PRINCIPLES AND APPLICATIONS OF ATOMIC

ABSO R PTI0N SPECTROSCOPY

Alfred ZeHner
Clinical Laboratories. Yale-New Haven Medical Center.
New Haven. Connecticut

Page
1. Introduction .................................................... 1
2. Historical ...................................................... 2
3. Theoretical ..................................................... 3
3.1. Atomic Absorption and Beer’s Law ........................... 4
3.2. Atomic Activation and Excitation ............................ 5
3.3. Flames ................................................... 6
4 . Instrumentation and Techniques ................................... 8
4.1. Commercial Instruments ..................................... 8
4.2. Basic Principles of the Atomic Absorption Spectrophotometer ..... 9
4.3. The Hollow Cathode Tube ................................... 9
4.4. The Burner-Atomizer ....................................... 12
4.5. Other Means of Sample Vaporization ......................... 14
4.6. Wavelength Selection ....................................... 16
4.7. Absorption Measurement and Recording ....................... 17
5 . Methodology ................................................... 19
5.1. Sensitivity ................................................. 22
5.2. Interferences ............................................... 26
5.3. Interference Control ........................................ 33
5.4. Individual Elements ........................................ 36
References ......................................................... 55

I . Introduction
One decade ago the term atomic absorption spectroscopy was familiar
to only a small group of scientists. to a few chemists working on special
problems. and to astrophysicists. Within the last few years. however.
this analytical technique has found enthusiastic acceptance by science
and industry. Several hundred papers have been published in a short
time on basic research. instrumentation. development of methods. and
practical application of atomic absorption spectroscopy. Several good
reviews of the subject stressing various aspects have appeared (A13. D6.
E2. G6. L3. L8. M6. M7. R2. R5. W15). and it seems hardly necessary
to add to their number . The problems of medical laboratories and clinical
1
2 ALFRED ZETTNER

chemists are, however, unique in many respects and it was for this rea-
son that the present article was written.
When new analytical tools become available, more often than not
considerations of responsibility to the patient, practicality, and economy
will keep the clinical chemist from accepting such newly developed
techniques without careful deliberation. It appears that presently atomic
absorption spectroscopy is slowly finding entrance into medical research
and service laboratories, and there is reason to expect that this technique
will find wider use and greater application than emission flame spectros-
copy. Virtually all metals, with very few exceptions, can be determined
by atomic absorption spectroscopy, It is anticipated that this technique
not only will replace currently used analytical methods for metals, but
will also make feasible the routine determination of elements now im-
practical by conventional means. Furthermore, the operational stability
of available instruments and the simplicity of actual performance of
measurements make this technique well suited for automation, by addi-
tion of an automatic sample feed and automatic recording.
Since atomic absorption spectroscopy is based on an atomic property
unique for each element, the possibility exists to obtain “true” values,
a challenge not to be overlooked by workers attempting to develop
atomic absorption methods for a particular metal or material. Some of
the original theoretical expectations in regard to freedom of interferences
have not been fully substantiated in practice and almost every element
has been found to suffer from one type of interference or another.
It is the object of the writer to present in this review the theoretical
and instrumental basis of this analytical technique, discuss the practical
methods already developed and in use, and consider some of the poten-
tialities for the clinical laboratory.

2. Hidorical
The first to record a phenomenon caused by atomic absorption was
Wollaston in 1802 in England, who observed dark bands in the sun’s
spectrum, although he was unaware of their cause. The same observation
was made in 1814 by Fraunhofer, who, with improved instrumentation,
was able to map more than 500 absorption lines in the solar radiation.
Not until the days of Kirchhoff and Bunsen, however, was the true
nature of the Fraunhofer lines recognized. In a series of papers Kirch-
hoff and Bunsen laid the foundation for modem spectrochemical analysis
(KSKlO).
A typical atomic absorption experiment in the laboratory was carried
out by Foucault in 1849 in Paris, as related by Stokes (S9). Foucault
 ATOMIC ABSORPTION SPECTROSCOPY 3

passed a ray of sunlight through a carbon arc and then through a prism
to produce a spectrum. In the spectrum he found the expected dark lines,
but they were more completely blacked out than in sunlight without
traversing the carbon arc. The mysterious power of the carbon arc to
enhance these spectral characteristics of sunlight was simply due, of
course, to sodium atoms in the arc deriving from sodium contamination
of the carbon electrodes, causing absorption on the resonance lines of
sodium, the well-known D lines in the sun’s radiation.
Following the work of Lundegardh in the twenties, emission flame
spectroscopy became established as an analytical tool in almost every
branch of science. Although hollow cathode tubes were first studied by
Paschen (P2) in 1916, and although atomic absorption spectroscopy had
found occasional application, notably in the mercury vapor detector
(W20), it remained for Walsh (W2) in Australia in 1955 to recognize
the essential advantages inherent in absorption over emission methods
and revive general interest in this technique. Shortly thereafter but
apparently independently, Alkemade and Milatz ( A2, A3 ) in Holland
devised instruments and applied atomic absorption spectroscopy in their
laboratory. Walsh and his co-workers have since contributed a re-
markable volume of work on instrumentation and application, and patents
are held by Walsh on his method in Australia, Europe, and America.
It may be mentioned here that this decade probably will see the soft
lunar landing of an atomic absorption spectroscope (M8)intended to
analyze surface material of the moon and relay signals back to earth.
The system makes use of a solar furnace for sample vaporization and
depends on the sun’s spectrum as a light source.
3. Theoretical
Atomic absorption can simply be defined as the absorption of light by
activated atoms. Such absorption occurs on very narrow spectral lines,
the so-called absorption or resonance lines, their theoretical spectral
width being of the order of 0.001 A. The lines are entirely characteristic
and specific for each element and to date no two elements have been
found to possess an identical resonance line. If monochromatic light of a
specific wavelength is provided, it will be absorbed only by atoms of
that element whose resonance line is identical with the wavelength of
the light source and not by any others. A field of atoms is “opaque” for
monochromatic light when resonance line and source wavelength match,
but for other wavelengths it is translucent. The degree of opacity is
proportional to the total number of absorbing atoms. It follows then
that with a beam of specific monochromatic light the concentration of
an element can be determined in a mixture of atomic species.
4 ALFRED ZETTNER

Atomic absorption will take place only in a field of free, neutral,

activated atoms. Atomic absorption cannot be brought about by ions,
by atoms bound in compounds, or by a molecular gas. When metals are
heated to their boiling point, they vaporize as free atoms, provided that
interaction with other elements is prevented, and it is for this reason
that atomic absorption spectroscopy in its present form has found its
most extensive applications in the analysis of the metallic elements.
3.1. ATOMICABSORPTION AND BEER’SLAW

As stated above the degree of atomic absorption depends on the num-

ber of absorbing atoms. If the absorbing field is spatially defined, absorp-
tion is proportional to concentration of the absorbing species. But this,
in essence, is also the content of Beer’s Law. The latter holds true under
certain ideal conditions (L11) :
(1) The analyzing light should be perfectly monochromatic.
( 2 ) The light beam should be strictly parallel.
( 3 ) The light beam should travel in an optically homogeneous
medium that does not scatter radiation.
( 4 ) The absorbing units (molecules or atoms) should not be close
enough to one another or to other molecules that their structure or their
energy levels are affected.
When these points are critically examined, it appears that these con-
ditions theoretically are ideally realized in atomic absorption. First, the
light sources used (hoIlow cathode lamps, discharge lamps) produce
extremely sharp lines of 0.001-0.01-Awidth, unobtainable by filters or
monochromators. Second, parallel light beams can be obtained in most
instruments of conventional dimensions and the problem of chromatic
aberrations in the optical system is not encountered when working with
monochromatic light. Third, under most experimental conditions, flame
matrices are transparent to the analyzing light. Only where nonvaporized
or unburned particles occur in the flame can light scatter be observed.
Fourth, the absorbing units in a gaseous phase are considerably farther
apart from one another and from other atoms and molecules than in a
liquid phase, as in conventional molecular photometry.
It has been found, however, in practice that a perfectly straight
analytical working curve (-log T plotted against concentration) is
seldom obtained in atomic absorption spectroscopy. The reasons for this
are usually a combination of instrumental problems; broadening of the
emission line of the light source due to self-reversal, Doppler and pres-
sure broadening of the absorption lines of the atoms in the flame, failure
to exclude flame emission entirely, use of a focused instead of a parallel
 ATOMIC ABSORPTION SPECTROSCOPY 5

light beam, flame inhomogeneities, and possibly atomic fluorescence

(Section 3.2) contribute to interfere with the validity of Beer's Law.
The problem is discussed in detail under methodology (Section 5 ) .
3.2. ATOMICACTIVATIONAND EXCITATION
While mercury evaporates significantly at room temperatures, other
metals require a supply of some form of energy to produce an atomic
vapor. Most of the activated atoms in such a vapor are present in the
ground state. If the supply of energy is sufficient, a small percentage of
these atoms can be excited to a higher energy level, at which they remain
for a very small period of time and then return to the ground state,
thereby giving off their excess energy as light. Several discrete levels of
excitation are possible which are reached in a stepwise quantized proc-
ess. When a certain maximum level of excitation is attained, further ex-
citation gain results in loss of the outer electron and the atom becomes
an ion.
The return from higher energy states to the ground state of atoms
takes place in the same stepwise transition between energy levels, re-
sulting in emission of radiation of specific wavelengths, the characteristic
emission spectrum. Atoms in the ground state can absorb on any of the
wavelengths corresponding to transitions between energy levels, but it
is usually the first resonance line representing the transition from the
ground state to the lowest excited state where absorption is strongest.
Exceptions to this are elements (iron, manganese ) with complicated
emission spectra, where more than one strong resonance line is en-
countered.
The number of excited atoms in a flame is exponentially dependent on
flame temperature ( Boltzmann relation). For example, for sodium, one
of the most easily excited species, the ratio of excited to unexcited
atoms changes at temperatures of 2000 and 4OoO"C from the order of
to respectively. In regard to the population of excited atoms
this means an approximate 1000-fold increase, but only an insignscant
decrease (less than 0.1%) of atoms in the ground state. The dependence
of flame emission on flame temperature is well known (D10, P5, R6),
while absorption is much less affected. Other metals, which require
much higher levels of energy for excitation, produce such small numbers
of excited atoms at ordinary flame temperatures, that their measurement
by emission flame photometry is difficult or impossible. Most of these
elements, however, at the same temperatures produce significant con-
centrations of atoms in the ground state which are then amenable to
measurement by atomic absorption.
6 ALFRED ZETTNER

Most of the literature on atomic absorption neglects to mention that

the light that is absorbed on atomic resonance lines causes the absorbing
atoms themselves to undergo transitions to a higher energy state. Their
deactivation in returning to the lower energy states by emitting light
on the specific atomic lines is a form of atomic fluorescence, since excita-
tion is from radiational, rather than thermal, energy sources. The use of
atomic fluorescence flame spectrometry in the analysis of metals has
recently been described (W16, W17) although the principle has been
known for many years, One particular advantage of this method should
be emphasized, that sensitivity can be altered by changes in the in-
tensity of the exciting radiation.

3.3. FLAME9

Coincidentally with the revival of atomic absorption methods, renewed

interest in flame processes has arisen among many groups of workers
(Bl, F7, G3, M5). Absorption techniques including the line reversal
apparatus have contributed significantly to our understanding of atomic
activation, excitation, atomic population densities, and temperature
gradients. At the present time, the flame represents the most convenient
means to create an atomic vapor under reproducible conditions. For these
reasons, a short discussion of processes taking place in the flame is
needed. Other means of production of activated atoms will be men-
tioned under instrumentation and techniques ( Section 4.5).
Combustion of a fuel-air or -oxygen mixture progresses with a certain
speed, the burning velocity. If the combustible mixture flows against the
direction of combustion, and flow velocity and burning velocity equal
each other, a stationary flame can be obtained, for instance, at the
opening of a tube. When the flow velocity is greater the flame will lift
off the tube opening, and when the burning velocity exceeds the former
the flame will progress into the tube and cause a back-flash.
Stationary flames can be produced by premixing the fuel and air be-
fore they emerge from the burner opening; these are so-called premixed
flames. If the fuel and air are fed separately and mixing occurs largely
by diffusion of the two gases into each other at the burner opening, such
a flame is called diffusion flame.
Stationary flames are composed of different zones: a base cone of
nonluminous, unburned gases, corresponding to the transport zone in
which preheating up to about 350°C takes place. This is immediately
followed by a thin, highly luminous zone, the primary reaction zone,
with temperatures between 1500 and 2000"C, depending on the type of
 ATOMIC ABSORPTION SPECTROSCOPY 7

fuel. The large remainder of the flame plume is divided into the second-
ary reaction zone, where the temperatures reached are highest, and into
a large recombination zone. In the latter temperatures generally are
somewhat lower, due to entrainment of the surrounding air.
Burning of the fuel is a stepwise degradation leading to many inter-
mediate products and formation of radicals. Some of these are of interest
because of their ability to interact with metallic atoms, notably atomic
oxygen and the OH radical. If the available oxygen equals the theoretical
amount necessary to burn the fuel completely, such a flame is called
stoichiometric. Otherwise, depending on the amount of fuel, we speak
of a fuel-rich or a lean flame. Where fuel and metallic atoms compete
for 0 and OH radicals, the formation of metal oxides can be kept down
by making the flame fuel-rich. Fassel et al. (F2, F3) have recently shown
that satisfactory atomic vapor concentrations can be produced in fuel rich
flames with a variety of metals particularly prone to form refractory
oxides in stoichiometric flames.
TABLE 1
FLAMETEMPERATURES
OF COMMON
FLAMES~
~ ~~ ~

Temperature ( "C)
Fuel With air With oxveen
Coal gas 1700-1840 2700-2730
Propane 1925 2800
Hydrogen 2045-2100 2730-2780
Acetylene 2125-2200 30504100
a Composed from references (D11, M5, W19).

The temperature of flames is dependent on the type of fuel and, of

course, on the reactant air or oxygen. A large amount of nonreactant
nitrogen in air flames absorbs energy, resulting in lowering of the
general flame temperature. The maximum temperatures of some typical
flames are given in Table 1. Graphs of isothermic lines (D11, M5) show
that the maximum temperature is reached quite close to the top of the
base cone, an area usually coinciding with the highest sensitivity in ab-
sorption. Maximum temperature is confined to a narrow zone of a few
millimeters' cross section in lean or stoichiometric flames, but in fuel-
rich flames this zone is spread out over several centimeters.
Under ordinary conditions flames are transparent over a wide range
of wavelengths, a fact found with surprise by FoucauIt in 1849 (S9) and
probably by everyone else since with an interest in this subject. Most
flame constituents are gases in atomic or molecular form, and whatever
8 ALFRED ZETTh’ER

absorption is found of light shone through a stoichiometric flame is con-

fined to atomic (or possibly ionic) lines and molecular bands. Oxygen
flames, however, are opaque for wavelengths below 2000 A because of
the strong absorbing power of oxygen in the vacuum ultraviolet. The
latter reason appears to be the most serious limitation to the use of
absorption techniques for the nonmetals, most of which have resonance
lines located below 2000 A.
The gas velocities in flames are of the order of several meters per
second. Consequently, an absorbing atom will pass through the zone
monitored by the light beam with an approximate cross section of one
centimeter in a few milliseconds. This points toward the relative “waste-
fulness” of flame procedures, considering that ( a ) at least 10-20 seconds
of time are necessary for one measurement during which a steady stream
of activated atoms has to be maintained and ( b ) that in the ordinary
premix-atomizer-burner only a few per cent of the total sprayed sample
effectively reach the flame. Clearly, if sensitivity gains are sought, im-
provements in atomization and atomic activation methods appear to
represent the most logical approach.

