
This procedure is described earlier under the section on lead. (See
Figure 8 and 9.)
Several cadmium solutions were made up, each with a known concentration
of cadmium. A fixed quantity of the cadmium solution was placed onto
the platinum wire using a standard capillary. In this instance, the
volume used was 5 |o,l. The platinum wire with the 5 M1! °f solution was
lowered into the atomizer cell, with the air stream gently flowing at
a predetermined rate over the wire. The platinum wire was heated by a
separate electrical circuit to a bright white heat, i.e., about 2000°C.
The cadmium solution rapidly evaporated and the cadmium atoms were
swept through the furnace into the absorption cell. By using different
concentrations of cadmium, the quantity of cadmium introduced into the
cell was varied. The absorption signal for each cadmium concentration
was measured and a calibration curve obtained. A typical calibration
curve is illustrated in Figure 16.
Calibration Using Liquid Drop Technique. In this system a drop of
solution 2 u-1 in volume was dropped directly into the atomizing bed.
Here the drop evaporated rapidly. The liberated cadmium atoms were
swept into the absorption tube and the absorption measurement made in
the normal way.
In the preliminary work the difficulty involved with this technique was
that the volume of solution introduced into the bed was fairly difficult
to reproduce. The drop would sometimes fall off the tip of the syringe
leaving a small part behind, or some evaporation of the drop took place
leaving a small residue of cadmium metal on the tip of the needle. In
either case, a low result would be obtained. Unfortunately, the
residual metal would then cause a high result with the next drop, the
data obtained in this fashion being very inprecise.
The problem was solved by using a Drummond micro pipet modified for full
displacement. In this type of pipet liquid in the bore of the syringe
was completely displaced by the needle pushing all of the liquid out of
the tip by air displacement and forcing it to be ejected into the
-------
0.4
0.
s O.2
O.I
-O'
. D
O - Direct drop injection
O - Platinum loop injector
3
/O
b
8 9 /O
Figure 16. Calibration Curves of Cd (Pt Loop) and Liquid Drop
-------
atomizer. There was an significant increase in the precision of the
procedure and sample volumes could be decreased quite safely from 5 M-l
to 2 u.1 and still get precise data. It was much easier to handle this
volume of sample since when it hit the hot atomizer bed invariably the
evaporation took place at explosive rates and 5 M-l generated a signifi-
cantly greater volume than 2 ul. In practice it was easier to control
the explosion of a 2 p.1 than a 5 ^1 sample.
By making up standard solutions of cadmium of varying concentrations the
relationship between cadmium absorption and cadmium concentrations could
be measured. The resulting calibration curve is shown in Figure 16. It
can be seen that the slope of the calibration curve was significantly
steeper than that obtained when the platinum loop was used. However, if
instead of the maximum absorption the peak area was used as a measure of
the absorption signal, the two curves coincided much more closely. This
indicated that in the case of the platinum wire the cadmium evaporated
off at a speed relatively slow compared to when the drop was ejected
directly onto the hot carbon bed.
There were advantages and disadvantages to both techniques. When high
sensitivity was required, it was better to use the liquid drop tech-
nique, since the signal was sharp and could be fairly easily measured.
However, if high precision was necessary, it was better to use the
platinum wire and measure the area of the curve. The problems involved
were typically those of the gas chromogram where the peak height or peak
area can be used successfully for quantitative analysis. There was one
added difficulty with atomic absorption and that was that the relation-
ship between absorption and concentration was logarithmic, not linear.
A correction must be made for the logarithmic relationship of the area
in order to get precise results.
The Use of Carbon Discs. Some success has been achieved by the use of
carbon rods for analysis of various metals by atomic absorption
spectroscopy. In this technique the solution was put onto the carbon
rod when it was cold. The rod was then warmed up and the solvent
evaporated off and discarded. The rod was then heated to high tempera-
ture and atomization took place. The temperature control of this
procedure must be rigidly adhered to in order to get reproducible
results. It has always been a suspicion that some of the volatile
metals could be lost when the solvent was evaporated off.
