FLAME PHOTOMETRY

.
1 Maclntyre
Department of Chemical Pathology. Postgraduate Medical School.
London. England

Page
.
1 Introduction ..................................................... 1
1.1. Historical; Background ....................................... 2
2. The Flame Photometer ............................................ 3
2.1. The Atomizer ............................................... 3
2.2. The Burner ................................................. 5
2.3. The Optical System .......................................... 6
2.4. Photosensitive Detectors ...................................... 6
2.5. Measurement of Emission ..................................... 6
2.6. Internal Standard Instruments .................................. 7
2.7. Commercial Flame Photometers ................................ 8
3. Fundamental Principles ............................................ 8
3.1, The Flame .................................................. 8
3.2. Emission Spectra ............................................. 9
3.3. Absorption Spectra ........................................... 10
3.4. Organic Solvents ............................................. 13
3.5. Interference ................................................. 14
4 Practical Applications ............................................. 19
4.1. Calcium .................................................... 19
4.2. Magnesium ................................................. 21
4.3. Determination of Magnesium and Calcium in Biological Fluids and
Tissues ..................................................... 23
.
5 Working Conditions ............................................... 26
6. Conclusion ...................................................... 27
References ........................................................ 27

I . Infroduction
It is the object of this review to provide a critical account of modem
flame photometry in the light of the basic principles of the method. No
attempt has been made to provide a catalog of the innumerable articles
appearing on the subject. but it is hoped that this chapter will help
clinical chemists to decide whether flame photometry is applicable to a
particular problem and perhaps to suggest ways of overcoming difficulties
with the method .
1
2 I. MAC-

1.1. HISTORICAL;
BACKGROUND
Herschel ( H l ) studied the emission of salts introduced into a flame;
he recorded [quoted by Mavrodineanu (M2) ] that:
“Salts of soda give a copious and purely homogeneous yellow.
Salts of potash give a beautiful pale violet.
Salts of lime give a brick red, in whose spectrum a yellow and bright
green line are seen.
Salts of strontia give a magnificent crimson. If analyzed by the prism
two definite yellows are seen, one of which verges strongly to orange.
Salts of magnesia give no colour.
Salts of lithia give a red (on the authority of Dr. Turner’s experiment
with the blow-pipe ) .
Salts of baryta give a fine pale apple-green. This contrast between the
flames of baryta and strontia is extremely remarkable.
Salts of copper give a superb green, or blue green.
Salts of iron (protoxide) gave white, where the sulphate was used . . . .
The colors thus communicated by the different bases to flames afford
in many cases a ready and neat way of detecting extremely minute quan-
tities of them; . . .”
Talbot ( T l ) had previously studied the flame spectra of lithium and
strontium, but the full potentialities of Herschel’s observations for qual-
itative analysis were first realized in the work of Kirchhoff and Bunsen
( K 1 ). These authors discovered cesium in 1860 and rubidium in the
following year by observation of their flame spectra.
Janssen (J1) suggested that spectral analysis, until then used only for
qualitative observations, was suitable also for quantitative work. He felt
that such a development would be particularly advantageous in the case
of elements like sodium which were difficult to determine by classic
procedures. His suggestions bore fruit 3 years later when Champion et
ul. (C1) constructed an instrument for the determination of sodium in
plant ash. A solution of plant ash was introduced into the flame by
means of a platinum wire and the emission intensity measured by com-
paring it by means of a visual photometric attachment with light from
a reference constant-intensity sodium flame. This “spectronatromAtre”
was the first flame photometer; and when one considers that it was
capable of an accuracy of between 2 and 5 %, it is interesting that it was
not for more than 70 years that the method was applied to clinical
problems.
However, the principles of flame photometry were not fully developed
until the brilliant work of Lundegirdh ( L l , L2). This worker devised for
the first time a satisfactory method of introducing solutions into the
 FLAME PHOTOMETRY 3

flame at a constant rate. This was by means of a concentric atomizer

which dispersed the solution into droplets which were then led through
a spray chamber to remove the larger particles before passing into the
flame. Instead of the Bunsen flame used by the earlier workers, Lun-
degHrdh substituted an air-acetylene flame and photographed the emis-
sion on a photographic plate after it had been dispersed by means of a
spectograph. He also developed procedures in which the intensity of an
emission line was recorded directly by a galvanometer connected to the
amplified output from a photocell which had been placed so as to re-
ceive an appropriate portion of the flame spectrum dispersed by a prism.
Lundegkrdh was able to measure almost half of the elements in the
periodic table, and his work is the foundation of all modern work on
flame photometry. It is only in the last few years that substantial ad-
vances have been made beyond the techniques employed by Lunde-
gkdh.
Schuhknecht ( S1) produced a greatly simplified instrument by sub-
stituting a filter for the monochromator used by Lundegbrdh, and this
type of apparatus is probably the most widely used today. But for the
outbreak of war in 1939 this simple method would certainly have been
rapidly adopted by clinical chemists, but it was not until the work of
Barnes and associates ( B l ) and Doming0 and Klyne (D2) after earlier
work by Ells and Marshall ( E 2 ) , Ells ( E l ) , Griggs et al. ( G 2 ) , and
Cholak and Hubbard (C2) that the method became widely adopted.
In modern laboratories sodium and potassium are almost exclusively
determined by flame photometry and it seems likely that the same will
shortly become true 01 calcium and magnesium and possibly of iron,
copper, chromium, manganese, cobalt, lead, and zinc.
2. The Flame Photometer
The flame photometer consists essentially of an atomizer, a burner,
some means of isolating the desired part of the spectrum, a photosen-
sitive detector, sometimes an amplifier and, finally, a method of pre-
senting the desired emission, whether by galvanometer, null meter, or
chart recorder.
2.1. THE ATOMIZER
Two main types are in common use: ( a ) an atomizer which produces
an aerosol which passes through a spray chamber before reaching the
flame (Fig. 1); ( b ) an atomizer which sprays directly into the flame.
This type is sometimes an integral part of the burner, as with the Beck-
man or Zeiss atomizer-burner (Fig. 2 ) . Atomizers of type ( a ) are usually
4 I. MACINTYRE

FIG.1. An atomizer of the aerosol-producing type; aerosol passes through a spray

chamber before reaching the flame. ( 1) Inlet capillary; ( l a ) metal atomizer; ( 2 )
drainpipe; ( 3 ) fuel and air inlets; ( 4 ) spray chamber; ( 5 ) baffles; (6) burner. (By
permission of Evans Electroselenium Ltd., Halstead, Essex, England. )

4
H,
3 KI HI)

01

FIG.2. Schematic diagram of burner of Zeiss flame spectrophotometer, ( 1) Sample;