4. Instrumentation and Techniques

4.1. COMMERCIAL
INSTTWMENTS
The early workers in this field built their own apparatus, often assem-
bling these from suitable units available from instruments for other
purposes. Emission flame photometers have been converted into atomic
absorption spectrophotometers by appropriate attachments consisting of
specifk light sources, choppers, and lenses, a principle also employed by
some manufacturers.
A number of good commercially built atomic absorption spectrophoto-
meters are available now from the following companies: (1) Bausch
and Lomb Optical Company, Rochester, New York; ( 2 ) Hilger and
Watts, Ltd., London, England; ( 3 ) Jarrell-Ash, Newtonville, Massachu-
setts; (4)Messrs. Jobin-Yvon, France; ( 5 ) Optica U. K., Ltd., Gateshead,
England; ( 6 ) Optica S.p.A., Milan, Italy; ( 7 ) Perkin-Elmer Corporation,
Nonvalk, Connecticut; ( 8 ) Techtron Pty, Ltd., Melbourne, Australia.
The instruments by Nos. 6 and 7 are double beam spectrophotometers,
the others are single beam instruments. Instrument No. 8 is distributed
in the United States by Aztec Instruments, Inc., Westport, Connecticut.
For detailed information about the various instruments, the reader is
referred to the manuals available from the different companies and
descriptions in the literature (B7, C2, D1, G2, K l , M7, M12, S 3 ) .
 ATOMIC ABSORPTION SPECXROSCOPY 9

4.2. BASICPRINCIPLES OF THE ATOMIC ABSORPTION SPECTROPHOTOMETER

An atomic absorption spectrophotometer operates on the following

principle. Light from special monochromatic sources with the wavelength
of the resonance line of the element to be determined is passed through
a vapor of neutral atoms. The diminution of the light beam by the ac-
tivated atoms is monitored by a detector and recorded by a read-out
system. Because the wavelength of the light emitted from the excited
atoms in the flame is identical with the wavelength of the source line,
radiation from both, the lamp and the flame, is passed by the mono-
chromator into the light detector. In order to prevent the undesired
flame radiation from interfering with the measurement, the light source
is modulated and the currents from the detector are fed into an a.c.
amplifier. The latter will amplify only the a.c. components of the output
of the detector, while d.c. currents are thus excluded from measurement.
4.3. THE HOLLOWCATHODE
TUBE
Paschen (P2) in 1916 while experimenting with Geissler tubes found
that on lowering the pressure of the noble filler gas to 1-2 mm mercury,
the light emission would retreat into the hollow cathode, which he had
constructed as a small box of sheet aluminum. In the spectrum of the
hollow cathode light, he found the expected emission lines of the noble
filler gas, but also strong aluminum lines. Modem versions of hollow
cathode lamps are sealed vacuum tubes filled with a noble gas (neon,
argon) at a pressure of a few millimeters of mercury ( J 2 ) .The purity of
the filler gas and its exact pressure are quite critical. Mounted inside
the tube are a rod-like anode and a hollow cathode usually consisting of
a cylinder with an open front. The front window of the tube is preferably
made of quartz to transmit also light of short wavelengths. When a vol-
tage is applied across the two electrodes, the filler gas becomes partially
ionized, but ionization at the low filling pressure remains largely con-
fined to the interior of the cathode. The noble gas ions, traveling toward
the cathode, start to strike the interior of the cathode thereby sputtering
off atoms of the metal composing it, in sufficient numbers to give rise to
a cloud of metallic atoms. These collide with the idler gas and each
other and thus become excited to give off light on the atomic lines
characteristic for their species. In this way the emission spectrum of
any metal can be obtained that composes or lines the inside of the
cathode. Tubes with more than one cathode composed of different
metals have been constructed ( J 2 ) .
The harder the cathode is struck by the filler gas ions, the more atoms
are sputtered off, resulting in greater intensity of the emitted light.
 FIG. 1. Spectral scan of the emission of calcium hollow cathode tube (Ransley Class, Melbourne, Australia). The emis-
sion spectrum is dominated by the strong resonance line of calcium at 4227 A (reproduced from reference (23) by per-
mission from the editor of Clinical Chemistry).
 ATOMIC ABSORPTION SPECTROSCOPY 11

Therefore, intensity of cathode emission increases with greater cathode

current and with higher atomic weights of the Uler gas species. Helium,
with its low atomic weight and relatively high ionization potential,
generally is least suitable ( T l ) .
The lines emitted by hollow cathode tubes run at room temperature
generally have a spectral width of less than 0.01 A. This value, however,
is dependent on the tube current. As the current is increased, temperature
rises and a denser population of neutral metal atoms is obtained in the
cathode. This results in greater width of the emission lines, due to in-
creased Doppler broadening and increased self-absorption, respectively.
For optimal results, it is desirable to operate with line widths not greater
than one third the width of the resonance line of the particular metal
being measured in the flame. The latter generally is of the order of
0.03 A. A spectral scan of the output of a calcium hollow cathode lamp,
as shown in Fig. 1, is indicative of the spectral purity of these light
sources (€38). The use of iron hollow cathode lamps in wavelength
standardization has been suggested (C3, S7).
In order to determine optimum operating conditions for a lamp, ab-
sorption of a given test solution should be measured at various tube
TABLE 2
OF CALCIUMHOLLOW
PERFORMANCE CATHODETUBE AT VARIOUS
LAMP CURRENT.+
Operating current Absorption readings
(mA) (%I
5 49.5
8 47.6
10 45.8
12 44.0
14 42.5
16 42.0
18 41.5
Absorption readings obtained with 1 mg % aqueous calcium solutions (Z2).

currents. As the tube current is decreased, the line width decreases,

but absorption of this light in the flame increases. The line width can
be considered satisfactory when further lowering of the tube current
does not increase absorption; the current at that point should be chosen
as the operating current. Most lamps, however, require a certain mini-
mum operating current to deliver a steady discharge, and for this
reason some lamps have to be run at currents higher than those at which
maximum sensitivity is observed. In these cases a compromise must be
found between reduced sensitivity and improved stability of lamp dis-
12 ALFRED ZETTNER

charge and meter reading. Table 2 shows absorption obtained with a

10 ppm calcium solution at various lamp currents.
While the first hollow cathode tubes were constructed in such a way
that they could be repeatedly flushed with the purified noble gas, the
inconvenience connected with such equipment led to the development
of permanently sealed tubes. In order to insure a reasonable lifetime of
such tubes, they have to be of a certain minimum volume. One of the
reasons for the lifetime limits is leakage of air into the tube, but more im-
portant seems to be the loss of the filler gas which is slowly absorbed
by the metal and the glass surface. Since the lamp operates by the
sputtering off of the cathode lining, gradual loss of the latter leads to
eventual deterioration of the lamp. Lamps for metals that sputter abun-
dantly, like the alkali metals, or zinc and cadmium, have short lifetimes,
mostly well below a hundred hours.
4.4. THE BURNER-ATOMIZER
Since atomic absorption spectroscopy in its present form was largely
developed on the basis of techniques known from flame emission, many
of the original absorption studies were done with means of vaporization
available from flame photometry. It was soon realized, however, that
these burners had certain shortcomings in absorption and this led to a
number of modifications of existing burners or new designs. The flame
needed in emission is to concentrate a sample in an area as small as
possible, just large enough to illuminate the slit uniformly and as slender
as possible to minimize self-absorption. Absorption techniques, on the
other hand, require an elongated flame to provide an absorption path
as long as possible.
Both types of burner-atomizer, the total consumption type and the
discharge type, have been used in absorption. The total consumption
type consists essentially of three concentric tubes, all of which end
approximately at the same tip level. The sample is aspirated through
the central capillary by the shearing-off action of the oxygen passed
through the middle tube, and sample spray and oxygen mix into the fuel
emerging from the outer tube. Its advantage is the relatively low sample
consumption of about one milliliter per minute. Although the whole
sample is introduced into the flame, only a fraction of it contributes to
emission while larger droplets pass through the flame ineffectively. From
the foregoing, it can easily be seen that with this burner an elongated
flame is difficult to produce. Nearly everyone employing it in absorption
found it necessary to use several burners in series to obtain the desired
absorption path length, or to use multiple passes of the light beam
 ATOMIC ABSORPTION SPECTROSCOPY 13

through the flame, or both. The discharge type of atomizer employs two
capillaries that open into a spray chamber and are either arranged con-
centrically or at a right angle to each other. As air or oxygen is forced
at high pressure and usually with supersonic speeds through one capil-
lary, the liquid sample is sheared off the other tube and dispersed in a
fine mist. Larger droplets settle out and are drained off as waste. Only
the smallest droplets are carried by the air stream through a connector
into the burner, which in this case usually is of the premix type. The
small droplets evaporate readily in the premix area or preheating zone
of the flame so that probably all of the sample that reaches the flame is
effectively vaporized. The discharge atomizers have a relatively high
sample consumption, 3-20 or more ml per minute, but only 3 4 % of
it effectively reaches the flame. The sample flow rate together with
the waste rate permit calculation of the sample volume per minute
actually introduced into the flame. The figures obtained are similar to
those of the total consumption atomizer, indicating that population
densities of activated and excited atoms are of the same order in both.
A burner-atomizer with mechanically controlled sample flow has been
described by Robinson and Harris (R7). Herrmann and Lang (H2, H4,
L2) recently conducted extensive investigations of the efficiency of
atomizers and derived formulas for optimal conditions of capillary size,
volume of spray chamber, and gas and sample flow rates. The effect of
methanol on the efficiency of sample atomization was studied by Wine-
fordner et al. ( WlS).
While it is technically difficult to construct burner-atomizers of the
total consumption type to deliver an elongated flame, this is much more
readily accomplished by the combination of discharge atomizer and pre-
mix burner. When the burner head of the latter is given the shape of
a fish tail or modified into a narrow long slot, an elongated, thin, blade-
like flame is obtained, satisfying the needs for a long absorption path.
A versatile burner-atomizer in one integral unit has been described by
Willis (W9) and Slavin ( S 3 ) . It consists essentially of a bakelite spray
chamber mounted on top of which is a burner head of a T-shaped
barrel of stainless steel with a narrow slot of 10-cm length. The entire
air supply passes through the atomizer capillary and fuel, air, and sample
mist are premixed in the spray chamber. From 90 to 97% of the aspirated
sample drains through a waste tube, The stainless steel burner head
warms up considerably during operation, which effects evaporation of
the fine droplets carried forward by the fuel air mixture. This can easily
be demonstrated by aspirating a sample without the flame lit; a fine
mist is seen emerging from the burner slot of the cold burner, but not
14 ALFRED ZETTNER

when the flame is extinguished after a few minutes’ burning. This is

probably reflected in the fact that sensitivity more than doubles after
the warming-up period.
An interesting solution to the problem of lengthening the absorption
path has been described by Fuwa and Vallee (F10). The flame from a
total consumption burner-atomizer (Beckman) is directed into a long
narrow tunnel in which the flame is propagated over a distance of about
90 cm. The light beam of the hollow cathode tube is directed through
the tunnel which functions as absorption cell. Sensitivity increases of
10-100-fold are stated. The material lining the absorption tunnel and its
reflectivity for light were found to be related to the degree of sensitivity
gain.
A Beckman total consumption burner was modified into a premix type
by Kniseley et al. (K11). The burner is mounted into the lower end of
a brass tube and centered with aligning screws. A graphite tube is in-
serted into the brass tube from the top in such a way that a small space
remains between the top of the Beckman burner and the bottom of the
graphite tube. Premixing occurs in this space and in the bore of the
graphite tube. It is maintained that the advantages of the modification
are an improvement of sensitivities by an order of one magnitude,
probably due to the marked reduction in flame noise. The use of the
burner is limited to solutions containing at least 50% of organic solvents.

4.5. OTHERMEANSOF SAMPLEVAPORIZATION

Although the flame appears to represent the most versatile means to
vaporize a sample, it is limited to materials dissolved in a liquid that
can be aspirated. Furthermore, some elements fail to produce effective
vapors in ordinary flames. Several other methods of vaporization have
been described, which are applicable to samples in the solid state and
to elements not amenable to measurement in the flame. A short note on
the application of atomic absorption to solids has been published ( S l )
but no details are given,

4.5.1. Cathode Sputtering

The principle of the hollow cathode tube, production of a vapor of
atoms by cathodic sputtering, has been employed by Gatehouse and
Walsh ( G l ) for sample vaporization. The sample is introduced into a
vacuum chamber and is made the cathode which produces a cloud of
activated atoms. The light of a separate hollow cathode tube is passed
through this vapor and absorption is measured in a spectrophotometer.
 ATOMIC ABSORPTION SPECTROSCOPY 15

The method is said to yield reproducible results, and a straight working

curve was obtained with silver,
4.5.2. Flameless Crucibles
L'vov (L12) developed a flameless furnace applicable for atomic
absorption spectroscopy. A tube-shaped graphite crucible is inserted
into a chamber, fitted at both ends with quartz windows. A carbon
electrode is installed in the middle of the graphite tube and another
one outside. Solid samples weighing less than 0.1 mg are placed at the
tip of the electrode inside the tube and the chamber is closed, evacuated,
and filled with argon at low pressure. The graphite crucible is heated
electrically to 2000°C. Then an arc is struck between the carbon elec-
trodes whereby the small sample is suddenly evaporated and an atomic
vapor is created inside the graphite crucible. The concentration of a
metaI in the vapor is dependent upon the total amount present originally
in the sample placed on the carbon electrode. The beam of an appro-
priate hollow cathode tube is passed through the long axis of the
graphite tube and absorption is measured by a spectrophotometer. It is
stated that no interference could be found for any of the elements tested,
that reproducibility was excellent, and that only 3 minutes were required
for one determination.
4.5.3. Capacitor Discharge Lamp
Nelson and Kuebler ( N l ) described a modified capacitor discharge
lamp into which samples could be introduced for vaporization. The
single, intense light pulse of these lamps results in flash heating of the
samples to several thousand degrees Celsius, by which a dense cloud
of atomic vapor is produced that persists for a few thousandths of a
second. By spectrographic means the authors were able to record the
absorption spectra of about 20 metals including copper, iron, lead, cal-
cium, magnesium, and zinc.
4.5.4. Cool Vapor
Mercury, the only metallic element with significant volatility at room
temperature, has been conventionally determined for many years by
atomic absorption spectrometry, as the mercury vapor detector ( WZO)
is based on this principle. Lindstrom (L7) used a flame to volatilize the
mercury in the liquid sample, but determined its concentration in the
exhaust gases with the mercury vapor meter after cooling and purifica-
tion in a filter that removed particulate matter. The method is said to
be capable of detecting 0.1 pg "/o of mercury in the original liquid sample
16 ALFRED ZETTNER

sprayed into the flame. Urine could be sprayed directly and plant
materials were extracted.
4.6. WAVELENGTH
SELECTION
It should be pointed out here that wavelength selection in atomic
absorption spectroscopy is largely accomplished by the choice of the
monochromatic sharp line source, possessing the wavelength of a res-
onance line of the element to be determined, a specificity of selection
unobtainable by any other means. Any additional wavelength selection
can be considered merely secondary and the methods to this end should
be examined with this in mind.
If the line source emits radiation on the resonance line only or if the
line intensity is so great that any background radiation is rendered
negligible, no additional wavelength selection is necessary. This is the
case with sodium for which it was possible to construct an atomic ab-
sorption photometer without a wavelength selector ( B7).
Where only the determination of alkali metals is desired, a simple
instrument can be devised in which selection of a resonance line can be
obtained with the help of color filters. However, at present such
instruments would not appear to have an appreciable advantage over
flame emission methods, considering the sensitivity of the latter and the
simplicity of presently available flame photometers. Appreciable advan-
tages, on the other hand, are inherent in absorption over emission
methods in the determination of the alkaline earths and magnesium.
Since these metals have simple emission spectra, the use of filters,
notably interference filters, would be feasible in instruments limited to
the determination of these elements.
Whenever it is intended to determine a variety of elements with one
instrument-and versatility is one of the major advantages of atomic
absorption-a high dispersion monochromator of good resolving power
and variable wavelength selection is required. It should be capable of
transmitting wavelengths from 2000 to 9000 A and therefore has to con-
sist of either a quartz prism or a diffraction grating. Grating monochro-
mators have linear dispersion but their resolving power suffers from
light scattering and from interference from higher diffraction orders.
Quartz prisms have the disadvantage that dispersion is nonlinear (smaller
at the long wavelengths) and only one third that of glass. The light
emitted by hollow cathode tubes is composed of the sharp lines of all
excited atoms; the filler gas, the lining of the hollow cathode, and the
supportive metals of the cathode, especially when made of an alloy,
contribute their own specific emission spectra. Since the monochromator
 ATOMIC ABSORPTION SPECTROSCOPY 17

will have to separate just a few lines and not a continuum, interferences
from light scattering and other diffraction orders are minimized when
using a grating monochromator. The latter has been found to satisfy
most requirements even where complex spectra are involved.
4.7. ABSORPTION MEASUREMENT AND RECORDING

The measurement which is sought in absorption photometry is the

degree of attenuation of the light beam by a sample introduced into the
absorption path; this value is compared to the amount of light energy
arriving at the detector without the sample. The hollow cathode tube,
providing stable operation, emits Iight of a constant intensity. When
passing through the flame, this light is attenuated depending on the
presence of atoms of the corresponding species. This reduced light then
is transmitted by the monochromator to be measured by the detector.
But some of the sample atoms in the flame become excited to emit light
themselves on resonance Iines or wavelengths identical with those emitted
by the hollow cathode tube. Therefore, flame emission together with
cathode emission will be received by the detector. This is somewhat
similar to the situation in molecular absorption photometry when the
sample shows undesired fluorescence. But while fluorescence usually is
of different wavelengths than those of the incident light and can there-
fore be diverted from measurement, the lines emitted by the sample
atoms in the flame and by the cathode are the same and cannot be
separated from each other by the monochromator. In order to exclude
the undesired flame emission from measurement, several possibilities
exist. When working with metals possessing absorption lines in the deep
ultraviolet, flame emission on these lines is usually small enough to allow
for electrical correction with so-called “bucking circuits.” Another in-
teresting method (L9) is the use of lenses at both ends of the flame
which are positioned in such a way that they focus the light from the
sharp line source at the entrance slit of the monochromator, but light
emitted from the flame is maximally defocused there.
4.7.1. Light Beam Modulation
The most versatile and convenient method to overcome flame emission
applicable to all elements is the modulation of the light emitted by the
hollow cathode. This can be accomplished by running the lamp on al-
ternating current or by interrupting the light beam mechanically with
a rotating disk. The photoelectric current from the photomultiplier tube
is then amplified by an ax. amplifier. The d.c. component of the output
of the photomultiplier tube, deriving from the continuous flame emission,
18 ALFRED ZETTNER

can thus not be amplified and thereby is excluded from measurement.