Cadmium is a typical example of a volatile metal which might be lost
during this evaporation stage. In order to check this information the
cadmium solution was spotted onto a carbon disc. The disc was lowered
into the atomization chamber and the solvent evaporated off into the
flowing air stream. The air stream in turn drove the solvent through
the atomizer into the absorption chamber.
In the atomization chamber all organic materials were burned to carbon
monoxide and hydrogen and any water was also converted to carbon monoxide
and hydrogen. It was possible to correct for any molecular absorption
-------
due to hydrogen using the deuterium lamp as described in the mercury
section. However, in practice it was found that hydrogen did not
absorb significantly at the resonance wavelenght of cadmium, therefore,
no correction was necessary for molecular absorption.
A cadmium solution was injected onto the carbon disc, which was then
lowered to within two inches of the hot carbon bed. Here the tempera-
ture of the disc was approximately 200°C, a temperature comparable to
that used in the carbon rod drying step.6'7 The solvent evaporated off
within a relatively short period of time. When the carbon disc was
completely dry, it \»as Cropped onto the atomizing system ana the
absorption due to cadmium metal was recorded. A typical spectrum thus
obtained is shown in Figure 1J. As can be seen cadmium was indeed lost
during the drying stage and the bulk was found during the atomization
stage.
This indicated a direct error may be involved in the carbon bed method.
However, in the carbon bed technique which we used no drying stage was
necessary and no error is involved.
Results. Based on this technique and the calibration curves obtained,
typical results for the analyses of laboratory atmosphere in Baton Rouge
are shown in Table 6.
Table 6
Typical Cd Concentrations in the Laboratory Atmosphere
at Baton Rouge, La,
Concentration
Sample Date Absorbance (jj,g/m3)
Aug. 2k, 1972* 0.027+0.005 0.020+0.001).
Aug. 25, 1972** 0.0^+0.008 0.035+0.005
* Immediately after heavy rain
** Weather hot, clear, sunny
The sensitivity of the procedure was determined to be 10"13 grams of
cadmium using the liquid drop technique.
-------
90 ^
SO-
70
60
50
40
30
20
10\
CARBON
SLIVER
B
V'
DIRECT
DROP
Comparison of Carbon Sliver
and Direct Drop Injections
A) Drying at 6 cm above
carbon bed
B) Injection of dried sliver
onto carbon bed
Figure IT- Absorption Traces Cd Introduced on Carbon
50
-------
Conclusions. It was felt that this technique was successful in pro-
viding a method for the direct determination of cadmium in the atmo-
sphere in real time. There appeared to be no complications in the
technique and it was felt that the ease in the determination of
cadmium indicated the wide applicability of this method to metals in
air analysis.
-------
SECTION IX
DETERMINATION OF ARSENIC
Arsenic was chosen as the next element to study for several reasons.
First, it is known to be highly toxic and a dangerous air pollutant.
Secondly, the standard techniques for the estimation of arsenic involve
scrubbing large volumes of air and analyzing the trapped arsenic by
colorimetric methods of analysis or flame atomic absorption spectros-
copy.
Arsenic has proven to be one of the more difficult elements to determine
by any technique. There are virtually no colorimetric methods which
are highly sensitive and with a sufficient degree of selectivity to
give reliable quantitative data. Many techniques used are based on the
reduction of arsenic to arsine AsH3, which evolves from the system.
The separated arsine is then determined either by atomic absorption
spectroscopy or colorimetric analysis. In either case, the end point
is difficult to control and the results are always suspect.
For this reason it was decided that arsenic would be a challenging
element to study. The development of a reliable rapid method of
analysis would be highly desirable.
Detection of Arsenic in the Atmosphere. The system was set up in
the normal method using an arsenic hollow cathode and the monochromator
was set at 18197 A, the most sensitive absorption resonance line of
arsenic. Air was drawn through the atomizer and the absorption signal
measured as previously mentioned. Typical results are shown in
Figure 18.