( 2 ) cannula; ( 3 ) fuel gas inlet pipe; ( 4 ) guide piece for cannula; ( 5 ) flame; (6)
throttling device for regulation of pressure indicator; ( 7 ) pressure gage; (8 and 9 )
pressure regulators; ( 10) pressure gage. (By permission Carl Zeiss, Germany, through
Degenhardt & Co. Ltd., London, W.1.)
 FLAME PHOTOMETRY 5

concentric, as shown in Fig. 1, but are sometimes constructed of two

capillaries at right angles (Fig. 3). The spray chamber-atomizer is gen-
erally employed with simpler instruments and cooler flames, while the
integral atomizer-burner is usually used when hotter flames such as
oxyacetylene are burned.
In the author’s opinion the integral type is much superior for most
purposes. Unlike the spray chamber atomizer, the integral type reaches
equilibrium almost instantly when solutions are sprayed and, if correctly
designed, is extremely robust in use. A further important distinction is

1
FIG.3. Atomizer constructed from two capillaries at right angles. Atomizer (two
vertical cross sections at right angles): (A) lower part carrying air inlet (B)and drain
for waste ( C ) . ( A ) is joined by the ground glass joint ( D ) to the upper part (E),
which carries the solution inlet ( F ) , the b d e plate ( G ) , and the outlet to the
burner (H).( F ) and ( G ) are mounted on ground glass joints (1, K), ( F ) is joined
by a piece of narrow rubber tubing to the vertical tube ( L )which dips into the
solution to be analyzed. (By permission of Doming0 and Klyne ( D 2 ) . )

that organic solvents can be used only in the integral atomizer without
difficulty and without producing an unstable flame. On the other hand,
flames cooler than air-hydrogen are not usable with type ( b ) ,while the
spray chamber-atomizer can be used with any flame.
2.2. THE BURNER
As will have been gathered from the preceding description, two main
types of burners are employed. The Meker type burner is most often
used for cooler flames. In this type the flame gases are mixed inside the
burner tube and are prevented from striking back by a grid at the mouth
of the tube, Different grids are employed for different gas mixtures, but
6 I. MACINTYRE

the orifices in the grid for the hotter flames are generally inconveniently
small so that frequent cleaning may be necessary; the integral atomizer-
burner has already been mentioned.

2.3. THE OPTICAL

SYSTEM
Flame photometers can be divided into two groups on the basis of
their optical systems. In group 1, the required part of the spectrum is
selected by means of absorption or interference filters. These instruments
are suitable only for the determination of sodium and potassium in
biological fluids, whatever the claims of manufacturers. In group 2,
emissions are isolated by means of a prism or diffraction-grating mono-
chromator. When used with a higher temperature flame, such as oxy-
acetylene, instruments of this type are capable of determining sodium,
potassium, calcium, and magnesium in all biological fluids and tissues;
they are also capable of determining many other elements of biological
importance, such as iron, manganese, and cobalt, after a suitable pre-
liminary extraction into organic solvents.
2.4. PHOTOSENS~VE
DETECTORS
Three types are used. ( I ) Bader-layer cells. These are satisfactory
only for simple filter instruments. (2) Vacuum phototubes. These tubes
require an external power supply, unlike barrier-layer cells, and their
output is usually amplified before measurement. ( 3) Photomultiplier
tubes are easily the most satisfactory detectors for use in flame pho-
tometry. The photocurrent is amplified inside the tube in such a way
that much lower light levels can be detected and measured accurately
than is possible with vacuum phototubes with amplifiers. A stable source
of high voltage up to perhaps 2000 volts is required to operate the photo-
multiplier tubes, but these tubes are almost universally used in high-
performance instruments and are essential if the advantages of using
narrow band width are to be obtained.
2.5. MEASUREMENT
OF EMISSION

The emission of the selected element may be presented in several

ways. Easily the most satisfactory is a direct-reading galvanometer.
m i l e most simple instruments use this method, some of the more ex-
pensive instruments use a nullpoint method. In the author’s opinion this
latter method of measuring emissions is less suitable for the best flame
photometric work.
For more complete and advanced studies, recording devices which
 FLAME PHOTOMETRY 7

record the emission while the wavelength is continuously varied give

much more information than is easily obtained by other means.
2.6 INTERNAL STANDARD INSTRUMENTS
Znternal Standard Instruments. In some instruments a double-beam
principle is employed. An internal standard element, such as lithium, is
added to a constant concentration to all unknown and standard solutions.
Dual optical paths are employed and the internal standard emission, after

FIG.4. A flame photometer with double monochromator. (By permission of Carl

Zeiss, Germany, through Degenhardt, London, W . l . )

isolation by a suitable optical system, is focused on a photocell. The

current produced from this cell is opposed to the current from a separate
photocell which responds to the isolated emission from the element being
measured. The opposing currents are balanced by means of an accurate
potentiometer, the potentiometer readings to produce balance when
known solutions are sprayed is noted, and the unknown concentrations
are deduced. This method is effective in eliminating or minimizing spray
interference (see below) but is incompletely effective in the other types
of interference met with in biological samples. Unfortunately, it is just
these types of interference which occur most often and which present
8 I. MACINTYRE

the greatest problems. In the author's opinion the extra cost entailed in
the provision of internal standard operation would be better expended
by providing improved optical resolution and more sensitive photocells.
FLAMEPHOTOMETERS
2.7 COMMERCIAL
There are many excellent commercial instruments on the market, and
it should be fairly simple to select a suitable model from the considera-
tions mentioned in the preceding section.
When the requirement is for many routine analyses of sodium and
potassium, a simple filter flame photometer burning a low temperature
flame should be purchased. Many such models are on the market. On
the other hand, if analysis for calcium and magnesium in biological
fluids is also required, then only a fairly complex instrument with mono-
chromator, photomultiplier, and high-temperature flame is satisfactory
(Fig. 4). Compromise instruments between these two extremes lose the
simplicity of the first type without gaining the versatility of the second.

3. Fundamenfal Principles
3.1. THEFLAME
A thorough study of flames is described by Gaydon and Wolfhard
( G1) ; for practical analytical purposes their most important character-
istic is their temperature. This is because the proportion of atoms which
are excited in the % m edepends critically upon the flame temperature
as well as on the characteristics of the element. Table 1 lists the tem-
peratures of flames which have been used in flame photometry. When
an integral atomizer-burner is used, aspiration of an aqueous solution
results in some lowering of flame temperature, perhaps by as much as
TABLE 1
FLAME TEMPERATURES
Temperature
Fuel-oxidant mixture ("GI
Coal gas-air 1840
Propane-air 1925
Butane-& 1930
Acetylene-air 2050
Hydrogen-air 2115
Hydrogen-oxygen 2690
Acetylene-oxygen 3110
Hydrogen-perchloryl fluoride 3300
Hydrogen-fluorine 4000
Cyanogen-oxygen 4850
 FLAME PHOTOMETRY 9