Since the a.c. amplifier is tuned to the frequency of the modulation of
the light source (usually 50-120 cycles per second) , the random modula-
tion of flame emission due to flame noise will be rejected. Only that
component of flame noise possessing the modulation frequency of the
lamp theoretically can contribute to the output of the amplifier, but in
practice this is entirely negligible.
4.7.2. Single Beam us. Double Beam
Single beam as well as double beam systems have been employed in
atomic absorption spectrophotometers. The question of preferability will
depend on the applications of the instrument but in general probably
can be answered in favor of the double beam. The instruments first
built by Alkemade and Milatz (A2,A3) in Holland and Russell et al.
(RQ)in Australia were both based on the double beam principle. Later
workers seeking simplification of the equipment were able to do satis-
factory work with single beam systems. The problems of reducing lamp
drift are considerable and it seems that even with optimal conditions
the accuracy of measurements is distinctly below that achieved with
double beam instruments, a shortcoming readily admitted by most who
have worked with single beam. While drifts and noise levels of single
beam instruments are of the order of 1% or more, the double beam
system reduces this generally to 0.1%. Gidley and Jones (a), in order
to reduce short term fluctuations, used an integrating system in which
the output of the photo cell was fed to a condenser for a certain period
of time. The ratio of these signals, integrated in this way between the
two voltages obtained with no sample and with the sample sprayed, is
measured. Since the sensitivity of an instrument ultimately is limited by
the signal to noise ratio, the double beam system with its inherent
stability will have to be employed wherever small absorption measure-
ments are sought with high accuracy. High stability further permits the
use of electronic scale expansion techniques often with considerable
improvements of the limits of detectability.
4.7.3. Detectors and Read-Out
Barrier-layer cells and photomultiplier tubes have both been used for
photodetection in atomic absorption spectroscopy. The use of barrier-
layer cells of course is limited by their sensitivity and the difficulty en-
countered in amplifying their output. They will suffice where determina-
tion of the alkali elements is desired only. For most other work photo-
multiplier tubes are necessary. These are available for a broad spectral
 ATOMIC ABSORPTION SPECTROSCOPY 19

range from 2000 to 7000 A. Their sensitivity increases with their operat-
ing voltage; when working with low emission levels of the hollow
cathode tube, the shot-noise effect of the photomultiplier run under high
voltage becomes the limiting factor.
The problems of modulation of two light beams, their optical and
electronic separation, and ratio recording have led to interesting technical
solutions. In the atomic absorption spectrophotometer of Russell et a2.
(R9) the reference beam was chopped at twice the frequency of the
sample beam. The beams were optically recombined, the signals from
the detector electronically again separated, and their ratio recorded. In
a later developed commercial instrument ( Perkin-Elmer Atomic Ab-
sorption Spectrophotometer Model 214) the modulated hollow cathode
emission was split into sample and reference beams, which were passed
into two separate photomultiplier tubes and their amplified signals
recorded by a self-nulling digital read-out. In another version, light
modulation and beam splitting are elegantly accomplished by a rotating
chopper fitted with sectional mirrors. Both beams are recombined and
received by one detector, which eliminates the costly process of match-
ing a pair of photomultipliers. The recording is done by a manual nulling
slide wire connected to a digital counter. An interesting system of ratio
reading without beam splitting (!M6) is based on the use of two separate
lines emitted by the hollow cathode tube, of which only one line is
absorbed. The ratio of the intensities of these two lines is related to
sample concentration in the flame. Unfortunately, intensity changes
due to lamp drift are not always the same for both lines, so that a steady
intensity ratio is not maintained.
5. Methodology
The technique of actually performing measurements is simple and
almost identical to that in flame emission photometry. A liquid sample,
free of particulate matter and containing the sample metal in solution,
is aspirated into the flame. The degree of absorption, usually read out
as % transmittance ( T ) ,but more conveniently as % absorption ( 1 -
T ) , is converted to absorbance or optical density ( 2 - log T ) in the
conventional way. Sample concentrations are obtained from interpolated
readings or absorbance-concentration plots of working curves derived
from appropriate standard solutions. Even under most ideal instrumental
conditions, i.e., low lamp current, homogeneous flame, narrow slit width,
optimal monochromator efficiency, etc., analytical curves usually bend
toward the concentration axis. The reasons for deviation from a straight
line are complex, and maximal correction of one contributing factor
20 ALFRED ZETTNER

usually does not eliminate the curvature. The effects from Doppler
broadening and self-reversal of the source line can be minimized by
running the lamp on a low current as discussed above, but the con-
siderable broadening of the resonance lines due to the high temperatures
in the flame is difficult to reduce. Flame inhomogeneities, as mentioned
by Menzies ( M 6 ) , should be corrected wherever possible. These in-
homogeneities cause variations in absorption in different parts of the

Inches a b v e burner top

FIG. 2. Vertical absorption profile of 10-cm ahacetylene flame. Lower curve
obtained with aqueous calcium solution, upper curve with isopropanol (from refer-
ence (S5) by courtesy of Perkin-Elmer Corporation).

elongated flame in regard to its horizontal as well as vertical extension.

The sensitivity profiles obtained by scanning the vertical extension of
the flame (Fig. 2 ) indicate considerable absorption differences from
the base to the top of the flame and since the sample light beam pos-
sesses at least a few millimeters' width, different degrees of absorption
will occur in the lower and upper part of the light beam. The problem
may be slightly corrected by using fuel-rich flames in which the zone
of maximum absorption is widened vertically and by focusing the light
beam onto the center of the flame, but in the latter case the lack of a
parallel sample beam might worsen the problem. A T-shaped burner
 ATOMIC ABSORPTION SPECTROSCOPY 21

head may lead to higher sample concentrations in the middle third of

the flame; (23)a baffle of the shape shown in Fig. 3, placed into the
burner head below the slot, corrects this problem. When using hollow
cathode tubes with simple emission spectra, the monochromator slit
width influences neither sensitivity nor curvature of the calibration graph
significantly, but for metals with complex spectra (iron, manganese,
nickel) the slit setting is critical. Atomic fluorescence (Section 3.2)
elicited by the hollow cathode radiation in the flame would aIso tend to
curve the calibration graph toward the concentration axis.
When working with curved calibration graphs, a large number of
working standards of varying concentrations would be necessary to

FIG. 3. Shape of the baffle placed into the barrel of the burner head to produce
uniform gas and sample flow over the entire slot length. The dotted line indicates
the outline of the burner head and slot (from reference (23) by permission).

determine the curvature as closely as possible. In practice, four to five

calibrating solutions have to suffice but linear interpolation between
these few points may lead to considerable error. Since the degree of
curvature for a particular metal and a given set of instrumental con-
ditions undergoes less change from day to day than the slope of the
curve, it is possible to make a template possessing the “mean” curvature
of a number of runs which is then used instead of a straight ruler to
connect the points on the analytical curve. Since curves drawn in this
way are a much better approximation of the “true” curve than those
drawn by straight interpolation, sample values read off the former will
be more accurate.
While industrial or agricultural chemists for the most part have large
samples available for analysis, the clinical chemist is faced with an ever
increasing number of tests to be performed on a small single sample,
for instance a few milliliters of serum. The sample size in atomic absorp-
tion spectroscopy depends on necessary sample dilutions, aspiration
rate, and time of aspiration required to obtain one reading, These are a
function of sensitivity and instrumental stability. Since a few seconds
(usually 10-30) of sample aspiration suffice for one reading, the total
22 ALFRED ZETTNER

amount of samples sprayed usually is 1-5 ml. Sodium, potassium, calcium,

and magnesium, because of their relatively high concentration in bio-
logical materials, require high dilutions and only small samples of the
original material, the minimum size of which is restricted only by the
impracticality of measuring small volumes rapidly and accurately. Iron,
copper, zinc, manganese, and other “trace” metals have to be determined
on undiluted or even concentrated materials. In these cases, removal of
proteins, ashing, or extraction of the analyte from blood, urine, or tis-
sues will be part of the sample preparation for aspiration.
5.1. SENSITIVITY
Sensitivity or detection limit in atomic absorption spectroscopy has
arbitrarily been defined as that concentration of a particular metal,
usually in aqueous solution, to give an absorption signal of 1%. The
limits were largely derived from work with single beam instruments
where 1% absorption may represent the minimum signal distinguishable
from a noisy background. The improvements of instrumental stability
with double beam systems make it possible to read much smaller ab-
sorptions with certainty, and the limit of detection then is set by the
limit of “readability.” The possible use of scale expansion may enhance
the detectability by one order of magnitude ( K l ) .
The sensitivities for various elements differ considerably from each
other, magnesium probably being the most sensitive element (0.01 ppm)
(Z4), but with most metals satisfactory work can be done in the range
of one to a few parts per million. When determination of a number of
elements of varying sensitivities is desired in solutions containing mix-
tures of the analytes, it may occasionally become necessary to reduce the
sensitivity for one or more elements without diluting the sample. It is
clear, however, that sensitivity should not be lowered by changes in the
optimal instrumental conditions, as such alterations not only would re-
duce sensitivity but also increase the curvature of working curves.
Sensitivity reduction without working curve deterioration, in such cases,
is best achieved by shortening the absorbing path of the flame, either
by turning the burner head (W11) or by closing the burner opening
partially with a suitable piece of metal.
While sensitivity reduction may be desirable only rarely, the necessity
for sensitivity increases arises on numerous occasions and methods to
this end have been studied by many. Double beam instruments and
scale expansion techniques provide optimal experimental conditions and
presently little gain can be expected from further instrumental im-
provements. Elongations of the absorption path for sensitivity increases
 ATOMIC ABSORPTION SPE(;TROSCOPY 23

have been employed by Fuwa and Vallee (F10) by the use of the
burner tunnel.
The effective absorption path length can also be increased by multiple
passes of the light beam through the flame with a system of mirrors
(W5). This or similar systems have been employed by several workers
(B2, R9, Z1) and are incorporated into one commercial instrument
(Jarrell-Ash). Actual sensitivity gains, however, seem to fall short of
gains expected from a certain number of passes; Russell et al. (RQ) with
12 passes achieved only a &fold sensitivity increase. Lang and Herrmann
( L l ) recently showed that the use of a mirror system has no significant
influence on the signal to noise ratio,
5.1.1. Organic Solvents
The first to show the enhancing influence of organic solvents in flame
photometry appear to have been Berry et al. (B5), who considered it
an interference. Since then the use of organic solvents has been studied
by many workers in emission flame photometry (B9, D11, D12, K4,
W3) and sensitivity increases up to 100-fold (D11) have been reported,
depending on the metal and the type of solvent used. Since atomizing
conditions in emission and absorption are similar, sensitivity gains
by the same means could also be expected in absorption. This was
confirmed and extensively studied by Allan ( AlO), Lockyer et al. ( LlO),
and Robinson ( R3). Although sensitivity enhancements of emission have
been obtained with organic solvents in discharge as well as total con-
sumption burner-atomizers, the mechanism involved is thought to be
different. Aspiration of an organic combustible liquid instead of water
into the flame of a total consumption burner results in higher flame
temperature. If an excitable metal is dissolved in the solvent, considerable
emission enhancement can be expected since emission intensity depends
exponentially on the flame temperature (Section 3.2). If too much of
the solvent reaches the flame, unburned portions of it will lower the
flame temperature and even a decrease of intensity might be observed.
On the other hand, raising of the flame temperature does not contribute
to increasing the population of atoms in the ground state, since many
metals are completely atomized from aqueous solutions in most flames.
Therefore, little sensitivity gain might be expected in absorption by the
use of organic solvents in total consumption burners. The experimental
findings of Robinson (R3) do not seem to corroborate this reasoning.
Using a Beckman burner (total consumption type) with an oxygen-
cyanogen flame, he found absorption enhancements for nickel up to
36-fold with a variety of solvents, while emission enhancement was
24 ALFRED ZE'ETNER

considerably smaller. The effect of self-absorption may be responsible

for the latter. Since both, emission and absorption, increase, the enhance-
ment cannot be due to a temperature effect only. One is led to conclude
that, with organic solvents, a larger sample reaches the flame and
atomization in the flame is more complete, due to smaller droplet size
and faster vaporization. A scheme of the behavior of aqueous and
organic solvent solutions in the flame is discussed by Robinson.
Allan (A10) conducted an extensive study of the effect of organic
solvents on absorption in an air-acetylene flame, employing a discharge
atomizer. In this system, where only a small fraction of the aspirated
sample reaches the flame in the form of the smallest droplets, the con-
centration of free atoms in the flame can be increased mainly by a larger
sample reaching the flame and by facilitating the dissociation of metal
compounds into free metallic atoms in the flame. Allan was able to show
that the rate of absorption enhancement achieved with one solvent was
approximately the same for metals completely atomized in the flame
(copper and zinc) and those forming compounds not completely dis-
sociable (iron and manganese). Since the dissociation of these com-
pounds is temperature dependent, flame temperature measurements were
carried out concurrently with absorption measurements. While flame
temperature, as compared to water aspiration, was found to be slightly
lower during solvent aspiration (due to the necessary acetylene flow re-
duction), absorption increased with all four elements tested. The con-
clusion that absorption enhancements in the flame by organic solvents
are indeed due to a larger sample reaching the flame was confirmed
by an experiment in which the spray issuing from the burner slot was
collected in glass wool and the eluate analyzed. When methylisobutyl
ketone solutions of copper were aspirated, 3.35 times as much copper
was found in the eluate as with aspiration of aqueous solutions. The
absorption measurements achieved with the same organic solvents were
3.5 times that of aqueous copper solutions. Zettner and Seligson (23)
obtained a 2S-40% increase of sensitivity by the admixture of 7%
butanol to aqueous calcium solutions ( Fig. 4). Other studies conducted
by Lockyer (L8) and Elwell and Gidley (EZ), using a system with a
separate spray chamber, showed that an admixture of isopropanol to
aqueous solutions also was capable of enhancing sensitivity. For a
variety of elements tested a sensitivity gain of =-fold was reported,
but there was considerable variation from element to element, iron
exhibiting an enhancement of up to 10-fold. Allan (A10) states that
the increases in sensitivity obtained with the solvents miscible with
water would be counterbalanced by the sample diIution incurred in
adding the organic solvent. The author feels that Allan's view is not
 ATOMIC ABSORPTION SPECTROSCOPY 25

entirely justified, because in some instances solvent dilution of originally

critically small samples would facilitate the analysis by providing
sufficient volume for aspiration without loss of sensitivity. Furthermore,
the use of solvent dilution would also be of advantage in samples con-
taining a mixture of the test element and interfering substances. By
adding the organic solvent the interference might be reduced to a
tolerable level due to the additional dilution, while retaining if not
increasing the sensitivity for the analyte. Zettner and Seligson (Z3),

FIG.4. Enhancement of calcium absorption by admixture of organic solvent to

aqueous calcium solutions: --o-o- aqueous calcium standards; -0-0- aque-
ous calcium standards with 7% n-butanol (upper curwe). Actual calcium concen-
trations are only one tenth of those given on abscissa (from reference (23) by
permission).