The results were very encouraging and indicated that arsenic in
the ambient atmosphere could be detected by this system. It now
remained to calibrate the method in order to translate the
absorption signal to a concentration of arsenic in the air.
Methods of Calibration. The Direct Drop Technique. A solution of
arsenic was made up to contain a 1000 parts per million of arsenic.
From this stock solution a number of standard solutions were prepared
with concentrations varying from one part per million to a hundred
parts per million.
Samples varying from 2 |il to 5 M-l were measured accurately using a
syringe and dropped directly into the atomizer. Results showed that
the absorption signal obtained was highly unreproducible. Even though
an absorption signal could be obtained for solutions containing one part
-------
.05
.03
UJ
o
ce
o
.01
O 184.9 nm
O 193.7 nm
A 197.2 nm
BED DEPTH - 4 in.
SOURCE - AS DHC
SAMPLE - AMBIENT AIR
200
400
FLOW RATE, crc
600
Figure 18. Absorption by as in the ambient air using three different resonance lines.
-------
per million of arsenic the signal obtained from the solutions containing
a hundred parts per million of arsenic was greater but did not conform
to Beer's Law.
Frequently, there was an absorption signal immediately after injection
of the arsenic followed by a second absorption signal some period later.
This is illustrated in Figure 19. The delay between the first signal
and the second signal was not reproducible. Sometimes it was one or
two minutes and sometimes it was up to ten minutes. The data seem to
indicate that the arsenic was held up somehow on the carbon bed.
A search of the literature indicated that no stable arsenic carbon
compounds had ever been isolated. It was, therefore, concluded that if
there was a reaction between the arsenic and the carbon the nature of
the product was not known. Similar results were obtained when the
platinum wire technique was used.
The Use of Arsine in Calibration Procedures. One possible problem area
was that the solutions used were in some way losing the arsenic on the
sides of the vessels or that the metal was plating out on the walls of
the syringe. Precautions were taken to eliminate this problem by
making up calibration solutions freshly (within hours) before all
studies were made.
As an extra precaution we investigated the use of arsine as a reagent
for calibrating the procedure. The first attempt to do this was made
by generating arsine from an acid solution. This proved to be a very
inefficient and unreproducible technique. Ihe technique could perhaps
have been standarized to provide a reproducible amount of arsine, but
considerable time would have been necessary to provide such a procedure.
It was, therefore, decided to buy a small cylinder of arsine gas from a
commercial source.
It is known that arsine is very sensitive to light and water vapor,
particularly if the two are in combination. Therefore, all precautions
were taken to prevent light from reaching the arsine and to insure that
the compound was protected from water vapor in the ambient atmosphere.
The first approach taken was to flush arsine into a bottle fitted with a
rubber cap. A glass syringe was pushed through the rubber cap and a
known volume of arsine was extracted. This arsine vapor was then
injected directly into the atomizer bed and the absorption signal
measured. As before, the signal was very unreproducible.
Atomic Form of Arsenic. A literature survey indicated that arsenic in
the gaseous form existed primarily as As4 at room temperature. This
conclusion was based on the experimental observation that the vapor
pressure of arsenic in a gaseous state coincides with a molecular weight
of As4. As the temperature was increased the concentration of As
increased, and the concentration of As2 and As4 progressively decreased.
-------
en
In
A - DIRECT DROP
AS IN HCOOH
14
TIME, minutes
Figure 19. Absorption trace of as indicating delayed signal.