200" in the case of an acetylene-oxygen flame. In cooler flames or with

the spray chamber-atomizer, cooling is probably much less. One of the
cooler flames is optimum for the determination of sodium and potassium;
where calcium and magnesium or other biological elements are also to
be determined then acetylene-oxygen is the flame of choice.
3.2. EMISSION
SPECTRA
When salt in solution is sprayed into a flame a whole complex series
of reactions rapidly ensues: the water envelope of the droplet is evap-
orated, and the resulting solid salt particle is heated and then vaporized
into gaseous salt molecules. These molecules are then dissociated by the
heat of the flame into free atoms; some of these atoms recombine re-
versibly with other components of the flame, such as OH, and still others
lose an electron and become ionized, In this mixed flame population of
molecules, atoms and ions, each may absorb energy from the flame and
pass to a higher or excited state. This absorbed energy may later be
released as light when it forms the characteristic emission spectrum of
the element. When an atom absorbs energy from the flame and becomes
excited, one of its outer valence electrons is displaced further from the
nucleus to a position of higher energy. According to modern theory,
energy can only be absorbed by the atom in discrete amounts, so that
only certain definite higher energy states can occur. The atom cannot
exist in a state of energy intermediate between the ground state and
the next higher energy level. The difference between these t w o energy
levels in each atom is usually expressed in electron volts (an electron
volt = 1.6 x 10-l2 erg) and is defined as the excitation potential. This
absorbed energy may be emitted as light and from the relation
Ej-Eo = hv
where E, = energy of higher level; Eo = energy of lower level; h =
Planck's constant, and v = frequency of radiation, it follows that each
transition is associated with the emission of light of a definite wave-
length. It is also apparent from this equation that higher energy tran-
sitions will result in emissions at the blue and ultraviolet end of the
spectrum, while lower energy transitions will occur at the longer wave-
lengths. Molecules and ions may also absorb energy from the flame, and
in the case of ions energy transitions will, like atomic transitions, result
in emission concentrated over a very narrow band width. In the case of
molecules, however, the changes in electronic energy of the constituent
atoms are usually accompanied by changes in the energy of vibration of
the constituent atoms or of the rotation of the whole molecule. These
10 I. MACINTYRE

energy changes in vibration-rotation may also result in emission of light,

and the sum of these emissions is a broad band extending through a
considerable spectral range.
There exists an upper limit of energy for each atom, and when this is
reached the atom becomes ionized; the energy required for ionization is
known as the ionization potential. With the cooler flames only a very
small proportion of atoms in the flame is ionized, but in the hotter
flames this is not true, and in some cases ions may constitute the greater
proportion of the element in the flame.
Whether one considers atoms or ions, the proportion of atoms which
is in the excited state is extremely small. In the emission of a spectral
line due to the transition from an excited state i of excitation energy Ei
to a ground state of energy Eo = 0, if Pj and Po are the statistical weights
for the excited and ground states, respectively, the number of atoms in
the excited state iVj is related to the number of atoms in the ground
state N o by the relation
N,= No P,/Po exp (-E,/kT)
(Wl). Table 2 is taken from Walsh's paper and shows the small propor-
tion of atoms in the excited state for various elements in transitions to
the ground state (resonance lines). This table shows the dependence of
the number of atoms in the excited state, and therefore of the emission
and sensitivity of the method, on the flame temperature.
3.3. ABSORPTIONSPECTRA
Walsh ( W l ) pointed out that if one could measure some function of
the unexcited atoms in the flame, rather than of the much smaller propor-
tion of excited atoms as in emission flame spectrophotometry, one should
have a potentially sensitive method. He showed that this was possible if
light of the appropriate wavelength were directed through a flame and
onto a photocell, after passing through a monochromator. Atoms in the
ground state in the flame absorb light of the same wavelength as is
emitted in transitions to the ground state, and the proportion of light
absorbed will bear a relationship to the atoms in the ground state
and therefore, except where ionization is appreciable, to the total number
of atoms in the flame and hence to the total concentration of the element
in solution. Light of the appropriate wavelength is obtained from a
hollow-cathode tube, and the absorption of this wavelength is measured
using an apparatus of the type shown in Fig. 5. Any light emitted at this
same wavelength by the excited atoms in the flame can be eliminated by
modulating the monochromatic light source and arranging that only
 TABLE 2
VALUES OF N , / N , FORVARIOUSRESONANCE LINES^ M

Resonance NjIN,
line PjIP, T = 2000°K T = 3000°K T = 4000°K T = 5000°K
Ca 8521 A 2 4.44 x 10-4 7.24 x 10-3 2.98 x 10-2 6.82x 10-2
Na 5890 A 2 9.86 x 10-6 5.88 x 10-4 4.44 x 10-3 1.51 x 10-2
Ca 4227 A 3 1.21 x 10-7 3.69 x 10-5 6.03 x 10-4 3.33 x 10-3
Zn 2139 A 3 7.29 x 10-16 5.58 x 10-10 1.48 x 10-7 4.32 x 10-6
a By permission from Walsh (Wl).
12 I. MACINTYRE

light modulated at this frequency will be detected by the recorder. The

absorption recorded is much less dependent on temperature than is
emission because the Doppler broadening which decreases peak ab-
sorption increases only as T112
[ cf.N,/No = P#o exp ( - E,/kT) 1.
This is a most elegant and attractive method, which is likely to be
used increasingly in flame photometric analysis. As will be discussed
later, interference effects are not inherently absent with this method;
anionic interference is equally or more troublesome with this method
than with emission flame work, while cationic interference is less prom-
inent, not because of the inherent nature of the method, but because of
the lower flame temperatures usually used with this method. Its attrac-
tion is that it has some claims to be an absolute method in certain cir-
Amplifier and

/( ( fl
rectifier
. A

Light Detector Recorder

s o u r c )-.
~
Monochromator

4
Chopper

FIG.5. Apparatus for measuring atomic absorption. (By permission of Russell et

al. ( R l , p. 318).)

cumstances, but its great practical advantage is in its ability to measure

elements whose excitation potential is too high (i.e., 5.5 ev or above)
for effective measurement in emission work even using an oxygen-
acetylene flame with an organic solvent. Thus elements such as zinc and
cadmium, with resonance lines in the extreme ultraviolet (and therefore
involving high-energy transitions) have the same order of sensitivity as
sodium with this method, although the excitation potential of the sodium
line at 589 is much lower (2.1 ev). This is in marked contrast with
emission spectra, where zinc is almost impossible to measure.
However, the method is so far in a relatively undeveloped stage, and
in the author’s opinion the best instrument for exploitation of this prin-
ciple has still to be designed. Further research into the use of this
method with hotter flames seems desirable; most applications so far
described have depended on Meker-type burners with a spray chamber
atomizer with its attendant disadvantages. The method should be equally
applicable to the integral atomizer-burner, and it seems possible that a
monochromator instrument of this type using either absorption or
 FLAME PHOTOMETRY 13

emission, whenever appropriate, would allow the use of these two

principles in a complementary manner.