while studying the effect of organic solvents, found that the addition of
n-butanol to an aqueous diluent for serum considerably reduced the
troublesome clogging of the burner opening by coagulated and half-
burned protein. It was actually this elimination of burner clogging that
made practical the development of a procedure for serum calcium
without deproteinization. This effect may be related to changes in
surface tension.
5.1.2. Solvent Extraction
The use of water-immiscible solvents for increase of sensitivity led
to the development of the selective extraction method by Dean and his
26 ALFRED ZETTNER

associates (D11, D12). The method is based on the formation of a com-

pound between the metal to be analyzed and a chelating agent added
to the original aqueous solution, with subsequent extraction into a
suitable organic solvent. The latter is then aspirated for analysis. Several
advantages are apparent: ( a ) the selective extraction makes it possible
to concentrate the desired element manyfold; ( b ) the concentrating
effect is enhanced by the sensitivity increase that the solvent yields when
aspirated into the flame; ( c ) interfering substances are left behind in
the aqueous phase. Trace quantities far beyond the sensitivity in the
original sample can thus be made accessible to rapid determination.
The method was applied by Allan (A10) to the determination of
copper, zinc, manganese, and iron, which were quantitatively extracted
from aqueous solutions after compIexing with ammonium pyrrolidine
dithiocarbamate ( M1) into ethyl acetate or methylisobutyl ketone. With
ethylamyl ketone an extraction ratio of more than 100 was achieved. The
same chelator was used by Willis (W14) to extract lead, mercury, bis-
muth, and nickel from urine into methyl-n-amyl ketone with similar
concentrating ratios and sensitivity gains. The strict requirements of
element selectivity of the extraction methods as necessary for colori-
metric work can be considerably relaxed in flame photometry, an ad-
vantage fully realized by Dean (011).The absence of spectral inter-
ferences in atomic absorption permits accurate determination of one
metal in the presence of many others, and therefore specificity of the
extraction procedure is not needed. Furthermore, several metals can
possibly be determined in one extract. The substituted dithiocarbamates,
as described by Bode and Neumann (B6), permit chelation over a
wide range of pH values. Since the sample preparations for extraction
of biological materials in most cases entail hydrolysis or oxidation in
acid solutions, chelation has to be carried out at very low pH values.
Dean (D11) described the extraction of iron as the iron-acetylacetonate
chelate at a pH of 0. When working with resonance lines of very short
wavelength, the absorptive properties exhibited in the ultraviolet by
many organic solvents should be taken into consideration. The enhanced
excitation of other metals by organic solvents, which might lead to in-
creased spectral interference in emission flame photometry, is not a
limiting factor in absorption spectroscopy.
5.2. INTERFERENCES
Interferences can be defined as any physical or chemical agents
capable of either increasing or decreasing the degree of absorption
usually achieved with the test element in aqueous solution. All inter-
ferences in flame emission have one common effect, namely, changing
 ATOMIC ABSORPTION SPECTROSCOPY 27

the state or number of excited atoms. In atomic absorption, interferences

can act in two ways, i.e., by changes in the number of activated atoms
in the ground state, or by attenuation of the monitoring light beam by
processes other than atomic absorption. Fortunately, the latter is rarely
observed and it is actually the high transparency of flames over a wide
spectral range that makes absorption flame photometry feasible.
Spectral interference, a well-known common di5culty in emission
flame photometry, arises from excitation of other metals with emission
lines on wavelengths too close to be effectively separated from the line
undergoing measurement. In addition, the emission bands of molecular
compounds, notably metal oxides and hydroxides, often overlap with the
atomic lines. There are no similar effects in atomic absorption, as dif-
ferent elements with common resonance lines are unknown except per-
haps in the special case of isotopes (Section 5.4.3). A,form of spectral
interference is encountered with hollow cathode tubes emitting a com-
plex spectrum, from which it may be difficult to separate the resonance
line. This is circumvented by the use of high dispersion monochromators.
Light emitted from the flame itself is eliminated from measurement by
the modulation of the resonance line source or by chopping of the light
beam. The possible interference from atomic fluorescence should also be
mentioned here (Section 3.2). This effect, if significantly strong, would
be particularly difficult to exclude from measurement, if resonance fluo-
rescence is involved. The latter not only possesses the same wavelength
but also the modulation frequency of the hollow cathode emission.
Excitation interferences appear to be entirely negligible in absorption,
since the fraction of excited atoms of the total population in most flames
is insignificantly small (Section 3.2). A signscant reduction of the num-
ber of atoms in the ground state, however, can be brought about by
excessive ionization, as observed with the more readily ionized metals.
The degree of ionization is exponentially dependent on flame tempera-
ture ( Saha equation).
Potassium, rubidium, and cesium possess especially low ionization po-
tentials and at the temperature of the commonly used air-acetylene flame,
for instance 30-70% of the total number of these atoms may be ionized
(D11, F6). The degree of ionization of an alkali metal, however, is re-
duced by the presence of other easily ionized elements. The admixture
of such elements affords one means of controlling this type of interfer-
ence.
5.2.1. Anionic and Anion-like Interferences
A number of anions forming refractory salts with metals retard the
dissociation into free atoms in the flame. The result is a reduction in the
28 ALFRED ZETTNER

L h P mgo% 50 loo 300

FIG.5. The depressive effect of phosphate on calcium absorption. All samples

contained 1 mg ”/o calcium. Maximum depression is reached at 0.4 mg “/o phosphate
(concentration as phosphorus) (from reference (23) by permission).

0:

0.:
Lrl
V
z
a
go.
0
cn
m
a
0

rc ,
I 2 10 50 100
I’

SULFATE m g %
FIG. 8. The depressive effect of sulfate on calcium absorption. All samples con-
tained 1 mg % calcium. Maximum depression is reached at 1.3 mg % suIfuric acid
(from reference ( 23) by permission).
 ATOMIC ABSORPTION SPECTROSCOPY 29

+n+nimi
OXALIC ACID mg %

FIG. 7. The depressive effect of oxalic acid on calcium absorption. All samples
contained 1 mg % calcium. Maximum depression is reached at 2.5 mg % oxalic acid
(from reference (23)by permission).

0.I

FIG. 8. The depressive effect of EDTA on calcium absorption. All samples con-
tained 1 mg ”/. calcium. Maximum depression is reached at 9 mg % EDTA. At higher
concentration a moderate enhancement of calcium absorption is seen (from reference
( 23 ) by permission).
30 ALFRED ZETTNER

total number of vaporized atoms of the test element, including atoms

in the ground state and in the excited state as well, and therefore a re-
duction in the emission and absorption intensity. This phenomenon has
been known for many years in flame photometry. A crystal of calcium
phosphate fails to color an alcohol flame ( T2). Among all metals calcium
appears to be most severely subjected to this type of anionic interference,
and many other anions, i.e., sulfate, nitrate, aluminate, etc., are similarly
effective. The phosphate effect is the most pronounced and has been
studied by many workers. Dippel et aZ. (D15) based an indirect flame
photometric method for the determination of phosphate on this anionic
depression, and similar methods for other anions are certainly feasible
(M9).
When a series of aqueous calcium chloride solutions is prepared with
increasing phosphate content, and the absorbance readings obtained are
plotted against phosphate concentrations, a curve results as shown in
Fig. 5. Between phosphorus concentrations of 0 and 0.3 mg %, there
is a sharp depression of the calcium signal with a linear relationship be-
tween phosphorus concentration and depression. Once maximum depres-
sion is reached, no further depression can be produced even with a 1OOO-
fold excess of phosphate. Although the depression by phosphate is the
strongest of all anions tested, very similar curves can be obtained with
sulfate (Fig. 6 ) , nitrate ( P l ) , and oxalate (Fig. 7 ) . Zettner and Seligson
(23) also investigated the combined action of anions. As anticipated,
phosphate-sulfate mixtures did not produce absorption depression be-
yond the expected maximal phosphate interference; an additive effect,
however, was demonstrable with anion concentrations lower than those
producing maximum depression with phosphate or sulfate alone. The
behavior of phosphate-oxalate and sulfate-oxalate mixtures is similar. Ex-
periments were also carried out with a chelator, disodium ethylenedi-
aminetetraacetic acid (EDTA) (Fig. 8 ) . A sharp depression is seen at
low EDTA concentrations, but after the point of maximum depression is
reached the curve again rises moderately. An interesting finding is the
fact that proteins, made calcium-free by ion exchange, are capable of a
similar characteristic absorption depression at surprisingly low concen-
trations, as shown in Fig. 9. At protein concentrations higher than 100
mg "/., however, absorption depression becomes progressively less, and
above 1000 mg "/. protein the calcium signal is even enhanced.
The mechanism of the anionic depression has been the subject of many
investigations. It is generally believed that as the aerosol droplet dries
up in the preheating zone or in the flame itself, a compound is formed
between the metal and the anion which interferes with or retards the
 ATOMIC ABSORPTION SPECTROSCOPY 31

production of the vapor of free, neutral atoms, Since the dissociation of

such compounds is heat and time dependent, the magnitude of the de-
pressive effect should be related to these two variables. This reasoning
is well supported by experimental evidence. Anionic depression is most
pronounced in cooler flames and anions forming salts with higher melt-
ing points are the strongest depressors ( M 6 ) . For instance, calcium
phosphate with a melting point of 1670°C,when introduced into a coal
gas-air flame of 18OO0C,will melt slowly; even at the temperature of the
acetylene-air flame, 220O0C,disintegration of this compound will not be
rapid enough to be complete near the base of the flame. Dissociation,

o ' I ~ ~ 1 320 2 650

0 1300 2600
WiOTEIN m g %
FIG. 9. The depressive effect of protein on calcium absorption. All samples con-
tained 1 mg % calcium. The protein used was made calcium-free by ion exchange.
Maximum depression is reached at approximately 80 mg % protein. Higher protein
concentrations enhance absorption (from reference (23)by permission).

however, progresses as the particle travels upwards through the flame

and it has, indeed, been shown, that the anionic depression is smaller or
absent in the top of the flame (A4, F8).Vaporization of the aerosol par-
ticles is facilitated by decreasing their size. Filcek (F5) was able to
abolish the depressive effect of phosphate on calcium almost completely
by certain changes in the atomizer design, resulting in greater fineness
of the aerosol. The size of the dry particles formed from the drying drop-
lets in the aerosol, however, does not depend only on droplet size but
also on the concentration of solutes. By lowering the concentration of
calcium phosphate solutions, Gibson et al. ( G 3 ) were able to show that
a minimum concentration can be reached at which no anionic depression
is seen. The critical maximum particle weight at which complete and
sufficiently fast disintegration in the flame is possible was calculated at
6x gram calcium phosphate.
32 ALFRED ZE’ITNER

If the test element and the depressant are sprayed separately into the
flame, as was done by Alkemade and Voorhuis (A4, AS) and Fukushima
(F8, F9), the depressive effect is not seen. This indicates that the rela-
tively undissociable compound between analyte and depressant must be
formed in the aerosol phase or while the droplets dry up. One wonders
then, however, about the mechanism of the depressive action of organic
substances such as oxalate, EDTA, and proteins that are capable of bind-
ing cations (23). These substances must certainly disintegrate rapidly
in the base of the flame if not already in the preheating zone, certainly
below the melting temperature of calcium chloride, 770°C; and yet the
chloride anion has no depressive effect while the oxalate, EDTA, and
protein are characteristic interfering agents. It may be that the organic
particle before complete disintegration in the flame first undergoes some
“charring” process during which release of the metal atoms is delayed.
It is difficult to explain, however, the seemingly contradictory finding
that low protein concentrations are more depressive than high ones.
Clearly, further investigations are needed in this area.

5.2.2. Matrix Interferences

Various interferences may derive from changes in the composition of
solution and flame matrices. The simplest matrix is encountered when
an aqueous solution containing the analyte only is aspirated. In biological
work this may very rarely be the case.
Relatively large increases of the viscosity of the sample are necessary
to lead to reduced vaporization, but the effect is usually combined with
changes in the surface tension. If the length of the aspirating capillary
of the burner as used by the author (23,24) is increased, very little
change in the absorption signal is seen although sample flow is reduced.
For instance, an increase of the capillary length from 3 to 25 inches re-
sulted in a gradual absorption change of only 2%, the longer capillary
producing the higher absorption reading (Z2). The results with other
atomizers are not necessarily the same (S8). Longer aspirating capil-
laries may be employed with advantage when working with critically
small samples. Absorption readings become lower only when the capil-
lary reaches a length just short of the point when the aspirating pressure
is insufficient to aspirate any fluid at all.
The use of organic solvents, a significant change in solution matrix,
has been discussed above.
Where discharge atomizers are used, the presence of acids and salts in
the sample interferes with the evaporation of the aerosol (P4). The
 ATOMIC ABSORPTION SPECTROSCOPY 33

mild depression of calcium absorption by sodium chloride (23)in phys-

iological concentrations may be related to this.
When solutions of high salt content (2% or higher) are aspirated, the
salt particles formed from the aerosol are of sufficient size to pass through
the flame without disintegrating (W15). These particles are capable of
scattering the light from the hollow cathode tube, which will show up
in the measurement as erroneously high absorption.
The opacity of flames below 2000 A, of course, can be considered a
matrix effect and should be mentioned here as an obstacle in determin-
ing elements with resonance lines in the vacuum ultraviolet. Organic
solvents are capable of imparting to the flame additional absorptive prop-
erties in the ultraviolet.

5.3. INTERFERENCE
CONTROL
There are several different possibilities to circumvent, compensate for,
or suppress interferences.
5.3.1. Separation of the Analyte from the Solution Matrix
This method with the attractive advantage that the test element at
once can be isolated from all interferences, of course, is fully realized in
the solvent extraction technique, although the latter appears to have
been employed mainly for reasons of sensitivity problems. Separation by
precipitation has been applied in sample preparations for atomic absorp-
tion spectroscopy ( W9, 23, ZA),but the presence of an anion in the pre-
cipitate capable of binding the cation again poses the problem of anionic
suppression. It can be said in general that very accurate results are ob-
tained with this method but where a great number of analyses is in-
volved it may prove too cumbersome. Hinson (H5) described the re-
moval of depressing anions by an ion-exchange technique. Leyton (L6)
suggested the use of ion exchange to remove calcium by absorption onto
a resin followed by elution for analysis.

5.3.2. Addition in Excess

This method used extensively in flame photometry is based on the
principle that the interfering agents are added in equal but large amounts
to standards and samples alike, and that all determinations are carried
out in their presence. The curves obtained with the anionic depressors
(Section 5.2.1) indicate that this approach could be used even in cases
where the approximate concentration of the interfering agent in the
samples is not known. In practice, however, sensitivity is considerably
34 ALFRED ZE'ITNER

curtailed and working curves appear to be less linear. Small interferences

like that of sodium on calcium absorption, can easily be compensated for
by this method.
5.3.3. Protective Chelation
If a strong chelator is added to a sample containing the test element
and anionic depressors, the metal is preferentially bound by the former
and thereby kept from interacting with the anion in solution or in the
aerosol phase. When aspirated, the metal is released from the disintegrat-
ing chelator in the flame where the anions are incapable of acting upon

2 4 6 8 10
P rng %
FIG. 10. The addition of 0.65% calcium-free protein to 1 mg "/o calcium solu-
tions reduces the depressive effect of phosphate on atomic absorption of calcium
(compare with Fig. 5 ) . Phosphate concentrations given as phosphorus (from refer-
ence (23)by permission).

the metal. This method was applied successfully in emission (W4) and
absorption (W9, W10, 23). It should be kept in mind, however, that
chelators themselves act as depressants (see Fig. 8). The troublesome
clogging of the burner slot with carbonaceous crusts, whenever high con-
centrations of a large molecular organic compound are aspirated, could
make this method impractical for prolonged use.
There is considerable evidence that the concept of protective chela-
tion may not be a correct one. Far greater concentrations of EDTA than
molar unity between EDTA and calcium are needed for full protection
(W4). Furthermore, the action of EDTA is independent of the pH of the
solution but dependent on its cation, the sodium salt being more active
than the ammonium salt. Baker and Garton (B2) reasoned that the dry
particles deriving from the aerosol consist mainly of a matrix of EDTA
 ATOMIC ABSORPTION SPECTROSCOPY 35

in which calcium or other solution constituents are evenly dispersed. In

the flame the organic matrix disintegrates rapidly, and when the metal is
released the production of the metallic vapor is greatly accelerated by
the high dispersal of the cation in the original particle. Thus, the effect
of EDTA in abolishing phosphate depression would not be related to its
chelating properties, but simply to the formation of a bulky matrix easily
decomposed in the flame. This reasoning is supported by an earlier find-
ing that with very fine aerosols the anionic depression is markedly re-
duced (A4). In order to prove their assumption, Baker and Carton re-
placed EDTA by sodium chloride or sucrose in similar concentrations
(1-2%), both of which were as effective as the chelator in abolishing
phosphate depression. Zettner and Seligson (23) showed that the addi-
tion of calcium-free protein to calcium phosphate solutions diminished
the phosphate effect significantly, although absorption depression was
not completely eliminated ( Fig. 10).
5.3.4. Competitive Cation Technique
Following the observation of Mitchell and Robertson (M9)that the
anionic depression of calcium emission in the flame could be abolished
by the addition of strontium, many other cations were found to have
similar effects. Lanthanum, neodymium, samarium, yttrium, magnesium,
beryllium, barium, scandium, iron, and other cations are also capable of
releasing calcium and other metals either completely or partially from the
depressive action of anions (D14, W6, Yl). It is thought that the releas-
ing or protecting mechanism depends on the competition between the
cations for the anionic depressants during salt formation as the droplets
of the aerosol evaporate. In order to suppress the anions completely, that
is, to keep them from interacting with the test element, the releasing cat-
ion has to be added in large excess over the concentration of both, of
the metal to be determined and of the interfering anion. Since the de-
pressive action of the anions does not originate from compound forma-
tion in the flame, but rather in the aerosol (A4, F8, F9),the protective
action of competitive cations cannot be due to the refractoriness of their
compounds in the flame, but must depend on other factors such as their
concentration in the solution and the soIubiIity of the salts they form
with the depressant anions. For instance, if an excess of lanthanum chlo-
ride is used to control the depressive effect of phosphate on calcium (Fig.
I l ) , the drying salt particles forming from the aerosol will consist of
lanthanum chloride, lanthanum phosphate, and calcium chloride, the
latter being readily dissociable in the flame. The true competitive nature
of this phenomenon is discussed by Dinnin (D14).
36 ALFRED ZETTNER

Where applicable, the use of competitive cations for the control of

anionic depressors appears to represent the method of choice. No spec-
tral interference arises in atomic absorption from the addition of another
cation, an objection often raised in emission. The concentrations of the
added salt required for full anion control usually are less than 1%, a salt
level well below that at which light scattering is observed. When work-
ing with serum, denaturation or precipitation of proteins may occur from
the addition of high concentrations of lanthanum chloride or other salts,
and the concomitant changes in solution properties should be taken into

2z 0.3.
a
m
a
0
v)
30.2-

0.1' i i j 4 6 i l e
? mg %
FIG. 11. The protective action of 0.5% lanthanum chloride on atomic absorption
of 1 mg % calcium against the depressive effect of phosphate. Compare with
Fig. 4 (from reference (23) by permission).

consideration ( Fig. 12). Lanthanum chloride causes a true enhancement

of calcium absorption (Fig. 13), a phenomenon not entirely explained
but possibly related to the reduction of calcium oxidation in the flame
(Z2).
5.4. INDIVIDUAL ELEMENTS
In the following subsections the application of atomic absorption spec-
troscopy to the determination of the more important elements of biolog-
ical and clinicaI interest is presented, and special problems and interfer-
ences encountered with individual elements are discussed in detail. The
resonance lines given at the beginning of each subsection are those show-
ing greatest absorption, although many elements possess several reso-
nance lines that can be used in analysis. The sensitivity limits quoted are
the lowest reported in the literature, usually defined as that concentration
of the test element in aqueous solution which produces 1% absorption.
The reproducibility of results by most atomic absorption techniques lies
 ATOMIC ABSORPTION SPECTROSCOPY 37

0.5.