-------
At a temperature of about l600°C and under equilibrium conditions, the
vapor was estimated to consist of As4, 55$> As2, ^5$> As, 10^. It is
important to note that neither As4 or As2 would be expected to absorb
strongly at the resonance lines of elemental As. Under these circum-
stances the degree of absorption of freely liberated As would depend
very much on the equilibrium conditions under which the arsenic was
reduced to free metal. This would be an important variable in flame
atomizers. Also it was anticipated that any delay in taking the atomic
absorption measurement (as in our instrument) would result in a rapid
loss of free As from the system and a rapid decrease in analytical
sensitivity. This proposal was tested experimentally on the equipment
which was designed in this work. Indications were the arsenic was
reduced in the atomizer section. Since the absorption is measured in
the absorption tube there was evidently a delay between formation of
free arsenic and measurement of the atomic absorption signal.
Two attempts were made to verify this affect. The first was to increase
the flow rate and the second was to decrease the carbon bed size. It
was proposed that if the flow rate were increased, then more free atoms
would be forced into the absorption tube before they could recombine to
form As2 and As4. The second proposal was if the bed was decreased in
size, then again more atomic As atoms would reach the absorption tube
before recombining to As2 and As4.
The results of the experiment are illustrated in Figures 20 and 21.
The data clearly indicated that as flow rate was increased, the
absorption signal did increase, indicating more free arsenic atoms
reached the absorption bed. Also, a decrease in bed size resulted in an
increase in absorption signal, again, supporting the thesis proposed
that the equilibrium between As2 and As4 very drastically affected the
absorption signal.
One of the conclusions drawn from this experimental work was that if the
bed temperature was greatly increased (e.g., to 2500°C), the equilibrium
would be thrown over very drastically to cause an increase in As concen-
tration. Further, if the absorption tube was kept at a very much
increased temperature, e.g., 2500°C, the percentage of As would remain
high and the absorption signal would be increased accordingly. It was
determined, therefore, that increasing the temperature of the entire
system would be very advantageous in the case of elemental arsenic
determination.
At this point studies on arsenic were discontinued because we had
already embarked on making a new instrument to operate at a much higher
atomization and absorption temperature. It was felt that it would be
more fruitful to devote our efforts to building this instrument and
evaluate arsenic at a later date.
-------
UJ
O
CQ
ce
o
CO
CO
SOURCE - AS DHC
X - 235.0 nm
FLOW RATE - 300 cm3/min
SAMPLE - 5 jjg AS IN 5 ul
.10
.05 —
BED DEPTH, inches
Figure 20. Effect of bed depth on arsenic absorption.
-------
en
oo
CO
o:
o
SOURCE - AS DHC
A-235.0 nm
BED DEPTH -1 in.
SAMPLE - 5 jig AS IN 5 ul
200
FLOW RATE, cm3/min
Figure 21. Effect of flow rate on arsenic absorption.
-------
SECTION X
OTHER ELEMENTS STUDIED
Preliminary studies on a number of other elements were carried out.
These studies were primarily to get an indication of analytical
sensitivity and of problems involved with these elements. The results
are outlined below.
Selenium. Selenium is an element similar in many respects to arsenic,
both in some of its biological effects and in its chemical properties.
It is known to form Se2 and Se4 at room temperature, in much the same
way as arsenic does.
Preliminary studies showed that selenium was similar to arsenic in its
behavior in this instrument. On ocassions, sensitive atomic absorption
data were obtained and other times it was difficult to detect the
selenium even under fairly high concentrations. In light of our
experience with arsenic, it was decided not to pursue our studies with
selenium at this point but to delay further studies until the high
temperature instrument was completed.
Copper and Zinc Absorption signals from copper and zinc in the
laboratory atmosphere were detected by drawing air through the
instrument as described previously.
Typical absorption signals are shown in Figure 6.
The wavelengths used for these measurements were copper, J2^8 A and
zinc, 2138 A.