3.4. ORGANIC
SOLVENTS
Organic solvents give a greatly increased sensitivity in emission work
and are most successfully used with an integral type of atomizer. The
organic solvents exert their enhancing effect either in aqueous-organic
solvent mixtures or still more when the organic solvent is used in 100 %
concentration. While several factors are of importance, the main reason
for this increased sensitivity is the increased flame temperature possible
with organic solvents, so that temperatures 200-300" higher than is pos-
sible with aqueous solvents can be attained. The exponential dependence
of sensitivity in emission work on flame temperature makes this increased
flame temperature very s i e c a n t . This subject has been most fully
studied by Dean (Dl). Table 3 records the enhancement possible when
organic solvents are used in place of aqueous solutions. By dissolving a
suitable complexing agent in an organic solvent, for example, 4-methyl-
2-pentanone, many elements can be extracted directly into the organic
solvent, which is then directly sprayed into the flame. This procedure is
TABLE 3
EMISSIONINTENSITY OF CERTAIN ELEMENTS
IN
WATERAND ORGANIC SOLVENTS~
Wave- Excitation Sensitivity
length potential (pg/ml)/(% T ) Enhance-
Element (mP) (ev) Water Organic men€
A1 396.2 3.14 67.0 0.5b 134-fold
Cr 425.4 2.91 5.0 0.16 50
cu 324.7 3.81 0.6 0.06c 10
Fe 372.0 3.3 2.5 0.125b 20
La0 560 2.21 0.5 0.05b 10
Pb 405.8 4.37 14.0 1.3d 11
Li 323.3 3.83 46.0 2.08b 22
460.3 4.52 12.5 0.7b 18
610.4 3.87 4.4 O.6b 8
670.8 1.85 0.067 0.007b 9
Mg 285.2 4.34 1.0 O.lb 10
Mn 403.4 3.08 0.20 0.018b 11
Na 330.2 3.75 12.5 0.9b 14
818 3.61 1500 125* 12
By permission from Dean (Dl, p. 63).
b 4-Methyl-2-pentanone.
0 Chloroform.
Gasoline.
14 I. MACINTYRE

an extremely powerful method which will certainly be more widely

used in the future; determinations of very small quantities can be made
by this means of iron, chromium, aluminium, lanthanum, magnesium,
copper, yttrium, nickel, manganese, and others. This type of procedure
is the method of choice at the moment for trace analysis because the
sensitivities obtained are rarely exceeded by the absorption method; the
latter is somewhat less convenient in use where many elements are to
be determined. This situation will probably change when new instru-
ments are developed which allow the use of either absorption or emission
and either aqueous or organic solvent, as seems appropriate.
3.5. INTERFERENCE
An interferent is a substance or factor which distorts the relationship
in pure solution and low concentration between salt concentration and
emission. Interferents can be classified as follows.
3.5.1. Spectral Interference
( a ) Background interference. All metals in large amounts give rise to
a continuous spectrum. Thus, sodium radiates from 360 to 602 mp, po-
tassium from 340 to 570 mp, and lithium from 320 to 460 mp ( D l ) . This
type of interference can be minimized by using more selective filters or
narrower monochromator band widths. In extreme cases, chemical sep-
aration may be required.
( b ) Ovedapping or adjacent emissions. When the emissions of elements
present in solution are closer together than can be separated by the
optical system used, again the only effective remedy is to use more
selective filters, or narrower band widths. Thus, the manganese
emission at 403 mp is difficult to separate from the potassium emis-
sion at 404 mp, while the magnesium emission at 285.2 mp cannot
readily be separated from the sodium emission at 285.3 mp, save
with very high resolutions. In some cases the emissions actually overlap,
so that complete separation is impossible whatever the optical means
employed. Thus, calcium, which has band systems of CaOH in the
green and red portions of the spectrum, will give rise to severe interfer-
ence with sodium when calcium is present in excess because the CaOH
molecules also emit at the sodium wavelength. In such cases chemical
separation is the only solution, although interference can sometimes be
minimized by choosing flame conditions least suitable to excitation of
the interferent. For example, in a cool flame calcium in excess will inter-
fere less with sodium than in a hotter flame. It is sometimes also possible
to minimize interference of this nature without chemical separation by
 FLAME PHOTOMETRY 15

adding substances which will depress the emission of interferent. In the

example discussed here, calcium interference with sodium can be min-
imized by adding excess of phosphate or aluminium, with which calcium
forms difficultly vaporizable compounds.
3.5.2. Cationic Interference
The ionization of an element in the flame can be considered as a
dissociation process:
A=A++e-
where A is an atom in the flame, A+ its positive ion and e- a free
electron. This reaction has an equilibrium constant K:

Ionization in various flames for alkali and alkaline earth metals is shown
in Table 4 ( a and b). As the dissociation of an atom into ion and electron
has a dissociation constant, it follows that increase in the flame content
of electrons will depress the ionization. This is accomplished when
another element, also capable of ionization in a flame, is present in solu-
tion. This depression of ionization will lead to an increase in the number
of atoms in the ground state, with a consequent increase in the number
of excited atoms. The resultant increase in emission intensity of the
element being measured is referred to as cationic interference or enhance-
ment. Since this phenomenon depends on ionization, it will not occur in
cooler flames or with elements of high ionization potential. Thus, sodium
and potassium which have fairly low ionization potentials may be deter-
mined together without cationic interference in the coal gas-air flame,
although in the air-acetylene flame the presence of sodium will enhance
potassium emission and vice versa (Figs. 6 and 7). This type of inter-
ference is quite common and is to be distinguished from spectral inter-
ference. The effects can be readily minimized by adding excess of
interferent, so that ionization is fully depressed and independent of any
further variation in interferent concentration; or when convenient, the
interference can be entirely removed by a suitable selection of flame
temperature. Thus, sodium and potassium cause mutual cationic inter-
ference in air-acetylene but not in coal gas-air flames, while sodium
markedly enhances calcium emission in an oxygen-acetylene flame but
to a small or negligible extent in an air-acetylene or oxygen-hydrogen
flame. This type of interference has been claimed to be absent from
atomic absorption spectrophotometry, but this is rather a reflection of
the fact that the absorption method has been used with elements with
higher ionization potentials in relatively cool flames rather than to an
16 I. MACINTYRE

inherent characteristic of the method. One can predict that if the

absorption method is used in a flame such as oxygen-hydrogen with either
rubidium or potassium, then enhancement effects should readily be
observed. A further method of overcoming this type of interference is
by adding an excess of an easily ionizable element such as cesium; this
is effective whether the interfering element is cesium or some other
element of low ionization potential.
TABLE 4a
PER CENT IONIZATION OF h K A L I AND METALSIN
ALKALINEEARTH FLAMES5

Ionization
potential Air- Hydrogen- Acetylene-
Element (ev) propane OXY OXY g*
Lithium 5.37 < 0.01 0.9 16.1
sodium 5.12 0.3 5.0 26.4
Potassium 4.32 2.5 31.9 82.1
Rubidium 4.16 13.5 44.4 89.6
Cesium 3.87 28.3 69.6 96.4
Calcium 6.11 < 0.01 1.0 7.3
Strontium 5.69 < 0.1 2.7 17.2
Barium 5.21 1.0 8.6 42.8
5By permission from Dean ( D l , p. 42).
From W. H. Foster, Jr., Ph.D. Thesis, Massachusetts Institute of Technology,
May, 1959.