0.4.
U
0

24681012H416
CALCIUM mg %
FIG. 12. Protein in the presence of 0.5% lanthanum chloride slightly depresses
atomic absorption of calcium: -0-0- 0.1% protein; 0.4% protein;
-A-A- 0.8% protein. Calcium concentrations of samples actually aspirated were
only one tenth of those given on the abscissa (from reference (23)by permission).

0.6

0.5.

2 4 6 8 10 12 14 16
CALCIUM m g %
FIG. 13, Enhancement of calcium absorption by lanthanum chloride: ---
aqueous calcium standards; -0-0- aqueous calcium standards with 0.5% Ian-
thanum chloride (upper curoe). ( Calcium concentrations of samples actually aspi-
rated were only one tenth of those given on the abscissa (from reference (23) by
permission ) .
38 ALFRED ZETTNER

between 1and 2% but in many cases is better than 1%.Where available,

graphs of typical working curves are shown for some of the individual
elements.
5.4.1. Sodium
Resonance Line: 5890 A
Sensitivity Limit: 0.03 ppm ( G 2 )
The determination of sodium by atomic absorption spectroscopy has
been applied successfully by several workers using a variety of equip-
ment. The first element to be determined by Alkemade and Milatz ( M ) ,
in fact, was sodium. While sensitivity in emission is slightly higher for
this metal than sensitivity in absorption, sodium still counts as one of the
most sensitive elements in atomic absorption spectroscopy. The absence
of any spectra1 interference (P3) and the relative freedom from other
interferences appear to offer promising advantages of absorption over
emission also for this element.
Sodium is completely atomized in the flame and since the higher en-
ergies of hot flames are not needed for excitation, a cool flame such as the
coal gas-air flame is sufficient and perhaps superior because the appre-
ciable ionization of sodium in the hotter flames is avoided.
Most workers used a vapor discharge lamp as the sharp-line source
but sodium hollow cathode tubes are also available. The relatively short
lifetime of the latter owing to the high volatility of sodium and its rapid
loss from the cathode is, however, a limiting factor. Since both light
sources at higher currents show considerable line broadening due to self-
reversal, most workers recommend a low operating current. Because of
the strong emission of these light sources at the yellow sodium doublet
and the absence of other strong lines, wavelength selection with filters
has been shown to be sufficient (M2). In the preparation of tissues for
analysis, the loss of sodium and potassium during dry-ashing has to be
considered ( G8).
David (D5) employed an air-acetylene flame for the determination of
sodium in soil extracts with a detection limit of 0.05 ppm. No effect was
seen from the presence of phosphate, aluminum, sulfate, and silicate ex-
ceeding the sodium level 100-200-fold.
Malmstadt and Chambers (M2), using an air-propane flame, obtained
high accuracy by their standard addition method in the 1-100 ppm range
with their nullpoint instrument. It is stated that a small decrease of so-
dium absorption was found when potassium equaled or exceeded the so-
dium concentration, and that depression was also seen with high con-
 ATOMIC ABSORPTION SPECTROSCOPY 39

centrations of hydrochloric and sulfuric acids. Samples and standards

contained 2S% isopropyl alcohol.
Robinson ( R l ) worked with an oxy-hydrogen flame in which inter-
ferences appeared to be less than in other flames, as no interference was
seen from a 100500-fold excess of potassium or lithium.
Willis ( W l l ) , employing a sodium hollow cathode lamp, determined
sodium in blood serum. With a 10-cm long flame, sensitivity was so high
as to make the necessary dilution (500-fold) impractical and subject to
contamination. He reduced sensitivity simply in shortening the absorp-
tion path by turning the flame 90 degrees, but he also showed that the
use of another, much weaker resonance line at 3302 A reduced dilution
requirements to only 10-fold. No interferences were encountered from
any of the serum constituents.
Herrmann and Lang (H3) studied various atomizers and recorded
best results with a laboratory-built high pressure vaporizer. No ioniza-
tion interference was seen in an air-propane flame and calibration curves
were straight from 1to 10 mg sodium per liter. Determinations were per-
formed on serum diluted 1:20-1:200 and results agreed well with those
concurrently obtained by emission flame photometry.
5.4.2. Potassium
Resonance Line: 7665 A
Sensitivity Limit: 0.03 ppm (G2)
Some of the workers who published absorption methods for sodium
also include data on potassium. While certain differences exist between
sodium and potassium in regard to sensitivity and interferences, the
equipment used by some authors was of the same simplicity for both.
Filters and barrier-layer photocells suffice, but when using photomulti-
plier tubes their reduced sensitivity in the red, where the potassium reso-
nance line is located, has to be taken into account.
Potassium is one of the more easily ionized metals and the type of
flame used will be of even greater influence than in sodium work. Since
the degree of ionization depends also on other solution constituents, i.e.,
alkali metals, significant interference from sodium can be expected. The
relative enhancement of potassium emission and absorption is shown by
Baker and Garton ( B2). Atomization of potassium in the flame, however,
is not only reduced by ionization but also by compound formation, nota-
bly hydroxide.
David ( D 5 ) determined potassium in the air-acetylene flame in ammo-
nium chloride extracts of soil and found no interference from a 40-fold
excess of phosphate, aluminum, sulfate, and silicate,
40 ALFRED ZETTNER

Malmstadt and Chambers ( M 2 ) with their nullpoint instrument and

standard addition method obtained high accuracies; they state that no
interference was seen from an excess of sodium. The latter is difficult to
understand since the degree of mutual ionization depression between
alkali metals at a particular flame temperature can be predicted (F6),
and has been experimentally demonstrated in absorption with flames of
comparable or even lower temperature (B2, W l l ) . On the other hand,
the use of the standard addition method may have minimized the sodium
interference and, since the mutual enhancement effect in absorption
seems to be smaller than in emission (B2), it may have been negligible
in the measurements.
Willis ( W l l ) , using a potassium hollow cathode tube instead of the
commonly employed discharge lamp, determined potassium in blood
serum. At the 1:50 dilution no interference was encountered from cal-
cium, magnesium, and phosphate at serum levels, but sodium gave a
small enhancement. The sodium interference was controlled by the ad-
dition in excess of sodium chloride or of the disodium salt of EDTA to
samples and standards alike.
Herrmann and Lang (H3) also determined potassium in blood serum.
Calibration curves were straight from 1 to 100 mg sodium per liter. One
wonders about the sensitivity of their technique since it is stated that the
serum dilution used was 1:5.
5.4.3. Lithium
Resonance Line: 6708 A
Sensitivity Limit: 0.03 ppm (G2)
Lithium ranks with the easily excited metals and is also distinguished
by high sensitivity in absorption. Alkemade and Milatz (A2) determined
lithium by an absorption flame photometric method, but instead of em-
ploying a lamp as light source, a second flame was used into which lith-
ium was aspirated. The radiation from this exciter flame was passed
through the second flame functioning as absorption cell.
Lithium is also one of the very few metals to form isotopes with sig-
nificantly shifted resonance lines, the difference amounting to 0.15 A.
Zaidel and Korennoi, as quoted by Gilbert ( G 6 ) ,were able to exploit the
isotopic shift of the resonance line in the analysis of an isotopic mixture
of Lie and Li7. With a Li7 hollow cathode lamp, Li7 was determined in
solutions containing both isotopes when aspirated into an acetylene
flame.
Manning and Slavin (M3)also reported lithium isotope analysis by
atomic absorption. The emission of an open, lithium-containing oxygen-
 ATOMIC ABSORPTION SPECTROSCOPY 41

hydrogen flame was used as line source. Despite considerable line broad-
ening in the flame, measurement of Li6 in the presence of a large amount
of Li7 was possible, although sensitivity was less than that reported for
a hollow cathode source (Fig. 14). The spectrum of lithium at 6708 A
consists of a doublet whose lines are separated by 0.15 A. Since the iso-
topic spectral shift between Lia and Li7 is also 0.15 A, one line of the Lia
doublet is superimposed on one of the doublet of Li7. This represents
true spectral interference, a very rare occurrence in atomic absorption.

/ Hollow cathode

I 2 3 4 5 6 7 8 9 10
ppm LI
FIG. 14. Absorbance-concentration curves obtained with aqueous lithium solu-
tions. Note the difference in sensitivities depending on the type of line source (from
reference ( M3) by courtesy Perkin-Elmer Corporation).

The difficulty was overcome by adding Li7 in excess to the solutions,

whereby one line of the Lis doublet was virtually completely absorbed
and measurements could be carried out on the other line of the doublet.

5.4.4. Calcium
Resonance Line: 4227 A
Sensitivity Limit: 0.06 ppm (23)
The development of fast and accurate procedures for the determina-
tion of calcium in biological materials represents one of the important
early achievements of atomic absorption spectroscopy. The difEculties en-
countered with calcium in emission flame photometry are well known
(D11, L6, S6, SlO), but spectral interferences and extreme dependency
on flame temperature, serious obstacles in emission, are either nonexistent
or of lower importance in absorption. Chemical interferences, however,
42 ALFRED ZETTNER

are equally severe and it is for this reason that calcium is one of the
most difficult elements to deal with in atomic absorption spectroscopy.
The choice of flame appears to be important and best results were ob-
tained with the air-acetylene flame (Fig. 15). For optimal sensitivity, it
is necessary to work with a fuel-rich, reducing flame, because of consid-
erable oxide and hydroxide formation in lean flames, an effect even more
pronounced with other elements (S4). The concentration of neutral cal-
cium atoms is greatest in a narrow, clear zone of the flame just above

0.5.

0.4
W
0

2 4 6 8 10 12 14 16
CALCIUM mg %
FIG. 15. Typical absorbance-concentration curve obtained with aqueous calcium
solutions in 10-cm air-acetylene flame. Note the slight deviation from linearity.
Actual calcium concentrations of standards aspirated were only one tenth of those
given on abscissa (from reference (23)by permission).

the luminous base cone (see Fig. 2 ) , and it is recommended that all
measurements be performed in this zone despite the greater interference
from anionic depression. A similar distribution of calcium atoms is ob-
served in the air-coal gas flame, but while sensitivity is lower, the anionic
effects are even greater and more difficult to overcome than in the air-
acetylene flame. If possible, the use of the air-coal gas flame for calcium
should be avoided, unless the samples are free of anionic depressors.
David (D3) determined calcium in plant materials, employing an air-
acetylene flame. After wet-ashing of the samples, magnesium and sulfuric
acid were added to overcome effects of anions. It was demonstrated that
high concentrations of phosphate, aluminum, and silicate were com-
pletely controlled. In soil samples David (D5) determined calcium by
extraction with ammonium chloride, but used strontium and lanthanum
 ATOMIC ABSORPTION SPECTROSCOPY 43

chloride to suppress the effect of phosphate, aluminum, sulfate, and sili-

cate. Methods for calcium in animal feeds and tissues are described by
Parker ( P l ) . A strong depressive effect of nitrate on calcium absorption
is shown.
Willis (W8, W9) studied the behavior of calcium in various flames and
found the air-acetylene flame most suitable for absorption work. He in-
vestigated the influence of phosphate, sodium, and protein and found
the expected anionic depression by phosphate, but ascribed a mild en-
hancing effect to sodium and protein, the latter also reducing phosphate
depression. Interference control was obtained by dilution of the sample
with strontium or lanthanum chloride, or the sodium salt of EDTA. The
results were related to those obtained by the Clark-Collip method and to
analysis of calcium oxalate precipitates and deproteinized samples by the
atomic absorption method, Although acceptable results could be obtained
by all modes of sample preparation, deproteinization and dilution with
strontium gave the most accurate values. The results by atomic absorp-
tion spectroscopy generally were 2% higher than those by the Clark-
Collip method, which wa's related to incomplete precipitation of calcium
by oxalate. Willis (W13) also reported the determination of calcium in
urine. Because of the occasionally very high phosphate content, high con-
centrations of lanthanum or strontium chloride had to be used in the
diluent. The present writer feels, however, that the composition of urine
is too variable in respect to anions, sodium, and protein to permit a simple
dilution technique. Separation of calcium by oxalate precipitation is pref-
erable wherever accurate results are desired,
Newbrun (N2) reported the determination of calcium in saliva by
atomic absorption. When samples were diluted with water only, results
were obviously too low, but with strontium, lanthanum, or EDTA the
values agreed with those obtained by oxalate permanganate titration.
Calcium was higher in the excretion from the submaxillary glands than
in that from the parotid. Depletion of saliva calcium content was not
seen after 1Sminute stimulation.
Decker et al. (D13) analyzed cerebrospinal fluid for calcium by
atomic absorption. Excellent results are claimed on samples diluted
1:20 with water only, but anionic depressors are not mentioned.
Zettner and Seligson (23) conducted an extensive study of calcium
interferences deriving from serum constituents and other substances.
In the air-acetylene flame, no effect was seen from excess concentrations
of the ions of potassium, ammonium, magnesium, chloride, bicarbonate,
and hydroxide. Phosphate, sulfate, oxalate, and EDTA acted as strong
anionic depressors. Sodium caused a small but distinct depression of
44 ALFRED ZETTNER

about 3%. The effect of calcium-free protein was seen to be rather

complicated, low concentrations acting like anionic depressors, but
concentrations above 1% enhancing calcium absorption. The combined
action of interfering agents was also investigated. Mixtures of phosphate
and sulfate showed an additive effect only at concentrations lower than
those producing maximum depression with phosphate or sulfate alone.
Mixing of phosphate and protein resulted in reduction of the expected
depression, but the degree of depression progressed with phosphate
concentration ( see Fig. 10). Interference control was achieved with the
competitive cation technique and by incorporating sodium and calcium-
free protein into the calibrating solutions. With a diluent consisting of
lanthanum chloride, butanol, octanol, and hydrochloric acid, accurate
calcium determinations could be carried out on serum samples directly
diluted 1 : l O or 1:20. Proof of method was obtained from comparisons
with the Clark-Collip method, from ashings and recovery experiments,
but mainly from oxalate precipitation studies. It was possible to show
that calcium precipitation at pH 4.2 in the cold is virtually complete,
contrary to statements often repeated in the literature. When the super-
natant was evaporated, dry-ashed, and taken up in a small volume of
diluent, the amount of calcium found was only slightly higher than that
expected from the solubility product of calcium oxalate.
5.4.5. Magnesium
Resonance Line: 2852 A
Sensitivity Limit: 0.005 ppm (24)
At least sixteen papers have appeared up to the present time on the
determination of magnesium by atomic absorption spectroscopy and
nine of these deal with biological materials. This pronounced interest
undoubtedly derives from the fact that while wet chemical ( H 7 ) and
flame emission methods ( Al, D11, F4, M10) are unsatisfactory in many
respects, atomic absorption allows for rapid and accurate analysis of
magnesium with sensitivities unexcelled with other elements ( Fig. 16).
Magnesium, like calcium, i s subject to the effect of anions, although
to a lesser degree. The most serious interference derives from refractory
acidic oxides formed in the flame from a number of elements, particularly
aluminum and silicon. The effect of phosphate and sulfate is much less
marked than with calcium (Fig. 17), but if aluminum or silicon is also
present in the solutions, magnesium depression is much more severe
than with either of these interfering agents alone. These interferences
can be overcome by the addition of strontium, lanthanum, or calcium.
Leithe and Hofer (L4, L5) showed that magnesium could be determined
 ATOMIC ABSORPTION SPECTROSCOPY 45

in various industrial materials with a magnesium-aluminum ratio of

1:1OOO, if calcium was added in excess of 2.5 times the amount of
aluminum. No interferences were seen from iron, copper, manganese,

0.6 -
0.5-
u
o 0.4 -
z
a
m
0.3 -
v)
m
a0.2-

Mg mg %
FIG. Typical absorbance-concentration curve obtained with aqueous ma -
nesium solutions in IO-cm air-acetylene flame. Note the slight deviation from lin-
earity (from reference 24).