Attempts to produce calibration curves in order to relate the absorption
signals to the quantity of zinc or copper in the atmosphere were
unsuccessful. In the great majority of cases, the background signal
obtained when the instrument was turned on was 100 percent absorption
for each of these elements. This indicated that under the conditions
of operation, high concentrations of copper and zinc were reaching the
absorption tube. These impurities were probably coming from the carbon
rods used to make up the reduction bed. Collaborative studies with both
Pocographite and Ultracarbon (pure carbon manufacturers) were carried
out. In an effort to eliminate all copper and zinc from the carbon used,
each of these companies took special precautions to eliminate all
sources of these elements from their manufacturing system. This, of
course, included all brass items, copper items or galvanized items.
In addition, each company independently treated the resultant graphite
by extended heating at high temperatures (such as 2000°c) for 2k and
for kQ hours.
In each case there was little improvement in the background signal
obtained for either of these elements. In practice, it would be vital
59
-------
to be able to clean up the instrument to remove such background.
Otherwise, absorption signals from known air samples could not be
measured.
The perplexing thing about this study was that occasionally a batch of
carbon rods would be obtained which generated very little background
and which cleaned up rather quickly when put into the atomizer section
and heated up. We were unable to trace the reason for this inconsis-
tency in the manufacturing process and discontinued studies at this
time on these two elements.
Another source of possible contamination was the quartz used for making
the atomizing system. It is planned to use various types of quartz from
different manufacturers to see if this eliminates the problem. It is
also anticipated that when we go to a high temperature system, the
copper and zinc contaminants will be greatly reduced or eliminated from
the system because of the elevated cleansing temperature available.
In a most important previous observation both copper and zinc were
detected and measured directly In the atmosphere with reasonanble
backgrounds. This illustrated the feasibility of using this instrument
for monitoring copper or zinc in the air.
Sodium. Preliminary studies indicated that an absorption signal can be
obtained from sodium in the laboratory atmosphere. This illustrated
the capability of the instrument for measuring this element directly.
In a similar manner to copper and zinc, a very high background was
frequently found when studies on sodium were carried out. It was quite
possible that this background came from the quartz used since cadmium
is generally at least a minor component in this type of material. It
was decided that the high temperature system may solve many of these
problems.
Similar observations were made for potassium, i.e., it could be detected
in the atmosphere by direct measurements but calibration was difficult
because of the high background of potassium.
Iron. This element has a considerably higher boiling point than many of
the elements examined hitherto. It was observed that the background
signal could be decreased to a reasonable level by heating the bed up to
operating temperature for an hour to ninety minutes, during which time
purified air was continually drawn through the instrument.
Attempts made to prepare a calibration curve showed that the absorption
signal was very erratic for a constant concentration of iron. It was
shown that the absorption signal x^as very sensitive to small changes in
temperature of the atomization bed. It was concluded that operating at
temperatures around 1400°C in the atomizing bed was too low.
Presumably, the temperature is at the critical low end of the permis-
sible temperature for operation of the atomization system.
60
-------
It was concluded that rather than try to optimize conditions and
control conditions rigidly, it would be better to operate at a high
temperature system wherein the temperature of the bed would not be
critical and reproducible analytical data could be obtained, even
with variations in bed temperature. It was concluded, therefore,
that the study of iron would be more profitable when the high temper-
ature system was built.
We were unable to detect Be, Mg, Ca_ or Mg in the air, or when a
solution containing 1 (J,g of the element was introduced into the
system. Each of these elements forms refractory oxides which may
not be reduced at these temperatures. A higher temperature system
was indicated.
61
-------
SECTION XI
DISCUSSION
Studies Involving the Use of High Temperature Atomizers and Absorption
Systems. Based on the results obtained from this project, it is
clear that certain elements present as air pollutants can be detected
directly without prior separation or concentration from large volumes
of air. Calibration procedures have been developed which enable us to
calculate the pollution levels from the measured absorption signals.
It is, therefore-, possible to measure metal pollution levels rapidly.
and in real time.
The three very important elements lead, mercury and cadmium have .all
been quantitatively determined in the laboratory atmosphere at
Baton Rouge. Other elements such as copper, zinc, arsenic, selenium,
sodium, potassium, and silver have been detected in the air but
quantitative analyses have not yet been developed for these techniques.