3.5.3. Anionic Interference

This is the most persistent and difficult type of interference met with
in flame photometry. It is due to the formation of difficultly vaporizable
compounds of certain metals with certain anions. Thus, calcium emission
is depressed by the presence of phosphate, sulfate, and dichromate.
Aluminate anions, silicate, iron, and beryllium also cause depression of
calcium emission. The mechanism with all these substances is the forma-
tion of a compound of the type Ca-1-0 of high melting point and low
volatility (where I is interferent). Unlike cationic interference, which
is most marked in the hotter flames, this type of interference is most
troublesome in cooler flames. It is equally troublesome in emission or
absorption work. This type of interference can be overcome either by
adding excess of the interfering ion, such as phosphate ( M I ) , or by
adding a substance which complexes with the metal being determined
and prevents the formation of the refractory compounds. Ethylene-
diamine tetraacetate has been successfully used in this way by Willis
(W2). In biological materials this type of interference is unimportant
with sodium and potassium, very marked with calcium, and slight or
 TABLE 4b
OF METALSIONUED~
FRACTION
~ ~~ ~ ~ ~~ ~~~~~ ~
~~~~~ ~ ~~ ~

Ionization Pressurea Temperature

Metal potential ( atm ) 1500°K 2000°K 2500°K 3000°K 3500°K
Potassium 4.32 10-2 2.7 x 10-6 2.5 x 10-4 4.7 x 10-3 0.028 0.11
10-4 2.7 x 10-5 2.5 x 10-3 0.047 0.27 0.75
10-6 2.7 x 10-4 0.025 0.42 0.94 0.99
sodium 5.12 10-2 1.2 x 10-7 2.7 x 10-5 7 x 10-4 6.2 x 10-3 0.031
10-4 1.2 x 10-6 2.7 x 10-4 7 x 10-3 0.062 0.30
10-6 1.2 x 10-5 2.7 x 10-3 0.07 0.53 0.95
Calcium 6.11 10-4 5.5 x 10-8 3.0 x 10-5 1.4 x 10-3 0.019 0.12
10-6 5.5 x 10-7 3.0 x 10-4 0.014 0.19 0.77
By permission of Gaydon and Wolfhard (Gl, p. 283).
b This is the flame metal content expressed as a partial pressure.
18 I. MACINTYRE

Atomic rotio* No:K

0 I00 2 00
10

0 100 200 300

No concentration (mg/IOOrnl)

FIG.6. Flame photometer readings for potassium; effect of sodium. Curve A, solu-
tions containing 2 mg K/100 ml (as KC1) +
Na (as NaCl), as shown by abscissas;
Curve B, solutions containing no K, but Na, as shown by abscissas. Asterisks denote
ordinate and abscissa for curve A only. ( B y permission of Domingo and Klyne
(D2).)

Atomic rotio K : No -

K concentration ( rng / 100 rnl )

FIG.7. Flame photometer readings for sodium; effect of added potassium. In all
solutions 2 mg Na/100 ml (as NaCl); K (as KCl) added as shown by abscissas. ( B y
permission of Domingo and Klyne ( D 2 ) . )
 FLAME PHOTOMETRY 19

absent with magnesium. Phosphate does not depress magnesium emission

provided a hot flame such as oxygen-acetylene is used ( A l ) . It should
be noted that it is often impossible to select a flame which will minimize
anion interference without incurring the risk of cationic enhancement.
3.5.4. Spray Interference
If there is a difference in physical properties between the test solution
and the standards it is likely that erroneous results will be obtained.
This is quite well recognized and can be readily overcome. Together with
variation in flame characteristics between test and standard reading, this
is the only type of interference in which a lithium internal standard is
really effective.
3.5.5. Self-absorption
This refers to the failure of emission to increase in proportion to
solution concentration above a certain limiting concentration. It is due
to the absorption by atoms in the ground state in the cooler outer part
of the flame of light emitted by excited atoms in the central hotter part
of the flame. This effect is generally seen only with transitions from the
ground state, and higher level transitions ending in a state above the
ground state do not suffer from this type of interference. This is because
in the outer part of the flame only atoms in the ground state are present
and higher energy states do not occur.
This type of interference is not troublesome with sodium, potassium,
or calcium but does occur with magnesium in the oxygen-acetylene flame
at concentrations above 0.1 meq/liter. Error is avoided by preparing
standards sufficiently close together to eliminate any error by interpolat-
ing readings, by working at concentrations at which self-absorption does
not occur, or by using emissions involving higher energy level transitions.
4. Practical Applications
The determination of sodium and potassium in biological fluids and
tissues is so widely practiced that no detailed discussion is necessary.
A simple filter instrument with absorption or interference filters is per-
fectly adequate for most purposes. However, the determination of
calcium and magnesium in biological fluids and tissues has proved more
difficult and merits further discussion.
4.1. CALCIUM
Calcium has a higher excitation potential than sodium and potassium
(Table 5 ) , and consequently a higher flame temperature is necessary for
its determination. In addition, anionic interference is particularly trouble-
 TABLE 5
FLAMESPECTRA^
DETAILED
Excitation
Wavelengthc potential
Elementb (mv) Emitterd (ev)
Na 285.37 I 4.06
(5.14;33.3) 330.27 I 3.75
330.37 I 3.75
568.3 I 4.03
568.8 I 4.03
589.07, x I 2.10
589.5, x I 2.10
818.3 I 3.61
819.5 I 3.61
K 344.67 3.59
(4.34;20.6) 344.77 3.59
404.47 3.06
404.77 3.06
693.9 3.40
696.4 3.40
766.57,~ 1.61
769.97, x 1.61
Ca 422.7~,x I 2.93
(6,11;3,12) 393.4T I1 3.15
396.87 I1 3.12
622 CaOH 1.97
554 CaOH
Mg 277.7 I 7.17
(7.64;4.42) 277.8 I 7.17
278.0 I 7.17
278.1 I 7.17
278.3 I 7.17
285.2r, x I 4.34
333.0 I 6.43
333.2 I 6.43
333.7 I 6.43
382.9 I 5.85
383.2 I 5.85
383.8 I 5.85
279.6~ 11 4.43
280.3~ 11 4.42
37 1 MFP 3.49
383 MgO 3.38
(1 Reprinted from “Biochemists Handbook,” 1961, edited by C. Long; by permis-

sion of the publishers, E. N. Spon & Co. Ltd., London.

b The ionization potential of the neutral atom followed by the excitation potential
of the singly ionized atom (expressed in ev) are given in parentheses.
0 r represents a transition to ground state. The most suitable wavelengths for

analysis are given as x .

d I represents the neutral atom, I1 the singly ionized atom.