FIG. 17. The depressive effect of phosphate on magnesium absorption (compare

with Fig. 4 ) . Phosphate concentrations given as phosphorus (from reference 2 4 ) .

chromium, and zinc. Phosphate and sulfate effects were marked in the
air-coal gas flame, but negligible in the air-acetylene flame. Andrew and
Nichols (A14) demonstrated that the aluminum and silicon interference
was overcome by the addition of nickel, and thus were able to analyze
46 ALFRED ZETI'NER

nickel alloys for magnesium. Strontium was used by Belcher and Bray
(B3) for the suppression of aluminum interference in the analysis of
iron for magnesium. Wallace ( W l ) was able to investigate aluminum
alloys by complexing the magnesium with 8-hydroxyquinoline.
Allan (A6) in determining magnesium in plant materials encountered
interference from aluminum. The addition of 10% sodium acetate had no
effect on absorption. David analyzed plants (D2) and soils (D5) and
used lanthanum and strontium for the suppression of interferences.
Wunsch and Teicher ( W 2 l ) estimated magnesium in plant substance
after dry- and wet-ashing. In the propane flame, depression from phos-
phate, sulfate, and aluminum was seen, but in the acetylene flame only
aluminum had such an effect.
Several papers deal with magnesium determination in blood and
urine. Willis ( WlO) analyzed serum in the air-acetylene flame and found
no effect from the presence of sodium, potassium, calcium, or phosphate,
but states that an enhancement was seen in serum diluted with water
only, probably due to serum proteins. This interference was controlled
by addition of strontium or EDTA. Sensitivities were the same in the
coal gas-air and air-acetylene flame, indicating complete atomization of
magnesium. In urine (W13) no interference was encountered and deter-
minations were performed on samples directly diluted with water.
Dawson and Heaton ( 0 1 0 ) determined magnesium in plasma and
urine. A water-cooled burner was used with an air-acetylene flame.
Good agreement was found with an ammonium phosphate precipitation
method when allowance was made for the loss of magnesium during
precipitation.
Stewart et al. (S8) estimated magnesium in serum and urine. Of four
different methods of sample preparation ( i.e., wet-ashing, deproteiniza-
tion, simple dilution with water, and dilution with hydrochloric acid),
deproteinization with trichloroacetic acid was found to be most satis-
factory. No interference was seen from sodium, potassium, or phosphate,
but sulfate produced depression. With protein a 6% decrease in the
apparent magnesium concentration was seen. Calcium and sulfate were
added to standards and samples to control sulfate depression,
Horn and Latner ( H 6 ) analyzed blood and urine, using an air-
propane flame. No interferences were encountered from other serum
constituents. The fuel mixture of the flame was found to be critical.
Decker et aZ. (D13) also include data on magnesium in cerebrospinal
fluid.
Zettner and Seligson (24)determined magnesium in serum, using an
air-acetylene flame. A study of interferences showed that phosphate and
 ATOMIC ABSORPTION SPECXROSCOPY 47

sulfate produced a slight depression at magnesium levels comparable to

those in serum diluted 1:SO. Lanthanum chloride was used for inter-
ference control. Occasional lanthanum samples from commercial sources
were inapplicable because of high magnesium contamination.
Cheek et al. ( C l ) used an atomic absorption spectroscopy method to
analyze rat skeletal and cardiac muscle for magnesium. Griffith et al.
(G7) analyzed liver, kidney, muscle, heart, plasma, and bone of grow-
ing chicks for magnesium levels following the addition of fluoride to
the diet. Only the bone magnesium content was found to be increased.
Dietary supplement of magnesium, however, increased magnesium in
plasma, bone, liver, and kidney, but magnesium in muscle and heart
was constant regardless of dietary intakes. Parker (P1) described
methods for the determination of magnesium in animal feeds and tissues.
Only minor effects were seen from hydrochloric and nitric acids.
5.4.6. Copper
Resonance Line: 3247 A
Sensitivity Limit: 0.1 ppm (A12)
Copper is distinguished by a complex spectrum and although the
3247-A Iine is the most sensitive resonance line, absorption can also be
observed on several other lines. Copper is completely atomized in the
flame. Allan (A10) made use of this fact in his study of organic solvents,
in which he could show that sensitivity increase was not due to better
atomization in the flame but related to an increase in the size of the
sample reaching the flame. For solvent extraction, the metal was com-
plexed with ammonium pyrrolidine dithiocarbamate. Copper is re-
markably free of interferences but the complex spectrum requires high
resolution monochromators for adequate wavelength separation. The
metal itself may become an interference, as in the instance of Gidley's
work on zinc (G4) during which some enhancement was encountered
when hydrochloric acid was sprayed. This was later shown (G5) to be
due to absorption by copper atoms liberated from the brass burner by
the acid. Since a brass hollow cathode tube was used, also emitting a
copper spectrum, absorption took place on weak copper lines at 2024,
2165, and 2182 A, in the vicinity of the zinc resonance line at 2138 A.
This interference was eliminated by neutralizing the samples or by using
a steel burner.
Details of copper determination are also discussed by Menzies (M6)
and Strasheim et al. (S11). Elwell and Gidley (E2) examined the effect
of many elements including silicon and found no interference on the
absorption of copper.
48 ALFRED ZETTNER

Allan (AQ) described the analysis of soils, fertilizers, and plants by

acid extraction, or chelation with ammonium pyrrolidine dithiocarba-
mate and subsequent solvent extraction. No interferences were en-
countered. A photographic technique for the selection of the most
suitable resonance lines is described.
Berman (B4) reported the clinical application of atomic absorption
spectroscopy in the determination of copper in blood, urine, and tissues.
After acid digestion of the samples, copper was complexed with EDTA
and diethylthiocarbamate and extracted into a ketone. Copper levels in
blood and urine from normal persons were compared to those found in
patients with neurological disorders.
Zeeman and Butler (21)determined copper in wines. Following acid
digestion and dry-ashing the samples were taken up in nitric acid and
aspirated. None of the elements found in the ash, when tested in-
dividually, had any effect on copper absorption, but when synthetic ash
solutions were investigated, containing a combination of the ash con-
stituents, absorption depression occurred. This effect had to be com-
pensated for in the calibrating solutions.
5.4.7. Manganese
Resonance Line: 2794.8 A
Sensitivity Limit:0.05 ppm (A12)
Manganese possesses a spectrum of considerable complexity with
multiple ground state lines and, therefore, absorption can be observed
on several lines, The fact that the strongest lines in emission are not
always the strongest in absorption is well documented for manganese
by Allan (A7) in his photographic study of emission and absorption
spectra of flames. In order to find the most sensitive line for absorption
measurement, Allan positioned a hollow cathode tube and a burner in
front of a spectrograph, the light beam traveling through the flame be-
fore passing into the entrance slit. In this way photographs of the emis-
sion spectrum of the hollow cathode tube could be obtained while
distilled water or a series of calibrating solutions of manganese were
aspirated into the flame. When the differences in line intensity of the
various spectrograms were related to the manganese content of the
calibrating solutions, quantification of absorption intensity was possible.
Emission spectrograms from the flame alone with the same set of
calibrating solutions were used to correct for flame emission on manga-
nese lines. It was found that manganese in the air-acetylene flame
emitted at 4030 A but not at 2800 A, while the intensity of absorption
at 2800 A exceeded 10-fold that observed at 4030 A. Absorption at
2800 A takes place at three lines of a multiplet of varying absorption
 ATOMIC ABSORPTION SPECTROSCOPY 49

intensity. Absorption is strongest at the 2794.8-A line and for optimal

sensitivity and pertinence of Beer’s Law, and the next line at 2798.3 A
has to be separated as much as possible by the monochromator.
Manganese is not completely atomized in the flame ( A l O ) , which is
reflected by the sensitivity differences in air-acetylene and air-coal gas
flames. Elwell and Gidley (E2) report that large excesses of sodium,
potassium, copper, lead, iron, nickel, tin, and zinc have no effect on
manganese absorption while severe depression was seen with silicon.
The moderate depression encountered with calcium, magnesium, alumi-
num, titanium, and zirconium is in contrast to interference studies of
Allan (A7), who found no effect from phosphate, calcium, and mag-
nesium. These differences in the influence of calcium and magnesium on
manganese absorption appear to be related to the different behavior of
these elements in the air-coal gas ( E 2 ) and air-acetylene flame (A7).
Only a few reports have appeared up to the present time on the
application of atomic absorption spectroscopy to the determination of
manganese in biological materials. Allan (A7) analyzed plants after
wet-ashing with nitric and perchloric acids, and a similar project includ-
ing a study of interferences was carried out by David (D9). A manganese
recovery experiment from human plasma is reported by Manning ( M2a).
Owing to the low manganese content of tissues and biological fluids,
considerable concentrating of materials by ashing or solvent extraction
will be necessary when determinations of this element by atomic ab-
sorption are considered. Hedge et al. ( H l ) studied manganese levels in
human heart muscle and blood serum by spectrographic analysis and
showed that following myocardial infarction serum manganese rose 2.5
times above the normal level (1microgram/100 ml serum). In view of
renewed biological interest in trace metals, it can be anticipated that the
determination of manganese by as fast and accurate a method as atomic
absorption spectroscopy may attain new significance.
5.4.8. lron
Resonance Line: 2483 A
Sensitivity Limit: 0.1 ppm (A12, G 2 )
Iron also has a very complex spectrum and most other points that
have been made in regard to manganese apply to this metal. The choice
of the most sensitive among the many absorption lines again was made
with the help of the photographic technique by Allan (A7). The strong-
est line is that at 2483.3 A at which sensitivity limits of 0.1 ppm have
been obtained, but the line at 3720 A still permits the detection of iron
at the 1 pprn level. Especially narrow slit width and high resolution
monochromators are necessary for optimal results, because the resonance
50 ALFRED ZETTNEn

line at 2483 A and the next line at 2488 A are of different absorption
intensity and, therefore, have to be resolved from each other. Emission
on these lines from iron in the air-acetylene flame is negligible.
Allan ( A 7 ) during his studies of ashed plant materials also investigated
interferences. Using an air-acetylene flame, sodium, potassium, calcium,
magnesium, and phosphate had no effect. In the air-coal gas flame, as
employed by Elwell and Gidley (E2), recoveries of iron were only
80-90% when the test solutions contained an excess of calcium, copper,
aluminum, titanium, and zirconium. With silicon added, iron recovery
was 26%. Owing to incomplete vaporization of iron in the flame, sensitiv-
ities attained are higher in the air-acetylene flame and lower in the
air-coal gas flame. Since iron is subject to oxidation in the flame, fuel-
rich flames are preferable.
David (D9) determined iron in plants and gives detailed instructions
for the treatment of plant samples, entailing acid digestion and filtration
before aspiration. The determination of iron in protein solutions is
briefly mentioned ( M2a ) .
Clinical uses of atomic absorption spectroscopy in the analysis for
iron have not yet been published and the low iron levels in most bio-
logical fluids may very well be limiting. An interesting clinical applica-
tion was demonstrated by Briscoe (B8), who determined by atomic
absorption spectroscopy iron excretion in the urine of a patient treated
with desferrioxamine for hemochromatosis.
The determination of iron in serum is of great clinical interest and,
despite the existence of sensitive colorimetric methods, a direct and
accurate method of the reliability of atomic absorption spectroscopy
would be desirable. Unfortunately, serum iron levels range near the
sensitivity limit of this method, and either very large samples and solvent
extraction, or a 100-fold increase of sensitivity, would be required. Useful
applications can be expected in the analysis of tissues of normally high
iron content and in hemochromatosis. Another important application is
anticipated in the determination of iron in hemoglobin, where atomic
absorption spectroscopy may be a welcome asset in hemoglobin stan-
dardization. Only 0.01 gram of hemoglobin of original sample would be
required to yield 5 ml of a solution with an iron content suitable for
the sensitivity range of the method.
5.4.9. Zinc
Resonance Line: 2139 A
Sensitivity Limit: 0.03 ppm (A12, C2)
Zinc exhibits very favorable characteristics in its determination by
 ATOMIC ABSORPTION SPECTROSCOPY 51

atomic absorption spectroscopy. Early workers showed its high sensitivity

( R 9 ) by this method and the limit of the detection has since been
lowered (A12, G2) considerably. This compares with extremely low
sensitivity of zinc in emission flame photometry in which the detection
limit is 500 ppm in an oxygen-hydrogen flame (D11).
Zinc is completely atomized in the cool air-coal gas flame, but in
hotter flames, where a weak emission continuum appears, some com-
pound formation takes place. For this reason highest sensitivities are
observed in the air-coal gas flame ( A l l , G2).
Zinc in atomic absorption spectroscopy is remarkably free from inter-
ferences as contrasted to the difficulties encountered in polarography or
with colorimetric methods ( M4). Gidley and Jones (G4,G5) studied
the influence of 27 elements and the only effect seen was a depression
with silicon. The absorption enhancement encountered by these authors
with haloid acids could be traced back to the attack of the brass burner
by the samples and to the use of a brass hollow cathode tube as zinc
line source. Methods for the determination of zinc in various metals and
alloys are described by these authors.
David (D2) as early as 1958 made use of the high sensitivity of zinc
in determining this metal in plants, following acid digestion with a sul-
furic and perchloric acid mixture and nitric acid. In the air-acetylene
flame he found no interference from other plant constituents, but mild
depression from sulfuric acid, which was easily compensated for by
treating samples and standards alike. Zinc values obtained by atomic
absorption were lower than those by polarography, possibly due to inter-
ferences in the latter technique. A similar study was performed by
Allan ( A l l ) , who estimated zinc in plants, soils, and fertilizers. No
interferences were encountered and where zinc concentrations in the
original sample were too low to be sufficiently concentrated by ashing,
solvent extraction was used.
Willis (W14) was able to determine zinc in untreated urine, the sam-
ples being directly aspirated into the flame. The zinc content of normal
urine was about 1 mg/liter. Reproducibility was better than that of
colorimetric methods.
Zeeman and Butler ( Z l ) determined zinc in a variety of wines.
Twenty-five-ml samples were ashed, and the residue dissolved in nitric
acid was aspirated. No interference was seen from any of the ash con-
stituents when investigated individually but, when combined, zinc ab-
sorption was depressed, resulting in a change in the slope of the calibra-
tion curves.
Parker (P1 ) described methods of sample preparation for the deter-
52 ALFRED ZETTNER

mination of zinc in animal tissues. Only slight influence on zinc absorp-

tion was seen from hydrochloric and nitric acids.

5.4.10. Mercury
Resonance Line: 2537 A
Sensitivity Limit: 5 ppm (A12)
Mercury is the only metal with a significant vapor pressure at room
temperature and the mercury vapor detector, a true atomic absorption
spectrophotometer ( W20 ), is based on this property. The sensitivity in
the flame is disappointingly low when compared with the detection
limits in the mercury vapor meter in which 10,OOO times higher sensi-
tivities are attained. The reason for this difference is not entirely clear,
but expansion of the flame gases, broadening of the resonance line at the
higher temperatures, and the shorter absorption path in the flame may
be responsible. The resonance line at 1849 A shows much stronger ab-
sorption than the line at 2537 A, but owing to the absorptive properties
of oxygen below 2000 A the shorter line can neither be used in the
common mercury vapor detector nor in the flame.
Lindstrom (L7) used a flame to volatilize mercury-containing solutions
but determined mercury levels in the exhaust gases which, after filtering
and cooling, were passed into a mercury vapor detector. A variety of
materials, including urine, were analyzed with the astounding sensitivity
limit of 0.001 ppm of mercury in the aspirated sample.
Jacobs et al. ( J l ) developed a method for the determination of mer-
cury in 0.1 ml blood. The specimen is digested in the cold and extracted
with dithizone. The extract is decomposed in an electrically heated
furnace and the mercury vapor is passed into a mercury vapor meter.
The method is said to detect mercury in the nanogram range.
Willis (W14) determined mercury in urine by chelation of the metaI
with ammonium pyrrolidine dithiocarbamate and extraction into methyl-
n-amyl ketone. He showed that recovery of mercury added was complete
and levels encountered in normal urines were 0.02 mg per liter.