Preliminary experiments on less volatile elements such as iron, cobalt,
etc., indicate that the temperature of the atomizer is too low for
practical quantitative work. The degree of atomization is very
sensitive to the temperature of the atomizer and at the temperatures
used, i.e., 1350-1500°C, relatively small changes in temperature caused
significant changes in the absorption signals obtained.
It appears that the best way to overcome this problem is to increase
the atomization chamber temperature significantly, i.e., to an order
of 2JOO°C.
It is our intention at this time to design such an instrument and test
it for the less volatile metals. In addition, we will continue to
study calibration techniques and miniaturization of the present
equipment. It is hoped that based on these results the field instrument
capable of being moved easily from one location to another will be
developed.
62
-------
SECTION XII
REFERENCES
1. Robinson, J. W. , Atomic Absorption Spectroscopy, Marcel Dekker,
Inc. , New York, p 62 (1966).
2. L'vov, B. V., Atomic Absorption Spectrochemical Analysis, A.
Hilger, London (1970).
3. Altshuller, A. P., and Cohen, I. R., Anal. Chim., 32, p 80J (I960).
k. Scott, W. W., Standard Methods of Chemical Analysis, 6th Ed, V 1
(1966).
5. Koistyohann, S.R., and Pickett, E. E., Anal. Chim., 37, p 661
(1965).
6. Anderson, R. G., Johnson, H. N., and West, T. S., Anal. Chim.
Acta. 57, p 281 (1971).
7. Amos, M. D., Bennett, P. A., Grodie, K. G., Jung, P. W. Y., and
Matousek, J. P., Anal. Chim., ij-3. P 211 (1971).
-------
SECTION XIII
LIST OF PUBLICATIONS
1. "The Continuous Determination of Lead in Air," H. P. Loftin,
C. M. Christian and J. W. Robinson, Spectroscopy Letters.
161-1?^ (1970).
2. "The Direct Determination of Cadmium and Mercury in the Atmosphere,"
C. M. Christian, II and J. W. Robinson, Anal. Chim. Acta, 56(3), k66
(197D.
3. "Non-Flame Atomic Absorption in the Yacu«m UV. The Direct Determi-
nation of Mercury in Air Using the 18^9 A Resonance Line," J. W.
Robinson, P. J. Slevin, G. D. Hindman, and D. K. Wolcott, Anal. Chim.
Acta, 61, 11-31 (1972).
k. "Metallic Impurities in Spectroscopic Carbon Electrodes," J. W.
Robinson and G. D. Hindman, Spectroscopy Letters, j?, 169-178 (1972).
5. "Study of the Application of Atomic Fluorescence Spectroscopy to
the Direct Determination of Mercury and Cadmium in the Atmosphere,"
J. W. Robinson and Y. E. Araktingi, Anal. Chim. Acta, 63, 29 (1973).
6. "The Determination of Cadmium by Atomic Absorption in Air, Water,
Seawater or Urine Using the RF-Carbon Bed Atomizer," J. W. Robinson,
D. K. Wolcott, P. J. Slevin, and G. D. Hindman, Anal. Chim. Acta, in
press.
7- "Correction for Background Absorption in Atomic Absorption
Spectroscopy Utilizing Carbon Atomizers," J. W. Robinson, G. D. Hindman,
and P. J. Slevin, Anal. Chim. Acta. in press.
8. "Calibration Techniques Used in Direct Determination of Atmospheric
Metallic Pollutant Levels," J. W. Robinson and D. K. Wolcott, Anal.
Chim. Acta, in press.
Related Publications
1. "Recent Advances in Instrumentation in Atomic Absorption," J. W.
Robinson and P. J. Slevin, American Laboratory. 10-18 (August 1972).
2. "Spanish Moss as an Indicator of Lead in the Atmosphere Before the
Use of Leaded Gasoline," J. W. Robinson, C. M. Christian, J. D.