 FLAME PHOTOMETRY 21

some in biological fluids from the presence of phosphate. The preponder-

ance of sodium with its much greater emission intensity makes it quite
unsound to use the CaOH band emissions in the green or red portion of
the spectrum if one wishes the method to be applicable to urine as well
as to plasma. The constant amount of sodium in plasma makes it possible
to use for plasma analysis the band emissions of calcium at the red end of
the spectrum with a constant correction for the spectral interference from
sodium. This interference cannot be overcome even with the use of a
monochromator. However, it is much more satisfactory to measure the
calcium emission in the blue at 422.7 mp, using a monochromator instru-
ment with photomultiplier and a flame not cooler than air-acetylene. At
this wavelength sodium spectral interference is eliminated with a suitable
band width (0.3 mp or less, using an oxygen-acetylene flame and Zeiss
.
integral atomizer-burner ) Phosphate interference may be overcome
either by addition of excess phosphate ( M l ) or by complexing with
ethylenediamine tetraacetate ( W2) . Either the absorption (W2) or
emission methods may be used; the emission method is perhaps
more advantageous, as hotter flames with their lesser susceptibility to
anionic interference are generally used with the latter method. With
suitable precautions, calcium may be determined in all biological fluids
and tissues with the same ease and accuracy as sodium and potassium.
4.2. MAGNESIUM
The determination of magnesium has presented more difficulty than
calcium and this for several reasons: the excitation potential of the
magnesium line 285.2 mp is 4.3 ev compared with values of 2.1, 1.6, and
2.9 for the emissions of sodium at 589 my, potassium at 766.5 mp, and
calcium at 422.7 mp, respectively. For emission work it is therefore
necessary to use a hot flame such as oxygen-acetylene to attain sufficient
sensitivity. Further, even the emission obtained with an oxygen-acetylene
flame is small and requires high sensitivity for its detection and measure-
ment. In addition, a monochromator of very high resolution is necessary
if even partial separation of the emission of magnesium at 285.2 mp is
to be achieved from the adjacent sodium emission at 285.3 mp (Table 6).
This severe spectral interference is commonly neglected, but it can be
overcome by the use of sufficiently narrow band width (approximately
0.04 mp using the Zeiss burner and oxygen-acetylene flame with double
monochromator MM 12). Anionic interference is not troublesome with
magnesium using high-temperature flames, and Alcock et al. ( A l ) de-
scribed methods suitable for the determination of magnesium in all
biological fluids and tissues.
 TABLE 6
FLAME
L~NES
AND BANDSOF ANALYTICALIMPORTANCE@,*

Wavelength Wavelength Wavelength

(mv) Element (w.) Element (mp) Element
213.9 zn 374.8 Fe 484b Al
228.8 Cd 374.9 Fe 493.4 Ba
253.7 Hg 375.8 Fe 495b B
285.2 Mg 377.6 T1 497b Ti
285.3 Na 378.6 Ru 500b Zn
303.4 Sn 383b Mg 510b Be
307.6 zn 385.6 Fe 518b Ti
317.5 Sn 386.0 Fe 520.5 Cr
324.8 cu 387.3 co 520.6 Cr
326.1 Cd 387.4 co 520.8 Cr
327.4 cu 396.2 A1 521b B
328.1 Ag 403.3 Ga 535.0 TI
330.2 Na 403.5 Mn 540b Mo
330.3 Na 404.4 K 548b B
338.3 Ag 404.7 K 550b U
340.5 Pd 405.8 Pb 552b DY
341.2 co 407.8 Sr 553.6 Ba
341.5 Ni 410.2 In 554b Ca
343.5 Rh 417.2 Ga 560b La
344.6 K 420.2 Rb 562b Pr
344.7 ‘ K 421.6 Sr 565b Tb
349b Sn 422.7 Ca 57Ob Gd
350.2 co 425.4 Cr 570b DY
350.3 Rh 427.5 Cr 571b Pr
331.5 Ni 429.0 Cr 57627 V
352.4 Ni 430.4 Nd 589.0 Na
352.5 Ni 438b La 589.5 Na
353.0 co 442b La 600b Mo
360.5 Cr 444b Y 600b Tb
. 361.0 Pd 450b Nb 622b Ca
363.5 Pd 450b Gd 653b Srn
364b Te 451.1 In 660b Nd
368.4 Pb 455.4 Ba 670.8 Li
369.2 Rh 455.5 CS 681b Sr
371b Mg 460.7 Sr 715b Ti
37221 Te 460.9 sc 766.5 K
372.0 Fe 4626 Gd 769.9 K
372.3 Fe 462b Nb 780.0 Rb
372.8 Ru 466.2 Eu 794.8 Rb
373.3 Fe 467b Al 818.3 Na
373.5 Fe 471b Be 819.5 Na
373.7 Fe 472.3 Bi 852.1 cs
374.3 Fe 481b Ce 873b Ba
374.6 Fe 483b Y 894.3 cs
a The emissions are arranged in order of wavelength. Inclusion does not neces-
sarily mean that the emission is suitable for quantitative measurement of the element
concerned. Band emissions are given at the most sensitive wavelength and are marked b.
b Reprinted from “Biochemists Handbook,” 1961, edited by C. Long; by permis-
sion of the publishers, E. N. Spon & Co. Ltd., London.
 FLAME PHOTOMETRY 23

4.3. DETERMINATION OF MAGNESIUM AND CALCJUMIN BIOLOGICAL

FLUIDS AND TISSUES

The methods in use in the author’s laboratory for determination of

calcium and magnesium in plasma and urine and in fecal and tissue ash
are described below in detail. Willis (W2) has shown that absorption
methods are also suitable. The methods described here have proved
reliable in practice over several years of intensive use.
A Zeiss spectrophotometer PMQ I1 with flame attachment is used with
double quartz monochromator MM 12 replacing the single monochroma-
tor normally used with the instrument. The burner is modified slightly
by enlarging the fuel inlet to 1.3 mm. An oxygen-acetylene flame is used
for both magnesium and calcium. The wavelength settings are 285.2 and
422.7, while the acetylene pressures are approximately 200 mm and
150 mm of water for magnesium and calcium, respectively. Slit widths of
0.02 mm are used in each case, and the instrument is operated at full
sensitivity.
The point of peak emission is accurately located by rotating the wave-
length selector while spraying a suitable standard solution before analysis
for each element. The instrument is set to zero with the flame off but
with full sensitivity turned on. The flame background reading is recorded
after turning on the flame and while deionized water is sprayed; the
deflection of a 14 mM NaCl solution is next recorded followed by read-
ings for one of the working standard solutions (0.20 meq/liter for Mg
and 0.50 meq/liter for Ca). Conditions are satisfactory when the ratio
of emission due to working standard to emission due to flame background
is 1.0 or more for Mg and 5.0 or more for Ca. The deflection due to 14
mM NaCl should not exceed 2% of the 0.2 meq/liter standard for Mg nor