5.4.11. Lead
Resonance Line: 2833.1 A
Sensitivity Limit: 0.3 pprn (A12)
Lead shows atomic resonance on several spectral lines, the strongest
absorption line being that at 2833.1 A. Again, maximum emission is ob-
served on different lines, that at 4058 A being the most sensitive. If a
lead hollow cathode lamp is used, it must be operated on low current
 ATOMIC ABSORPTION SPECTROSCOPY 53

to prevent rapid loss of the metal from the cathode. If the hollow cathode
is made of leaded brass, higher currents can be used and Elwell and
Gidley ( E l ) obtained 250 service hours from an argon-filled tube run
at 40 milliamperes. Despite broadening of the resonance line, best
sensitivity was found at the highest possible operating current, 50 mA.
The 2833 line, however, must be separated from lines at 2824 and
2840 A.
Robinson (R4) described the determination of lead in gasoline and
found no interference in the oxy-hydrogen flame from a 90-fold excess
of tin, sodium, bismuth, copper, zinc, chromium, iron, nickel. No effect
on lead absorption was seen when solutions of lead tetraethyl in iso-
octane were mixed with equal volumes of carbon disulfide, o-thiocresol,
diethylamine, and other organic sources of nitrogen and sulfur. Elwell
and Gidley ( E l ) determined lead in alloys and steel and state that no
interference, except a slight viscosity effect, was seen from excesses of
nitric, hydrochloric, perchloric, and fluoroboric acids, or from a 20-fold
excess of sodium, calcium, copper, magnesium, iron, nickel, tin, or zinc.
Remarkable is the fact that neither aluminum nor silicon caused ab-
sorption depression. An unexplained enhancement, however, was seen
when traces of lead were determined in 2% iron solutions in hydro-
chloric acid; light scattering due to the highly concentrated solutions
seems not to have been the cause, since no absorption was found at the
neighboring copper lines of the brass hollow cathode lamps.
Willis (W12, W14) reported the determination of lead in urine. Lead
was chelated with ammonium pyrrolidine dithiocarbamate and extracted
into methyl-n-amyl ketone. Urine samples of 50 ml had to be extracted
into 1.5 ml of the solvent to obtain sufficient concentration of the metal.
Differences in the efficiency of extraction from water and urine at various
pH values were noted. The availability of a reliable and simple method
such as the one described will be welcomed by medical laboratories,
but unfortunately the limited sensitivity requiring large original samples
makes the method less applicable for the determination of lead in blood.
Zeeman and Butler ( Z l ) analyzed wines for lead content. Wine
samples of 50 ml were ashed to dryness and the residue taken up in
2 ml nitric acid, effecting a 25-fold concentration. The samples were
aspirated into a propane-butane-air flame 8 inches in length, and a
simple optical system with two flame traversals was employed. The
authors showed that a certain amount of lead was lost in the ashing
process, and the loss was smallest with lead sulfate. Lead concentrations
in a variety of wines were 0.2-1 ppm.
54 ALFRED zE?TNER

5.4.12. Other Elements

Many other elements can be determined by atomic absorption spec-
troscopy but little has been published about applications in biological
materials,
Noble metals were determined by Lockyer and Hames (L9). Silver
shows a sensitivity limit of 0.1 ppm. When working with gold salts, the
deposition of gold in the burner may lead to an absorption decrease in
the flame.
Cadmium is characterized by high sensitivity (limit 0.03 ppm) ( G2)
and by complete atomization in the air-coal gas flame. Willis (W14)
showed that cadmium when added to urine could be determined by
aspirating the urine directly; normal urine levels were at the sensitivity
limit of the method, but applications of this technique are certainly
feasible in toxicological work. The recent demonstrations of a specifk
cadmium-containing protein (K2,K3) adds importance to the availa-
bility of an atomic absorption method for cadmium.
Chromium was determined by Williams et d. (W7) in animal feces
samples to study pasture intakes. In the air-acetylene flame the sensitivity
limit was 0.15 ppm. Of a variety of substances tested individually, only
calcium, silicate, and phosphate depressed chromium absorption. How-
ever, when interferences were studied following treatment of solutions
with phosphoric acid, manganese sulfate, and potassium bromide, de-
pression was caused by silicon and aluminum, but calcium and magne-
sium enhanced absorption. Calcium was also capable of aboIishing the
effect of silicon and aluminum.
The determination of cobalt and nickel is discussed by Allan (A8).
Sensitivities are considerably higher than those in emission. Applications
of absorption methods to molybdenum and strontium are reported by
David ( D4,D7, D8).
Atomic absorption spectra of vanadium, titanium, niobium, scandium,
yttrium, and rhenium in the fuel-rich oxy-acetylene flame and spectra of
the lanthanides were studied by Fassel and Mossotti (Fl, M11). The use
of the flame as a line source for the study of atomic absorption spectra
of europium, thulium, and ytterbium was shown by Skogerboe and
Woodriff ( S2).
Bismuth added to urine was recovered by Willis (W14) with solvent
extraction and determined by atomic absorption spectroscopy. An ab-
sorption interference rarely encountered in atomic absorption spectros-
copy was seen from the absorption of the 3068-A line of bismuth by
the OH radical in the air-coal gas flame.
 ATOMIC ABSORPTION SPECTROSCOPY 55

REFERENCES
A l . Alcock, N., MacIntyre, I., and Hadde, I., The determination of magnesium
in biological fluids and tissues by flame spectrophotometry. J . Clin. Pathol. 13,
506-510 (1960).
A2. Alkemade, C. T. J., and Milatz, J. M. W., A double-beam method of spectral
selection with flames. Appl. Sci. Res. Sect. B 4, 289-299 (1955).
A3. Alkemade, C. T. J., and Milatz, J. M. W., Double-beam method of spectral
selection with flames. J. Opt. SOC. Am. 45, 583-584 (1955).
A4. Alkemade, C. T. J., and Voorhuis, M. H., Zur Frage des Phosphoreinflusses
auf die Calciumemission in der F l a m e . Z. Anal. Chem. 163, 91-103 ( 1958).
A5. Alkemade, C. T. J., and Voorhuis, M. H., On the problem of calcium depres-
sion by phosphorus in flame photometry. Spectrochim. Acta 12, 394 (1958).
A6. Allan, J. E., Atomic-absorption spectrophotometry with special reference to
the determination of magnesium. Analyst 83, 468-471 (1958).
A7. Allan, J. E., The determination of iron and manganese by atomic absorption.
Spectrochim. Acta 15, 800-860 ( 1959).
A8. Allan, J. E., Determination of nickel and cobalt by atomic absorption. Nature
186, 1110 (1960).
A9. Allan, J. E., The determination of copper by atomic absorption spectrophotom-
etry. Spectrochim. Acta 17, 459-466 (1981).
A10. Allan, J, E., The use of organic solvents in atomic absorption spectrophotom-
etry. Spectrochim. Acta 17, 476-483 ( 1961).
A l l . Allan, J. E., The determination of zinc in agricultural materials by atomic
absorption spectrophotometry. Analyst 86, 530-534 ( 1981).
A12. Allan, J. E., Atomic absorption spectrophotometry absorption lines and
detection limits in the air-acetylene flame. Spectrochim. Acta 18, 259-263 (1962).
A13. Allan, J. E., A rcview of recent work in atomic absorption spectroscopy.
Spectrochim. Acta 18, 605-614 ( 1962).
A14. Andrew, T. H., and Nichols, P. N. H., The application of atomic absorption
to the rapid determination of magnesium in electronic nickel and nickel alloys.
Afiolyst 87, 25-31 ( 1962).
B1. Baker, C. A., The flame as a source of atoms. Analyst 85, 461 (1960).
B2. Baker, C. A., and Garton, F. W. J., A study of interferences in emission and
absorption flame photometry. U . K . At. Energy Authority R e p . AERE-R 3490
(1961).
B3. Belcher, C. B., and Bray, H. M., Determination of magnesium in iron by
atomic absorption spectrophotometry. A d . Chim. Acta 26, 322-325 ( 1962).
B4. Berman, E., An application of atomic absorption spectrophotometry in clinical
chemistry: Determination of copper in biological materials. Proc. Intern. Congr.
Clin. Chem., Sth, Detroit, 1963.
B5. Berry, J. W., Chappell, D. G., and Barnes, R. B., Improved method of flame
photometry. Znd. Eng. Chem. Anal. Ed. 18, 19-24 (1946).
B6. Bode, H., and Neumann, F., Untersuchungen uber disubsituierte Dithiocarba-
midate. 2.Anal. Chem. 172, 1-21 (1980).
56 ALFRED ZE?TNER

B7. Box, G. F., and Walsh, A., A simple atomic absorption spectrophotometer.
Spectrochim. Acta 16, 255-258 (1960).
B8. Briscoe, A,, Personal communication (1963).
B9. Buell, B. E., Use of organic solvents in limited area flame spectrometry. Anal.
Chem. 34, 635-640 (1982).
C1. Cheek, D. B., Graystone, J. E., Willis, J. B., and Holt, A. B., Studies on the
effect of triglycerides, glycerophosphate, phosphatidylethanolamine on skeletal and
cardiac muscle composition, Clin. Sci. 23, 169-179 ( 1962).
C2. Clinton, 0. E., A burner for atomic absorption spectrophotometry. Spectro-
chim. Acta 16, 985-988 (1960).
C3. Crosswhite, H. M., Dieke, G. H., and Legagneur, C. S., Hollow iron cathode
discharge as source for wavelength and intensity standards. I . Opt. SOC. Am. 45,
270-280 ( 1955).
D1. Davey, B. G., Modification of a spectrophotometer for atomic absorption
spectrophotometry. Spectsochim. Acta 19, 1319-1322 ( 1963).
D2. David, D. J., Determination of zinc and other elements in plants by atomic-
absorption spectroscopy. Analyst 83, 655-661 ( 1958).
D3. David, D. J., Determination of calcium in plant material by atomic-absorp-
tion spectrophotometry. Analyst 84, 536-545 ( 1959).
D4. David, D. J,, Atomic absorption spectrophotometric determination of molyb-
denum and strontium. Nature 187, 1109 (1960).
D5. David, D. J., The determination of exchangeable sodium, potassium, calcium,
and magnesium in soils by atomic-absorption spectrophotometry. Analyst 85, 495-
503 ( 1930).
D6. David, D. J., The application of atomic absorption to chemical analysis.
A d y s t 85, 779-791 (1960).
D7. David, D. J., The determination of molybdenum by atomic-absorption spectro-
photometry. Analyst 86, 730-740 ( 1961).
D8. David, D. J., Determination of strontium in biological materials and ex-
changeable strontium in soils by atomic-absorption spectrophotometry. Analyst 87,
576-585 ( 1962) .
D9. David, D. J., Atomic absorption spectrochemical analysis of plant materials
with particular reference to manganese and iron. Atomic Absorption Newsletter
No. 9. Perkin-Elmer COT., Norwalk, Conn. (December 1962).
D10. Dawson, J. B., and Heaton, F. W., The determination of magnesium in
biological materials by atomic absorption spectrophotometry. Bfochem. I . 80, 99-106
(1981).
D11. Dean, J. A., “Flame Photometry.” McGraw-Hill, New York, 1960.
D12. Dean, J. A., and Lady, J. H., Application of organic solvent extraction to
flame spectrophotometry. Determination of iron in nonferrous alloys. Anal. Chem.
27, 1533-1536 ( 1955).
D13. Decker, C. F., Aras, A., and Decker, L. E., Determination of calcium and
magnesium in human cerebrospinal fluid by atomic absorption spectroscopy. Abstr.
Am. Chem. SOC., Dio. Biol. Chem., Dee. 1963 p. 7A.
D14. Dinnin, J. I., Releasing effects in flame photometry. Determination of cal-
cium. A w l . Chem. 32, 1475-1480 (1960).
 ATOMIC ABSORPTION SPECTROSCOPY 57

D15. Dippel, W. A,, Bricker, C. E., and Furman, N. H., Flame photometric deter-
mination of phosphate. Anal. Chem. 28, 533-558 (1954).
E l . Elwell, W. T., and Gidley, J. A. F., The determination of lead in copper-base
alloys and steel by atomic absorption spectrophotometry. Anal. Chim. Acta 24, 71-78
(1981).
E2. Elwell, W. T., and Gidley, J. A. F., “Atomic-Absorption Spectrophotometry.”
Macmillan, New York, 1982.
F1. Fassel, V. A., and Mossotti, V. C., Atomic absorption spectra of vanadium,
titanium, niobium, scandium, yttrium, and rhenium. And. Chem. 35, 252-253 ( 1983).
F2. Fassel, V. A., Kniseley, R. N., and Mossotti, V. C., Progress in the atomic
absorption spectrometric determination of previously “unrealized” elements. Symp.
Recent Deuelopments in Aes. Methods and Instrumentation, National Institutes of
Health, Bethesda, Md. (October 1963).
F3. Fassel, V. A,, Myers, R. B., and Kniseley, R. N., Flame spectra of vanadium,
niobium, rhenium, titanium, molybdenum, and tungsten. Spectrochim. Acta 19,
1187-1194 (1983).
F4. Fawcett, J. K., and Wynn, V., The determination of magnesium in biological
materials by flame photometry. I . Clin. Pathol. 14, 403-409 ( 1981) .
F5. Filcek, M., Die Ausschaltung des Phosphateinflusses bei der flammenphoto-
metrischen Calciumbestimmung. Z. Pflanzenernaehr. Dueng. Bodenk. 85, 112-117
(1959).
F8. Foster, W. H., and Hume, D. N., Mutual cation interference effects in flame
photometry. Anal. Chem. 31, 2033-2038 ( 1959).
F7. Fristrom, R. M., The mechanism of combustion in flames. Chem. Eng. News
41, 150-180 (1963).
F8. Fukushima, S., Mechanism and elimination of interferences in flame photom-
etry. Mikrochim. Acta, pp. 598-818 (1959).
F9. Fukushima, S., Mechanism and elimination of interferences in flame photom-
etry. Mikrochim. Acta, pp. 332-343 (1900).
F10. Fuwa, K., and Vallee, B. L., The physical basis of analytical atomic absorp-
tion spectrometry. A w l . Chem. 35, 942-940 (1983).
G1. Gatehouse, B. M., and Walsh, A., Analysis of metallic samples by atomic
absorption spectroscopy. Spectrochim. Acta 16, 002-804 ( 1980).
G2. Gatehouse, B. M., and Willis, J. B., Performance of a simple atomic absorp-
tion spectrophotometer. Spectrochim. Acta 17, 710-718 ( 1981).
G3. Gibson, J. H., Grossman, W. E. L., and Cooke, W. D., Excitation processes in
flame spectrometry. A d . Chem. 35, 208-277 (1983).
G4. Gidley, J. A. F., and Jones, J. T., The determination of zinc in metallurgical
materials by atomic-absorption spectrophotometry. Analyst 85, 249-256 ( 1960).
G5. Gidley, J. A. F., and Jones, J. T., The determination of zinc in metallurgical
materials by atomic-absorption spectrophotometry. Analyst 86, 271 ( 1981).
G8. Gilbert, P. T., Absorption flame photometry. Anal. Chem. 34, 210R-220R
(1982).
G7. Griffith, F. D., Parker, H. E., and Rogler, J. C., The influence of dietary
fluoride on magnesium distribution in growing chicks. Federation Proc. 22, 554
(1983).
58 ALFRED ZElTNER