Martinez, and M. Nathany, Environmental Letters, k, 87~93 (1973)-
-------
SECTION XIV
GLOSSARY OF TERMS, ABBREVIATIONS AND SYMBOLS
g - gram
cc - cubic centimeter
K - absorption coefficient at frequency v
e - charge on an electron
m - mass of the electron
c - speed of light
f - oscillator strength of the energy transition involved in the
absorption
N - number of unexcited or ground state atoms in the light path
rf - radiofrequency
A - Angstrom
mm - millimeter
W - watt
V - volt
°C - degrees Celcius
T.E.L. - tetraethyl lead
i.e. - id est (that is)
p.1 - microliter
Pt - platinum
$ - percent
jig - micrograrn
|j,g/m3 - microgram per cubic meter
AA - atomic absorption
a - standard deviation
e.g. - exempli gratia (for example)
65
-------
Sensitivity. Where sensitivity limits have been expressed they may be
considered as that amount of sample required to give one percent
absorption. This is sometimes referred to as absolute sensitivity. A
concentrational sensitivity is sometimes used in atomic absorption
literature which is defined as the concentration of the sample solution
required to give one percent absorption.
66
-------
BIBLIOGRAPHIC DATA
SHEET
1. Report No.
EPA-650/2-73-011
3. Recipient's Accession No. j
"4." Title "and Subtulc
DIRECT DETERMINATION OF METALS IN AIR
5- Report Date
August 1973. Date of Prjp
6.
7. Author(s)
J. W. Robinson
8. Performing Organization Kept.
No.
9. Performing Organization Name and Address
Louisiana State University
Baton Rouge, Louisiana 70803
10. Project/Task/Work Unit No.
26AEK-12 1AA010
11. Contract/Grant No.
800866
12. Sponsoring Organization Name and Address
National Environmental Research Center
Chemistry and Physics Laboratory
Environmental Protection Agency
Research Triangle Park. N. C. 27711
13. Type of Report & Period
Covered
Final A/1/70-5/31/73
14.
15. Supplementary Notes
16. Abstracts An instrument has been developed capable of the direct determination of
metals in air. No prior scrubbing or extracting of the metals from the air is
necessary. Consequently, the time necessary for analysis is a matter of minutes -
permitting real time analysis to be carried out on small volumes of air.
The method was based on atomic absorption spectroscopy and involved the develop-
ment of a highly efficient atomizer. The sensitivity of the method was determined to bje
about 10"1 g. Procedures for the direct quantitative determination of lead, mercury
or cadmium in the air were developed. Calibration techniques were studied and
reliable analytical techniques were developed.
17. Key Words and Document Analysis. 17o. ! U-scriplots
Atomizer
Atomic Absorption Spectroscopy
Trace Elemental Analysis
Metals
Air Analysis
Lead
Mercury
Cadmium
17b. IdcntiHers/Opcn-Ended Terms
17c. COSATI F.eld/Group
18. Availability Statement
Release unlimited
FORM NTIS-35 [REV. 3-72)
19. Security Class (This
Report)
UNCLASSIFIED
20. Security Class (This
Page
UNCLASSIFIED
21. No. of Pages
73
22. Price
USCOMM-OC 14052-P72
67
-------
INSTRUCTIONS FOR COMPLETING FORM NTIS-35 (10-70) (Bibliographic Data Sheet based on COSATI
Guidelines to Format Standards for Scientific and Technical Reports Prepared by or for the Federal Government,
PB-180 600).
1. Report Number. Each individually bound repon shall carry a unique alphanumeric designation selected by the performing
organization or provided by the sponsoring organization. Use uppercase letters and Arabic numerals only. Examples
FASF.B-NS-87 and FAA-RD-68-09.
2. Leave blank.
3- Recipient's Accession Number. Reserved for use by each rcp>orc recipient.