1%of the 0.5 meq/liter standard for Ca.
Stock Solutions
All solutions are made with deionized distilled water and are pre-
pared from AnalaR chemicals (British Drug Houses, Poole, England)
with the exception of the magnesium and calcium stock solutions. The
calcium solutions are made from “Specpure” calcium carbonate (John-
son, Matthey & Co. Ltd., London) dissolved in the minimum volume of
concentrated HCl. The magnesium stock solution is prepared from
“Specpure” magnesium sulfate. Polythene containers are used to store
all solutions.
The following solutions are used:
Magnesium Stock Solution I . A solution of 25 m M MgS04.7Hz0, the
concentration of which is checked by titration with disodium ethylene-
24 I. MACINTYRE

diamine tetraacetate which has been standardized with calcium stock

solution and with standard zinc acetate solution.
Magnesium Solution ZZ, 8 mM MgC1,.
Mixed Salt Solution A, 30 mM KC1, 5 mM KaS04, 1.4 M NaC1, 50 mM
KH2P04.
Mixed Salt Solution B, 50 mM KC1,5mM K2S04,50 mM NaCl, 150 mM
KHZPO?.
Calcium Stock Solution, 25 mM CaC12. This is prepared from “Spec-
pure” calcium carbonate in appropriate amount by dissolution in HCl.
Perchloric Acid, HC104, 60% (w/w ).
Hydrochloric Acid, HCl (sp. gr. 1.18).
Phosphoric Acid, H3POc (sp. gr. 1.75).
Phosphate Solution, 44.4 mM KH2P04.
Sodium Solutions, 200 mM NaC1; 14 mM NaC1.
Magnesium Standards for Plasma, Urine, and Feces. Solutions are pre-
pared by adding 10 ml mixed salt solution A, 10 ml calcium solution, and an
appropriate volume of magnesium solution I to 700 ml water in a liter
flask. Perchloric acid (50 ml) is now added, and the volume is made up
to 1 liter. A range of standards at intervals of 0.05 mM is prepared from
0.05 to 0.40 mM with respect to magnesium concentration.
Calcium Standards for Plasma, Urine, and Feces, Ten milliliters of
mixed salt solution A, 10 ml magnesium solution 11, and an appropriate
volume of calcium solution are added to 700 ml water in a liter flask.
Perchloric acid (50 ml) is now added, and the volume is made to 1liter.
A range of standards at intervals of 0.05 mM is prepared from 0.05 to
0.5 mM with respect to calcium concentration.
Magnesium Standards for Soft Tissues. Standard solutions are pre-
pared to cover the range 0.125-0.50 mM with respect to magnesium con-
centration at intervals of 0.125 mM. Ten milliliters of salt solution B, 2 ml
calcium solution, and appropriate volume of magnesium solution I are
added to approximately 700 ml water in a liter flask; HC1 (100 ml) is
added and the volume is made up to 1 liter.
Calcium Standards for Soft Tissues. Standard solutions are prepared
to cover the range up to 0.10 mM with respect to calcium at intervals of
0.01 mM. Ten milliliters of salt solution B, 10 ml magnesium solution I,
and an appropriate volume of calcium solution are added to approxi-
mately 700 ml water in a liter flask; HCl (100 ml) is added and the
volume is made up to 1 liter.
Combined Diluting and Deproteinixing Fluids. ( 1 ) Phosphate solution
(10 ml) is diluted to approximately 700 ml in a volumetric flask, mixed
with 55.5 ml HC104, and made up to 1 liter. This solution is used for
 FLAME PHOTOMETRY 25

plasma or serum. (2) Phosphate solution (10 ml) and sodium solution
(25 ml 200 mM NaCl) are diluted to approximately 700 ml in a volu-
metric flask. Perchloric acid (55.5 ml) is added and the volume is made
up to 1 liter. This solution is used for analysis of urine and fecal ash.
(3) H3PO4 ( 1 ml) is added to HCl (100 ml) and the volume is made up
to 1 liter. This solution is used for soft tissue analysis.
Procedure
The same dilution is used for analysis of both calcium and magnesium
in plasma or serum, urine, tissue ash, and, where possible, in fecal ash.
Occasionally a further dilution of the fecal ash solution may be appro-
priate where one element is in unusual concentration.
Readings are obtained for those standards just above and just below
each unknown sample reading, and the results are calculated by inter-
polation.
The normal (95% ) ranges for plasma or serum obtained with minimal
stasis from fasting subjects is 1.5-1.8 meq/liter for magnesium and
4.9-5.5 meq/liter for calcium. If heparin is used for an anticoagulant, it
should be purified by ion exchange resins, as samples of heparin are often
contaminated with calcium.
All glassware should be specially acid-washed and finally rinsed in
distilled water. All contact with filter paper or rubber stoppers should
be avoided.
The precision of a single routine estimation of calcium or magnesium
by this procedure is between 1 and 2%.
P l u m or Serum. One volume of plasma or serum is added drop by
drop to 9 volumes of combined diluting and deproteinizing fluid ( I ) .
The tube is capped with "parafilm" (A. Gallenkamp & Co. Ltd., London,
E. C. 2), and the contents are mixed by gently inverting the tube several
times, It is then centrifuged, and the clear supernatant is used.
Urine. Urine is diluted 10- or 20-fold with diluting fluid (2). Twenty-
four-hour urine specimens should be collected in a Winchester bottle
containing 10 ml concentrated HCl in order to prevent precipitation of
magnesium and calcium.
Feces. Specimens are homogenized, and a sample is dried at 105" for
16 hours. A known weight (approximately 1.0 g ) is ashed in a platinum
crucible at 400".The ash is dissolved in NHCl (200 ml HCl/g dry feces).
The solution is thoroughly shaken to ensure complete dissolution. The
solution is further diluted as required with diluting fluid (2).
Soft Tissues. Soft tissues are dried to constant weight at 105" and then
ground in an agate mortar with several changes of a mixture of equal
26 I. MACINTYRE

volumes of ethyl ether and light petroleum (b.p. 40-60"). Obvious tendon
is removed in the case of muscle. The powder is placed in the oven at
105"for 12 hours. A known weight (approximately 20 mg) of the dry fat-
free tissue powder is weighed into a platinum crucible and ashed at 400".
The ash is dissolved in diluting fluid ( 3 ) (4 m1/20 mg tissue powder).
The ash solution is suitable for the determination of sodium and potas-
sium by flame photometry as well as for calcium and magnesium.