G8. Grove, E. L., Jones, R. A., and Mathews, W., The loss of sodium and potas-
sium during the dry-ashing of animal tissues. And. B f o c h a . 2, 221-230 (1981).
H1. Hedge, B., Griffith, G. C., and Butt, E. M., Tissue and serum manganese
levels in evaluation of heart muscle damage, A comparison with SGOT. Proc. SOC.
Erptl. Biol. Med. 107, 734-737 (1981).
H2. Henmann, R., Untersuchungen an Zerstaubern fur die Emissions-und Absorp-
tions-Flammenphotometrie. Optik 18, 422-430 ( 1981 ).
H3. Henmann, R., and Lang, W., Analysen von Natrium und Kalium in Seren
mit Hilfe der Absorptionsflammenphotornetrie. Z. Ges. Erptl. Med. 134, 288-279
(1981).
H4. Herrmann, R., and Lang, W., Untersuchungen an Zerstauberkammern fur
die Emissions- und Absorptionsflammenphotometrie. Optik 19,208-218 ( 1982).
H5. Hinson, W. H., An ion exchange treatment of ash extracts for removal of
interfering anions in the determination of calcium by atomic absorption. Spectrochim.
Acta 18, 427-429 (N62).
H8. Horn, D. B., and Latner, A. L., The estimation of magnesium by atomic
absorption spectrophotometry. Clin. Chim. Acta 8, 974-978 ( 1983).
H7. Hunter, G.,Calcium and magnesium content of normal human blood serum.
Nature 182, 283-284 (1958).
J1. Jacobs, M. B., Goldwater, L. J.. and Gilbert, H., Ultramicrodetennination of
mercury in blood. Am. Ind. H y g . Assoc. J . 22, 278-279 (1981).
J2.Jones, W. J., and Walsh, A., Hollow-cathode discharges-the construction
and characteristics of sealed-off tubes for use as spectroscopic light sources. Spectro-
chim. Acta 18, 249-254 (1980).
K1. Kahn, H. L., and Slavin, W., An atomic absorption spectrophotometer. Appl.
Opt. 2, 931-938(1983).
K2. Kagi, J. H. R., and Vallee, B. L., Metallothionein: A cadmium- and zinc-
containing protein from equine renal cortex. J. B i d . Chem. 235, 3480-3465 (1980).
K3. Kagi, J. H. R., and Vallee, B. L., Metallothionein: A cadmium- and zinc-
containing protein from equine renal cortex. J. B i d . Chem. 238, 2435-2442 ( 1981 ).
K4. Kingsley, G. R., and Schaffert, R. R., Micro-flame photometric determination
of sodium, potassium, and calcium in serum with organic solvents. 1. Biol. Chem.
206, 807-815 ( 1954).
K5. Kirchhoff, G., Uber das Verhaltnis zwischen dem Emissionsvennogen und
dem Absorptionsvermogen der Korper. Pogg. Ann. 109, 275-301 (1880).
K8. Kirchhoff, G., On the relation between the radiating and absorbing powers of
different bodies for light and heat. Phil. Mag. [41 20, 1-21 (1880).
K7. Kirchhoff, G., and Bunsen, R., Chemische Analyse durch Spectralbeobach-
tungen. Pogg. Ann. 110, 161-189 (1880).
K8. Kirchhoff, G., and Bunsen, R., Chemical analysis by spectrum-observations.
Phil. Mag. [4]20, 89-109 (1880).
K9. Kirchhoff, G., and Bunsen, R., Chemische Analyse durch Spectralbeobach-
tungen. Pogg. Ann. 189, 337-381 ( 1881).
K10. Kirchhoff, G., and Bunsen, R., Chemical analysis by spectrum-observations
Phil. Mag. 141 22, 329-349 ( 1881).
K11. Kniseley, R. N., D'Silva, A. P., and Fassel, V. A,, A sensitive premixed
 ATOMIC ABSORPTION SPECTROSCOPY 59

oxyacetylene atomizer burner for flame emission and absorption spectrometry. Anal.
Chem. 35, 910-911 (1983).
L1. Lang, W., and Herrmann, R., Signal-zu-Rauch-Verhaltnisbei flammenphoto-
metrischen Absorptionsmessungen. Uptik 20, 347-352 ( 1963) .
L2. Lang, W., and Herrmann, R., Durchflussmengen pneumatischer Ringspaltzer-
stauber fur Flammenphotometer. Optik 20, 391-398 ( 1963) .
L3. Leithe, W., Die Absorptions-Flammenphotometrie in der analytischen Chemie.
Ang. Chem. 73, 488-492 (1961).
L4. Leithe, W., and Hofer, A., Uber Magnesiumbestimmungen durch Atomarab-
sorptions-Flammenphotometrie. Mikrochim. Acta, pp. 268-276 ( 1961) .
L5. Leithe, W., and Hofer, A., Uber Magnesiumbestimmungen durch Atomarab-
sorptions-Flammenphotometrie. 11. Bestimmung des Magnesiums in Aluminium-
Legierungen. Mikrochim. Acta, pp. 277-282 ( 1961).
L6. Leyton, L., Phosphate interference in the flame-photometric determination of
calcium. Analyst 79, 497-500 ( 1954).
L7. Lindstrom, O., Rapid microdetermination of mercury by spectrophotometric
flame combustion. Anal. Chem. 31, 461-467 (1959).
L8. Lockyer, R., Some factors affecting performance in atomic absorption spectros-
copy. A d y s t 85, 461 (1960).
L9. Lockyer, R., and Hames, G. E., The quantitative determination of some noble
metals by atomic-absorption spectroscopy. Analyst 84, 385-387 ( 1959).
L10. Lockyer, R., Scott, J. E., and Slade, S., Enhancement of atomic absorption
in the flame by organic solvents. Nature 189, 830-831 (1961).
L11. Lothian, G. F., Beer’s Law and its use in analysis. Analyst 88, 678-685
(1963).
L12. L’vov, B. V., The analytical use of atomic absorption spectra. Spectrochim.
Acta 17, 761-770 (1961).
M1. Malissa, H., and Schoffmann, E., Uber die Venvendung von substituierten
Dithiocarbamaten in der Mikroanalyse. Mikrochim. Acta, pp. 187-202 ( 19551.
M2. Malmstadt, H. V., and Chambers, W. E., Precision null-point atomic absorp-
tion spectrochemical analysis. A d . Chem. 32, 225-235 (1960).
M2a. Manning, D. C., Trace metals in blood plasma. Atomic Absorption News-
letter No. 11. Perkin-Elmer Corp., Norwalk, Conn. (March 1963).
M3. Manning, D. C., and Slavin, W., Lithium isotope analysis by atomic absorp-
tion spectrophotometry. Atomic Absorption Newsletter No. 8. Perkin-Elmer Corp.,
Norwalk, Conn. ( November 1962).
M4. Margerum, D. W., and Santacana, F., Evaluation of methods for trace zinc
determination. Anal. Chem. 32, 1157-1161 (1960).
M5. Mavrodineanu, R., Flame characteristics and emission. Spectrochim. Acta
17, 1016-1042 ( 1961).
M6. Menzies, A. C., A study of atomic absorption. Anal. Chem. 32, 898-904
( 1960).
M7. Menzies, A. C., Atomare Emission und Absorption in Flammen. Z. Instru-
mentenk. 68, 242-247 ( 1960).
M8. Miller, B., Project surveyor to seek solar origin. Aviation Week 75, 62-66
(1961).
60 ALFRED ZETI’NER

M9. Mitchell, R. L., and Robertson, I. M., The effect of aluminum on the flame
spectra of the alkaline earths: A method for the determination of aluminum, J . SOC.
Chem. lnd. (London) 55, 289T-272T (1938).
M10. Montgomery, R. D., The estimation of magnesium in small biological sam-
ples by flame spectrophotometry. j. Clin. Pathol. 14, 400-402 (1981).
M11. Mossotti, V. G., and Fassel, V. A., Atomic absorption spectra of the
lanthanide elements. Spectrochim. Acta. In press.
M12. Miiller, R. H., Developments in atomic absorption spectrometry. Anal.
Chem. 36, 147A-148A ( 1964).
N1. Nelson, L. S., and Kuebler, N. A., Vaporization of elements for atomic ab-
sorption spectroscopy with capacitor discharge lamps. Spectrochim. Acta 19, 781-
784 (1983).
N2. Newbrun, E., Application of atomic absorption spectroscopy to the deter-
mination of calcium in saliva. Nature 192, 1182-1183 ( 1961).
P1. Parker, H. E., Magnesium, calcium, and zinc in animal nutrition. Atomic
Absorption Newsletter No. 13. Perkin-Elmer Corp., Nonvalk, Conn. ( May 1983).
P2. Paschen, F., Bohr’s Heliumlinien. Ann. Physik 50, 901-940 (1918).
P3. Perkins, J,, The determination of sodium in halo-phosphate phosphors by
atomic absorption spectroscopy. Analyst 88, 324-338 ( 1963).
P4. Porter, P., and Wyld, G., Elimination of interferences in flame photometry.
Anal. Chem. 27, 732-738 ( 1955).
P5. Pungor, E., Hegediis, A. J., Thege, I. K., and Zapp, E. E., Uber die Rolle der
Flammentemperatur bei der flammen-photometrischen Analyse der Alkalimetalle.
Mikrochim. Acta, pp. 1247-1263 (1958).
R1. Robinson, J. W., Determination of sodium by atomic absorption spectroscopy.
Anal. Chim. Acta 23, 458-481 (1960).
R2. Robinson, J. W., Atomic absorption spectroscopy. Anal. Chem. 32, 17A-29A
(1980).
R3. Robinson, J. W., Effect of organic and aqueous solvents on flame photometric
emission and atomic absorption spectroscopy. Anal. Chim. Acta 23, 479-487 ( 1980).
R4. Robinson, J. W., Determination of lead in gasoline by atomic absorption
spectroscopy. Anal. Chim. Acta 24, 451-455 (1981).
R5. Robinson, J. W., Recent advances in atomic absorption spectroscopy. Anal.
Chem. 33, 1087-1071 (1981).
R6. Robinson, J. W., Flame photometry using oxycyanogen flame. Anal. Chem.
33, 1227-1230 ( 1981).
R7. Robinson, J. W., and Harris, R. J., Mechanical feed burner with total con-
sumption for flame photometry and atomic absorption spectroscopy. Anal. Chlm.
Acta 26, 439-445 ( 1982).
R8. Russell, B. J., and Walsh, A., Resonance radiation from a hollow cathode.
Spectrochim. Acta 15, 883-885 ( 1959).
R9. Russell, B. J., Shelton, J. P., and Walsh, A., An atomic-absorption spectro-
photometer and its application to the analysis of solutions. Spectrochim. Acta 8,
317-328 (1957).
S1. Sikorski, M. E., and Copeland, P. L., Application of atomic absorption spectros-
copy to solids. Spectrochim. Acta 9, 361 (1957).
 ATOMIC ABSORPTION SPECTROSCOPY 61

S2. Skogerboe, R. K., and WoodrifF, R. A,, Atomic absorption spectra of europium,
thulium, and ytterbium using a flame as line source. Anal. Chem. 35, 1977 (1964).
S3. Slavin, W., A burner-atomizer for atomic absorption spectrophotometry.
Atomic Absorption Newsletter No. 10. Perkin-Elmer Corp., Nonvalk, Conn. (Feb-
ruary 1963).
S4. Slavin, W., and Manning, D. C., Atomic absorption spectrophotometry in
strongly reducing oxyacetylene flames. A d . Chem. 35, 253-254 ( 1963).
S5. Slavin, W., Sprague, S., and Manning, D. C., The determination of calcium
by atomic absorption spectrophotometry. Atomic Absorptim Newsletter N O . 15.
Perkin-Elmer Corp., Nonvalk, Conn. ( September 1963).
S6. Spector, J., Mutual interferences in elimination of calcium interference in flame
photometry. Anal. Chem. 27, 1452-1455 ( 1955).
S7. Stanley, R. W., and Dieke, G . H., Interferometric wavelengths of iron lines
from a hollow cathode discharge. I. Opt. SOC. Am. 45, 280-286 (1955).
S8. Stewart, W. K., Hutchinson, F., and Fleming, L. W., The estimation of
magnesium in serum and urine by atomic absorption spectrophotometry. J. Lab.
Clin. Med. 61, 858-872 (1963).
S9. Stokes, G . G., On the simultaneous emission and absorption of rays of the
same definite refrangibility; being a translation of a portion of a paper by M. LBon
Foucault, and of a paper by Prof. Kirchhoff. Phil. Mag. [41 19, 193-197 (1860).
S10. Strasheim, A., and Nell, J. P., The flame photometric determination of cal-
cium in plant and biological materials. J . S. African Chem. Imt. 7 , 79-89 (1954).
S11. Strasheim, A., Strelow, F. W. E., and Butler, L. R. P., The determination of
copper by means of atomic absorption spectroscopy. J . S. African Chem. I&. 13,
73-81 (1960).
T1. Tolansky, S., The nuclear spin of arsenic. Proc. Roy. SOC. A137, 541-558
(1932).
T2. Torok, T., Einige flammenspektralanalytische Beobachtungen. Z . Anal. Chem.
116, 29-33 (1939).
W1. Wallace, F. J., The determination of magnesium in aluminum alloys by
atomic-absorption spectroscopy. Analyst 88, 259-265 ( 1963) .
W2. Walsh, A., The application of atomic absorption spectra to chemical analysis.
Spectruchim. Acta 7 , 108-117 (1955).
W3. West, A. C., Effects of aliphatic acids and their salts on the flame spectro-
metric emission of calcium. A n d . Chem. 36, 310-314 (1964).
W4. West, A. C., and Cooke, W. D., Elimination of anion interferences in flame
spectroscopy. Use of ( ethylenedinitrilo ) tetraacetic acid. Anal. Chem. 32, 1471-1475
(1960).
W5. White, J. U., Long optical paths of large aperture. 1. Opt. SOC. Am. 32,
285-288 ( 1942) .
W6. Williams, C. H., The use of lanthanum chloride to prevent interferences in
the flame photometric determination of exchangeable calcium in soils. Anal. Chim.
Acta 22, 163-171 (1960).
W7. Williams, C. H., David, D. J,, and Iismaa, O., The determination of chromic
oxide in feces samples by atomic absorption spectrophotometry. J. Agr. Sci. 59,
381-385 (1962).
62 ALFRED ZETTNER

W8. Willis, J. B., The determination of calcium in blood serum by atomic absorp-
tion spectroscopy. Nature 186, 249-250 (1960).
W9. Willis, J. B., The determination of metals in blood serum by atomic absorp-
tion spectroscopy-I. (Calcium). Spectrochim. Acta 16, 259-272 ( 1960).
W10. Willis, J. B., The determination of metals in blood serum by atomic absorp-
tion spectroscopy-11. (Magnesium). Spectrochim. Acta 16, 273-278 ( 1960).
W11. Willis, J. B., The determination of metals in blood serum by atomic absorp-
tion spectroscopy-111. (Sodium and Potassium), Spectrochim. Acta 16, 551-558
( 1960).
W12. Willis, J. B., Determination of lead urine by atomic absorption spectros-
copy. Nature 191, 381-382 (1961).
W13. Willis, J. B., Determination of calcium and magnesium in urine by atomic
absorption spectroscopy. Anal. Chem. 33, 556-559 ( 1961).
W14. Willis, J. B., Determination of lead and other heavy metals in urine by
atomic absorption spectroscopy. Anal. Chem. 34, 614-617 ( 1962).
W15. Willis, J. B., Analysis of biological materials by atomic absorption spectros-
copy. Methods Biochem. Anal. 11, 1-67 (1963).
W16. Winefordner, J. D., and Vickers, T. J., Atomic fluorescence spectroscopy as
a means of chemical analysis. Anal. Chem. 36, 161-165 (1964).
W17. Winefordner, J. D., and Staab, R. A., Determination of zinc, cadmium, and
mercury by atomic fluorescence flame spectrometry. Anal. Chem. 36, 165-168 (1964).
W18. Winefordner, J. D., Mansfield, C. T., and Vickers, T. J., Atomization effi-
ciency of total consumption atomizer-burners in flame photometry. Anal. Chem.
35, 1607-1610 ( 1963).
W19. Winefordner, J. D., Mansfield, C . T., and Vickers, T. J., Temperatures of
some typical flames used in flame photometry. Anal. Chem. 35, 1611-1613 (1963).
W20. Woodson, T. T., A new mercury vapor detector. Rev. Sci. Instr. 10, 308-
311 (1939).
W21. Wiinsch, A,, and Teicher, K., Wber die Bestimmung von Mg in Pflanzensub-
stanzen durch die Absorptions-Flammenspektroskopie. Z. Pflanzenernaehr. Dueng.
Bodenk. 97, 101-106 ( 1962).
Y1. YofB, J., and Finkelstein, R., Elimination of anionic interference in flame
photometric determination of calcium in the presence of phosphate and sulfate.
A d . Chim. Acta 19, 166-173 (1958).
Z1. Zeeman, P. B., and Butler, L. R. P., The determination of lead, copper, and
zinc in wines by atomic absorption spectroscopy. Appl. Spectry. 16, 120-124 ( 1962).
22. Zettner, A., Unpublished data.
23. Zettner, A,, and Seligson, D., Application of atomic absorption spectrophotom-
etry in the determination of calcium in serum. Clin. Chem. 10, 869-890 (1964).
24. Zettner, A., and Seligson, D., Determination of serum magnesium by atomic
absorption spectrophotometry. Clin. Res. 11, 406 ( 1963).