4- Title and Subtitle. Title should indic.tie clearly and briefly the subject coverage of the report, and be displayed promi-
nently. Set subtitle, if used, in smaller type or otherwise subordinate it to main title. When a report is prepared in more
than one volume, repeat the primary title-, .uM volume number .ind include subtitle fnr (hi- specific volume.
5- Report Dote. I'nch report shall carry j daie indicating at least month and year. Indicate the basis on which it was selected
{c.^.t date of issue, date of approval, d.ne of preparation.
6- Performing Orgomzotion Code. Leave blank.
7. Autrjor{s). Give name(s) in conventional order (e.g., John K. Doe, or J.Robert Doe). List author's affiliation if it differs
from the performing organisation.
8. Performing Orgoni zotion Report Number, insert if performing organization wishes to assign this number.
9. Performing Orgontzotion Name ond Address. Give name, street, city, state, and xtp rthte. List no more than tv/o levels of
an otg;ini?.nt ional hierarchy. Display thr name of the organization exactly as it should appear in Government indexes such
as UMJRDrM.
10. P ro jeci/Tosk/ Work Unit Number, Usr the project, task a nii work unii numbers under winch the report was prepared.
11. Controcf ^Grant Number. Insert conir.ut *M grant number under which report was prepared.
12. Sponsoring Agency Nome end Address. InfluJc zip rocir.
13- Type of Report ond Period Covered. InJ K .ite interim, f inn I, i- tc ., ;mj , if applicable, d.itcs covered.
14- Sponsoring Agency Code. Leave M.tnlc.
15. Suppiemcntory Notes. Hnu-r in/tumatitin not iiu luded i-lst-whert but useful, such a> : Prepared tn cooperation with . . .
Tunslatiun of ... Presented at conference of . . . To In- published in ... Super sciit-s . . . Supplements
16. Abstract. Include ;i brief (200 words or li-ss) factual .summ.uy of the most si^niftcjnt information contained ir» the report.
If the report contains a significant bibliography or literature survey, mention It here.
17. Key Words and Document Analysts, (o). Descriptors. Select from the Thesaurus of Engineering and Scientific Terms the
proper authorized terms that identify the rnajoi concept of the research and are sufficiently specific apt,i precise to be- used
as index entries for c.italo>:inp.
(b). Identifiers ond Open-Ended Terms. Use idcntifit-rs (or project names, code name--, equipment designators, etc. Use
open-ended terms written in descriptor form for those subjects for whjch no descriptor exists.
(c). COSATI Field/Group. Field and Group Assignments art-to be taken from the 1965 COSATI Subject Category List.
Since the majority of documents an- mult id isc iplinary in nature, the primary Field/Group assi£f»mcm(s) will be the specific
discipline, area of humun endeavor, or type of physical object. The appltcat ton(s) will be cross-referenced with secondary
Field/Group assignments that will follow the primary posting(s).
18- Distribution Statement. Denote relcasability to the public or limitation for reasons other than security for example "Re-
lease unlimited". Cue any av.u I ability i»> the public, with address nnci price.
19 & 20. Security Clossification. Ho not .submit classified reports to the National Technical
2L Number of Pages. Insert the cotal number of pages, including this one and unnumbered paj;cs, but excluding distribution
list, if any.
22. Price. Insert the price set by the National Technical Information Service or the Government Printing Office, if known.
FORM NT1S-3S (REV. 3-721 —— . . • (JSCOMM-OC 14032-P72
-------
ENVIRONMENTAL PROTECTION AGENCY
Technical Publications Branch
Off tC* of Administration
Triangle Park, N. C. 27711
POSTAGE AND FEES PAID
ENVIRONMENTAL PROTECTION AGENCY
EPA - 335
OFFICIAL BUSINESS
AN EQUAL OPPORTUNITY EMPLOYER
SPECIAL FOURTH-CLASS
BOOK
Return this sheet if you do NOT wish to receive this material
or if change of address is needed CD. (Indicate change, including
ZIP code.)
------- |