Microanu2ysi.s. The methods described are directly applicable even
when only a small volume of plasma or urine is available. Magnesium
or calcium can easily be determined on 25 microliters of plasma with
the same precision as' in the macro-procedure. This small volume contains
approximately 2.5 pg of calcium and 0.5 pg of magnesium.
Twenty-five microliters of plasma are pipetted into a microcentrifuge
tube of approximately 0.4 ml capacity (Beckman Instruments, Inc.,
Spinco Division, Palo Alto, California). Then 250 pl of diluting fluid
(1) are added and the tube centrifuged in a microcentrifuge (Beckman
Instruments, Inc. ) . The supernatant fluid is analyzed by aspirating
directly from the micro-tube above the packed protein precipitate. It is
convenient to make an adaptor which will hold the small tube in the
sample turntable (Fig, 4) in the correct position for spraying.
The procedure may be still further scaled down in the case of calcium.
One or two microliters of plasma (0.1 or 0.2 pg of calcium) are pipetted
into the micro-tube. Then 250 pl of diluting fluid ( 2 ) are added and after
centrifugation the supernatant fluid is analyzed as described above.

5. Working Conditions
There are several essentials for the satisfactory operation of a flame
photometer when many analyses are to be undertaken.
1. A period of at least several weeks must be allowed between the
receipt of a flame photometer from the manufacturers and the under-
taking of clinical analysis. This is an absolute essential if the analyst is
to become thoroughly familiar with the working of his instrument.
2. The flame photometer must be carefully maintained. It is the
author's experience that an integral burner atomizer should be thoroughly
and carefully cleaned every day. In addition, one or two spare burners
should be available. These should be ordered when the flame photometer
is being purchased.
3. All reagents should be stored in polythene vessels.
4. An ample supply of deionized distilled water should be available.
5. Control solutions of composition unknown to the analyst should be
 FLAME PHOTOMETRY 27

analyzed with every batch. Where the control solution gives a result
outside of the chosen confidence limits, the analysis should be repeated
and if the error persists the test solutions should not be reported and the
analyses should be discarded until the cause of the error has been defi-
nitely proven.
6. For satisfactory flame analysis it is essential that the analyst or his
supervisor should be conversant with the theoretical basis of the method.
Rule-of-thumb working will sooner or later result in serious errors being
made.
6. Conclusion
Relatively few reliable texts are available on the subject of flame
photometry, but recently a comprehensive textbook (D1) has appeared.
A full bibliography is given in this book and the whole subject is thor-
oughly covered. The reviewer has drawn heavily on this work in pre-
paring this review. For more detailed theoretical considerations the
reader should consult Mavrodineanu and Boiteux (M3).
REFERENCES
Al. Alcock, N., MacIntyre, I., and Radde, I., The determination of magnesium in
biological fluids and tissues by flame spectrophotometry. J. Clin. Pathol. 13, 506-510
(1960).
B1. Barnes, R. B., Richardson, D., Berry, J. W., and Hood, R. L., Flame photom-
etry; a rapid analytical procedure, Ind. Eng. Chem., Anal. Ed. 17, 605-611 (1945).
C1. Champion, P., Pellet, H., and Grenier, M., De la spectrombtrie; spectrona-
trom8tre. Compt. rend. acad. sci. 76, 707-711 (1873).
C2. Cholak, J,, and Hubbard, D. M., Spectrochemical analysis with the air-
acetylene flame. Ind. Eng. Chem., Anal. Ed. 16, 728-734 ( 1944).
D1. Dean, J. A., “Flame Photometry.” McCraw-Hill, New York, 1960.
D2. Domingo, W. R., and Klyne, W., A photoelectric flame photometer. Biochem.
J. 46,400408( 1949).
E l . Ells, V. R., The Lundegardh flame method of spectrographic analysis. J. Opt.
SOC. Am. 31, 534-542 (1941).
E2. Ells, V. R., and Marshall, C. E., The determination of exchangeable bases
by the Lundegirdh spectrographic method. Soil Sci. Sac. Am. Proc. 4, 131-135
( 1939).
G1. Gaydon, A. G., and Wolfhard, H. G., “Flames, Their Structure, Radiation
and Temperature.” Chapman & Hall, London, 1953.
G2. Griggs, M. A., Johnstin, R., and Elledge, B. E., Mineral analysis of biologic
materials; use of Lundegirdh spectrographic method. Ind. Eng. Chem., Anal. Ed.
13, 99-101 (1941).
H1. Herschel, Sir J. F., Treatises of sound and light. In “Encyclopaedia Metro-
politana; or System of universal knowledge on a methodical plan projected by
Samuel Taylor Coleridge,” 2nd ed., rev., p. 438. J. J. Griffin, London, 1848.
28 I. MACINTYRE

J1. Janssen, J., Spectroscopic.-Sur l’analyse spectrale quantitative. Compt. rend.

71, 626-629 (1870).
K1. Kirchhoff, C., and Bunsen, R., Chemische Analyse durch Spectralbeobach-
tungen. Ann. Physik Chern. 110, 181-189 ( 1860).
L1. Lundegirdh, H., “The Quantitative Spectral Analysis of the Elements,” Vol.
1. G. Fischer, Jena, 1929.
L2. Lundegirdh, H., “The Quantitative Spectral Analysis of the Elements,” Vol.
2. G. Fischer, Jena, 1934.
M1, MacIntyre, I., The flame-spectrophotometric determination of calcium in
biological fluids and an isotopic analysis of the errors in the Kramer-Tisdall procedure.
Biochem. J . 67, 164-172 ( 1957).
M2. Mavrodineanu, R., Bibliography on analytical flame spectroscopy, Part I.
Appl. Spectroscopy 10, 51-64 ( 1956).
M3. Mavrodineanu, R., and Boiteux, H., “L’Analyse Spectrale Quantitative par
la Flamme.” Masson, Paris, 1954.
Rl. Russell, B. J,, Shelton, J. P., and Walsh A., The atomic-absorption spectro-
photometer and its application to the analysis of solutions. Spectrochim. Acta 8,
317-328 ( 1957).
S1. Schuhknecht, W., Spektralanalytische Bestimmung von Kalium. Angew. Chem.
60,299-301 (1937).
T1. Talbot, W. H. F., Some experiments on coloured flames. Edinburgh J . Scl.
6, 77-81 (1826).
W1. Walsh, A,, The application of atomic absorption spectra to chemical analysis.
Spectrochim. Acta 7, 108-117 ( 1956).
W2. Willis, J. B., The determination of metals in blood serum by atomic absorp-
tion spectroscopy.-I. Calcium. Spectrochim. Acta 16, 259-272 ( 1980).